SAP Community - SAP Community

Український бізнес у режимі змін: інсайти від керівної директорки SAP Ukraine

2026-02-05T16:54:27.830000+01:00

Дві розмови з Ганною Мікулицькою, керівною директоркою SAP Ukraine, про те, як зараз працює бізнес в Україні і про роль технологій.

Процеси та гігієна даних

У розмові з KPMG Ганна розповіла, як цифрова трансформація стала ключовою умовою стійкості українського бізнесу в поточних (чи теперішніх) надскладних умовах:

Цифрова трансформація сьогодні — базова умова стійкості бізнесу. Війна прискорила перегляд ІТ-інфраструктури та впровадження сучасних цифрових рішень, що підвищують конкурентоспроможність на глобальному рівні.
Технологічний сектор — це не лише IT-компанії: сьогодні технології присутні майже в кожній організації — від банків і телекомів до державних установ.
Штучний інтелект та дані. Щоб AI працював ефективно, потрібні структуровані, якісні дані — це базова інвестиція перед автоматизацією.
Людський капітал змінюється: зростає роль жінок і перекваліфікація спеціалістів, зокрема ветеранів, а освітні ініціативи й програми розвитку цифрових навичок стають ключовими для майбутнього екосистеми.
Відмова від ворожого софту — це не лише безпека, а й ефективність. Перехід на сучасні цифрові продукти може значно підвищити ефективність економіки та створити нові можливості для бізнесу.

Оборонка, кібербезпека та ментальне здоров’я

У випуску Let’s Talk разом із Європейською Бізнес Асоціацією Ганна Мікулицька, керівна директорка SAP Ukraine, поділилася баченням того, як український бізнес адаптується до нової реальності та які зміни посилюють його конкурентоспроможність.

Темп змін в Україні випереджає європейські ринки.
Трансформації відбуваються швидко, і в цьому контексті технології стають не розкішшю, а стратегічною необхідністю.
Люди — ключовий ресурс.
Ганна обговорює, яким чином внутрішня стійкість, підтримка команд та увага до ментального здоров’я формують більш гнучку й ефективну організаційну культуру.
Україна — інноваційний хаб.
У відео йдеться про те, чому Україна має всі шанси стати майданчиком для тестування нових технологій і привертати міжнародний інтерес.
Кібербезпека та ШІ — про практичний вплив.
Це не абстрактні тренди: кібергігієна стає частиною бізнес-культури, а штучний інтелект — інструментом для реальних рішень.
Лідерство без дистанції.
Ганна також торкається теми сучасного лідерства — як воно змінюється та що означає бути в контексті молодших поколін

Переходьте до перегляду повних інтерв'ю за посиланням:

KPMG Review magazine

EBA, Let’s Talk

А після перегляду чекаємо на вас у коментарях. Що вас зачепило найбільше? З чим ви погоджуєтесь, а з чим — ні?

SAP PO to SAP CI Migration - Simplifying File-Based Scenarios with SMB Adapter in SAP CI

2026-02-09T11:20:57.235000+01:00

Every enterprise today is actively working on SAP PO to SAP Cloud Integration (CI) migration as part of their move towards modern, cloud-based integration platforms. While many interface types can be migrated smoothly, one of the most common challenges during SAP PO to SAP Cloud Integration (CI) migration is handling file-based interfaces that rely on internal on-premise file systems. In many landscapes, business applications exchange data by placing files on shared network folders, and SAP PO integrations are tightly built around these locations.

In SAP PO, this was quite straightforward. The File/NFS Adapter allowed direct access to internal network paths without the need for FTP servers or external exposure. As a result, many legacy integrations continue to depend on this simple and secure approach.

But when moving to SAP Integration Suite, teams quickly realize an uncomfortable truth: SAP CI does not provide a native NFS adapter. Since SAP CI runs in the cloud, it cannot directly access internal file systems, making a direct “like-for-like” migration of such scenarios difficult.

This creates a big question:

How do we migrate scenarios where files are stored inside the customer’s internal network?

Solution :

Option 1 – FTP/SFTP

A common workaround suggested in projects is to introduce an FTP/SFTP server and move files there first.

But highlight why this is not ideal:

FTP is not recommended for security reasons
SFTP might require extra infrastructure and additional costs

So we need something better.

Option 2 – SMB Adapter (Best fit)

SAP has released the SMB Adapter in SAP Integration Suite specifically designed to address this exact problem.

The SMB adapter facilitates file-based operations and accelerates implementation time to connect to a server using SMB (Server Message Block) protocol. The adapter is available only for Standard service plans and above of the SAP Integration Suite.

Feature highlights:

Secure authentication with Server Message Block (SMB) shares.
Offers compatibility with various SMB dialects.
On-Premise Connectivity via Cloud Connector.
Supports connections to SMB using Encryption, DFS, and Signing.
Provides different file handling options such as override, append, fail, and ignore.

High-Level File Based Architecture Explanation:

The above diagram illustrates the architectural shift when migrating file-based integrations from SAP PO to SAP Cloud Integration.

In SAP PO landscapes (Diagram 1), accessing files stored inside the customer’s internal network was relatively simple. The SAP PO File Adapter could directly read and write files from network locations using the NFS protocol.

Application servers typically mounted shared drives, and SAP PO accessed them seamlessly as local folders.

However, in SAP Cloud Integration ((Diagram 2) this approach is no longer possible. Since SAP CI runs in the cloud, it cannot directly mount or access internal network file systems. Moreover, SAP Integration Suite does not provide an NFS adapter like SAP PO.

This creates a major migration challenge for scenarios where:

Business applications drop files on internal network folders
SAP PO polls those folders using the File/NFS adapter
No FTP/SFTP servers are involved

To address this gap, SAP introduced the SMB Adapter in SAP Integration Suite.

The lower part of the diagram shows the modern CI-based approach:

Internal folders are exposed as SMB Shares on the customer network
SAP Cloud Connector establishes a secure tunnel between SAP CI and the on-premise network
The SMB Adapter in SAP CI connects through Cloud Connector
Communication happens over TCP using the SMB protocol
Files can be securely read from or written to internal locations without exposing them to the internet

This architecture provides a clean and secure replacement for NFS-based integrations in SAP PO, enabling a smooth migration path for file-based scenarios without introducing FTP servers or additional middleware.

Additional Scenario – When an SMB Server Already Exists

In many customer landscapes, an SMB file server is already available in the internal network (diagram 1), and applications directly place files on SMB shared folders. In SAP PO setups, even in such cases, the File/NFS adapter cannot consume these files directly. The SMB share first needs to be mounted on the SAP PO server and exposed as a local file path. Only after this mounting, SAP PO can access the location using the NFS/File adapter. This adds additional administrative effort and dependency on server configurations.

With SAP Cloud Integration, this limitation no longer exists. The SMB Adapter in SAP CI can connect directly to an existing SMB server through SAP Cloud Connector without any need for drive mounting. As long as the folder is available as an SMB share and properly exposed via Cloud Connector, SAP CI can immediately read from or write to it. This significantly simplifies the architecture, reduces infrastructure dependencies, and makes the migration of such scenarios from SAP PO to SAP CI much more straightforward.

In the Next Sections, We Will See:

How to : Create or update an SMB Share on the network
Configure: Cloud Connector for SMB access
Build: Configure SMB Adapter in SAP CI
Demo Scenario:

Pick a file from on-prem location
Process it in SAP CI
Delete the file as a part of post processing

Section A – Preparing the SMB Share

In our scenario, the internal file server is Linux-based. The following steps need to be executed by the Infrastructure team to create and expose a folder as an SMB share. These commands can be run on the Ubuntu server via SSH using tools like PuTTY or the command prompt.

# Update packages - before installing new components, ensure that the system packages are up to date
sudo apt update
 
# Install Samba and client tools required for SMB communication
sudo apt install samba smbclient -y
 
# Create the folder that will be exposed as an SMB share
sudo mkdir -p /srv/smb/smb_share
 
# Create SMB user
sudo adduser smbuser
 
# Set directory ownership and permissions
sudo chown -R smbuser:smbuser /srv/smb/smb_share
sudo chmod -R 775 /srv/smb/smb_share
 
# Backup Samba config
sudo cp /etc/samba/smb.conf /etc/samba/smb.conf.bak
 
# Configure the SMB Share - Add the new share definition to the Samba configuration
sudo bash -c 'cat >> /etc/samba/smb.conf <<EOF
 
[smb_share]
   path = /srv/smb/smb_share
   browseable = yes
   read only = no
   valid users = smbuser
   force user = smbuser
   create mask = 0660
   directory mask = 0770
EOF'

This configuration defines:

The physical path of the share
Read/write access
Authorized user
File and folder permission masks

# Create and enable the Samba-specific password for the SMB user:
sudo smbpasswd -a smbuser
sudo smbpasswd -e smbuser
 
# Restart Samba services
sudo systemctl restart smbd nmbd
 
# Allow Samba through firewall (if UFW is enabled)
sudo ufw allow samba
 
# Verify Samba configuration
testparm -s
 
# Verify share listing
smbclient -L localhost -U%
 
# Test share access
smbclient //localhost/smb_share -U smbuser

#Accessing the Share from Windows access: Once the SMB share is created, it can be accessed from any authorized Windows machine using the network path format using the credentials created above:
\\<server-name-or-ip>\smb_share

Deploy the credentials (smbuser) as basic credentials in SAP CI Security Artifact, these credentials will be configured in the SMB Adapter for authentication.

Section B – Cloud Connector Configuration

As we are connecting to On Premise Server, Cloud Connector comes into picture for the connectivity.

Configurations:

Internal Host: The IP address or the host name of the Internal Server on which SMB share is avaiable

Port: 445 (Used for TCP protocol)

Check out this article to setup TCP connection in Cloud Connector: https://help.sap.com/docs/connectivity/sap-btp-connectivity-cf/configure-access-control-tcp

Section C – Setup SMB Adapter in SAP CI Iflow

SAP CI IFlow:

Configuration of SMB Sender Adapter

Field	Description
Address	Specify the hostname or IP address of the SMB Server. Specify Virtual Hostname if Cloud Connector is used
Domain	Specify the domain used for authentication.
Proxy Type	Specify the proxy type: • Internet • On-Premise
Location ID	Location ID of Cloud Connector (proxy type = On-Premise)
SMB Dialect	Select the SMB dialect or version to be used for communication. Recommendation is to use the latest one available in the adapter, 3.1.1 for now An SMB dialect is a specific version of the SMB protocol that defines the supported features, capabilities, and communication formats between a client and a server
Share	Specify the SMB share folder which represents the network resource Example: Assuming the mount path is \\ServerName\ShareName, you must specify Share as ShareName.
Authentication	Select the authentication type for connecting to the SMB server. Currently only Basic Authentication is supported
Credential Name	User Credentials security artifact
Timeout (in secs)	Specify the maximum waiting time (in seconds) for a response message from the SMB Server.
Enable Encryption	Enable to turn on encrypted communication between client & server. Recommendation is to Enable Encryption for better security. This option is applicable only if the server supports it. This adds encryption to the data in transit, supported only for SMB 3.x dialect
Enable DFS	Enable Distributed File System DFS in SMB is a feature that provides a unified virtual file system namespace, allowing multiple SMB file shares across different servers to be accessed through a single logical path.
Enable Signing	Enable SMB Signing. Recommendation is to Enable Signing for better security. This option is applicable only if the server supports it. This adds signing to the data in transit. supported only for SMB 3.x dialect
Maximum Reconnect Attempts	Specify the maximum number of retries permissible for reconnection to the SMB server before message processing starts.

Recommendation : Use the latest 3.1.1 SMB Dialect with Encryption and Signing for better security

The processing tab contains the following fields:

This tab has parameters similar to what we see in SFTP Adapter Processing.

Please check this document for detailed information: https://api.sap.com/odata/1.0/catalog.svc/Files('2ee08d2557d84acfab3331448c2784b1')/$value

Section D - Demo:

In the Demo, the idea is to read a file from the SMB Share using SMB Sender Adapter.

SMB Share:

Content in the File:

The Sender Adapter Configuration is shown as above, Lets deploy the IFlow now.

Message run in TRACE mode:

File is received in SAP CI and deleted from the SMB Share.

In this similar manner, you can also use SMB receiver adapter to write a file on the folder, the connection tab configuration remains same as Sender Adapter, Operation Type in the Processing tab will be changed to “Create File”

Conclusion:

Migrating file-based scenarios from SAP PO to SAP CI no longer needs to be a roadblock.

The SMB Adapter provides a clean, secure, and SAP-recommended approach to connect SAP Integration Suite with internal network file systems making it the closest and most reliable replacement for the NFS adapter of SAP PO.

With minimal changes to existing infrastructure and no need for FTP servers, the SMB adapter truly simplifies one of the trickiest parts of PO to CI migration.

Hope this help,

Cheers!

Punith

Будуй майбутнє з SAP: стажування в команді Sales

2026-02-09T12:58:09.449000+01:00

Ти ще встигаєш, стажування в SAP Ukraine чекає на тебе!

Шукаємо амбітного студента, який готовий підкорювати світ Sales разом із лідером ринку. Це не просто стажування — це реальний досвід у міжнародному середовищі.

Що на тебе чекає?

Участь у бізнес-дослідженнях і аналізі ринку.
Допомога у створенні маркетингових кампаній та розвитку локальної SAP Community.
Підтримка команди Sales під час зустрічей із клієнтами та партнерами.
Розвиток власної мережі контактів і навичок у глобальному середовищі.

Ти наш ідеальний кандидат, якщо:

✅Навчаєшся на 4–5 курсі (маркетинг, менеджмент, IT, фінанси).

✅Вільно розмовляєш українською та англійською.

✅Прагнеш професійного драйву.

Чому саме SAP?
SAP — це глобальна технологічна компанія, що об’єднує понад 100 000 фахівців у всьому світі та підтримує 400 000 клієнтів у різних галузях. Приєднуйся, щоб отримати досвід у маркетингу, аналітиці та продажах світового рівня!

Локація: Україна
Статус студента — обов’язковий

Твій старт у SAP вже тут 👉 подати заявку

SAP Community Voice: SAP BW News, Ask Anything to SAP Champions, and Events from Tokyo to Brazil!

2026-02-10T17:33:19.414000+01:00

For this week's edition of the SAP Community Voice, we will walk through exciting featured blog posts that have been trending over the last weeks in Technology Blogs, plus a review of the top new January features for SAP Business Data Cloud and Datasphere. Also, are you looking for answers from experts? Have you always dreamed of connecting with our SAP Champions? Then this is your moment, because the Ask Me Anything initiative with SAP Champions launched this month!

And we know that many of you who are watching right now also participated in Devtoberfest last year. Guess what? The winners have been announced! And we will tell you more details today!

We have upcoming events taking place all over the world, including Brazil, Germany, and Tokyo, giving our community multiple opportunities to connect, learn, and engage globally.

Today's edition is super exciting and filled with news that we know you would love to know more about. Let’s dive in!

Featured Blog Posts

Basics Information you should know about Custom Filters in Fiori elements

SAP Champion Mio Yasutake wrote an informative blog post of the basics of creating the Custom Filters in Fiori elements. Starting with how custom filters are used in Fiori elements when standard filter fields are not sufficient, for example when a different control is required or when entered values must be transformed before being converted into OData filter conditions. Mio gradually explains how custom filters work, including the role of the internal Filter Values model, and walks through the different implementation approaches depending on whether custom logic is required. With clear explanations, code examples, diagrams, and references to a working GitHub repository, her post provides a solid foundation for anyone looking to understand and correctly implement custom filters in Fiori elements.

Read Mio Yasutake Blog Post here

Top January Features of SAP Business Data Cloud and Datasphere News

2026 started with many great new features in SAP Business Data Cloud and Datasphere, and SAP´s Klaus Peter Sauer made sure to collect the most important ones to our SAP Community in his latest YouTube Video (click here for the video) and in a blog post. From Task Chains and Scheduling to Replication Flow, the January top features contains many technical enhancements but also one of the major milestones for BW modernization scenarios: Metadata extraction of SAP BW 7.5 is now enabled for the catalog plus also a data lineage from SAC story down to the BW 7.5 objects. And it does not stop there, because at the end of his post you will find the dates and links to sign up for the Webinar Series: The Next Era of Business Data. Make sure to take a look and save in your calendar, because this can be an opportunity to have an overview of SAP Business Data Cloud and guide you through its integral pillars and strategic vision.

Read Klaus Peter Sauer Blog Post here

Guide on SAP PO to SAP CI Migration Simplified File-Based Scenarios

SAP Champion Punith Oswal wrote in his latest blog post about one of the most usual challenges during SAP PO to SAP Cloud Integration (CI) migration, which is handling file-based interfaces that rely on internal on-premise file systems. When teams move to SAP Integration Suite, many realize that SAP CI does not provide a native NFS adapter. Starting from the question, “How do we migrate scenarios where files are stored inside the customer’s internal network?,” Punith shares his solution by going through each necessary section step by step, with very useful images to make it easier to follow. If migrating file-based scenarios from SAP PO to SAP CI has turned into a roadblock for you, this content is must-read.

Read Punith Oswal Blog Post here

Less Code More Value: Declarative Programming in CAP

SAP´s Brigit Selbach wrote a blog post, introducing the declarative approach in the SAP Cloud Application Programming Model (CAP) to reduce repetitive validation handlers and complex imperative code by describing what should happen instead of how to implement it. Brigit walks through the core concepts of declarative programming in CAP, including declarative expressions, input validations using built-in annotations, expressions as annotation values, and the new @assert feature released in December 2025. With clear examples covering queries, calculated elements, validations, and authorization scenarios, the blog shows how declarative programming leads to less boilerplate code, consistent validations, improved readability, and better maintainability across both CAP Java and CAP Node.js applications.

Read Brigit Selbach Blog Post here

Community News

Ask Me Anything Sessions with SAP Champions

SAP Champion and SAP BTP architect Sebastiano Marchesini introduced a new monthly AMA (Ask Me Anything) initiative launched by the SAP Champions, starting this month. Based on experience at SAP TechEd, learning sessions, and live events, the idea is getting direct honest answers from experts, which creates more connection and high trust relationships with other SAP community members. Each month, a different SAP Champion with a specific area of expertise will answer both technical and non-technical questions, submitted openly or anonymously. For this first edition, selected questions will be answered in a recorded video published on the SAP Communities YouTube channel. This AMA session will be open for questions until the last week of February. Check Sebastiano latest blog post here, for more details on how to participate. This initiative helps us to shape an open space for real discussions, practical experience, and shared learning within the SAP ecosystem.

Hobbies Series: Growing Together Through Dog Agility

The Hobbies Series have been one of the greatest exchange of experiences inside out Community Corner Blog Posts, and last week we were impacted by the inspiring and adorable relationship between SAP Champion Patrizia Rossi and her dog, Joker, and their journey with Dog Agility Sport (read the full blog post here). What started as simple walks in the park evolved into a demanding and rewarding sport that requires discipline, focus, problem-solving skills, and a deep connection between handler and dog. In this sport, Joker and Patrizia are what is called a team (or handler–dog pair), and they train together every Friday, which has also turned out to be a wonderful way to end the working week and completely switch off. Patrizia compares this a bit to machine learning: step by step, the dog learns the mechanisms, and over time it takes little for him to understand what to do. In the end, her hobby is not just about competition, but about finding real success in stepping off the field tired but fulfilled, knowing they both gave their best, learned something new, and strengthened their connection once again.

Devtoberfest 2025: Winners Announcement!

The wait is finally over! The Devtoberfest 2025 winners have been announced, and we want to thank our SAP community for the incredible participation and engagement throughout the event. Congratulations to all the winners who will be joining us at SAP TechEd this year, and a big shout-out to all finalists for their dedication. Do you want to know if you are one of the winners?! Check out the full list of their names in the blog post here!

Events

SAP CodeJam Roadshow 2026 - Brazil edition

SAP CodeJam Roadshow 2026 is back in Latin America, with a special edition in Brazil. The tour kicks off on March 1, with stops in several states across the country. Our experts will travel across Brazil to deliver CodeJams on Generative AI, SAP Joule, and much more. In addition, during the same weeks, we will also have SAP Inside Tracks, organized by amazing members of the local communities. And, of course, to make sure there are plenty of opportunities to talk and connect, we will also have meetups!

Check out our agenda below:

March 2 - Campinas, Brazil
Build AI Services using CAP and Generative AI Hub: https://community.sap.com/t5/sap-codejam/build-ai-services-using-the-cap-and-generative-ai-hub-campinas-brazil/ev-p/14317691

March 3 - São Paulo, Brazil
Joule Studio: Create Joule Skills and Agents Without Coding: https://community.sap.com/t5/sap-codejam/joule-studio-create-joule-skills-and-agents-without-coding-s%C3%A3o-paulo-brazil/ev-p/14317678 + SAP Meetup – São Paulo: https://community.sap.com/t5/s%C3%A3o-paulo-events/meet-up-s%C3%A3o-paulo-2026-03-03-brazilroadshow2026/ec-p/14319322#M38

March 4 - São Paulo, Brazil
SAP Build: Create Event-Based Processes: https://community.sap.com/t5/sap-codejam/sap-build-create-event-based-processes-s%C3%A3o-paulo-brazil/ev-p/14317705

March 5 - Rio de Janeiro, Brazil
Getting Started with Generative AI Hub on SAP AI Core: https://community.sap.com/t5/sap-codejam/getting-started-with-generative-ai-hub-on-sap-ai-core-rio-de-janeiro-brazil/ev-p/14317716

March 13 - Joinville, Brazil
Getting Started with Generative AI Hub on SAP AI Core

March 14 - Joinville, Brazil
SAP Inside Track – Joinville: SAP Inside Track Joinville 2026 - SAP Community

March 20 - Belo Horizonte, Brazil
Getting Started with Generative AI Hub on SAP AI Core

March 21 - Belo Horizonte, Brazil
SAP Inside Track – Belo Horizonte

And it doesn’t stop there: the roadshow continues in April as well! Keep an eye on posts here and on the official SAP Community announcements page, as we still have more cities to announce, with events scheduled between April 6 and 17. You will not want to miss this opportunity, right? So, register using the official links, as seats are limited! For more details, check the full blog post here.

SAP Inside Track Tokyo 2026

SAP Inside Track Tokyo 2026 is coming soon, bringing the local SAP community together for two hybrid events focused on sharing knowledge and real-world experiences. The event will take place on February 18 and March 6, 2026, from 6:00 PM to 10:00 PM (JST), and will be held in Japanese. Participation is free of charge, and registrations can be completed via the SAP Community event page. This year’s agenda will cover Business Data Cloud and Business AI, with the official theme to be announced soon. Whether you join on-site in Tokyo or online, SAP Inside Track Tokyo offers a fantastic opportunity to connect, learn, and engage.

Code Connect 2026: Call for Proposals Open!

We are excited to announce the return of Code Connect for its third edition, taking place from July 13 to 16, 2026, in St. Leon-Rot, Germany, and online. While attendee registration opens on April 10, 2026, the Call for Proposals is already open and runs from January 26 to March 13, 2026, inviting speakers to actively shape the agenda across all three events. (Links to submit your application: Call for proposals UI5con, Call for proposals reCAP and Call for proposals HANA Tech Con). Code Connect offers a unique opportunity to experience three specialized events in one location: UI5con, reCAP, and HANA Tech Con, allowing you to dive deep into various aspects of SAP development. Code Connect is designed for developers at every level: Whether you are an SAP veteran or just starting out. For more details and FAQ, please make sure to check the latest blog post here.

The 2025 Developer Insights Survey: The Report

2026-02-10T22:24:13.580000+01:00

Role Overlaps for external SAP developers

Part I: Introduction and General Information

Welcome to the 2025 SAP Developer Insights Survey report. This annual survey targets SAP’s existing external developer audience. It is used to track demographics, usage patterns, and trends within that population. 2025 is the fifth year we have conducted the survey. The survey program is led by the SAP Ecosystem and Community team. Key support is provided by the SAP Global Experience Management Office (XMO). Several other SAP teams provide advice and inputs to survey content. Those teams are listed at the end of this report.

The 2025 survey was open for six weeks - from January 22nd through March 7th.

Using social media announcements, SAP's Developer News show, and a web-popup on Community web sites, we invited all technical professionals in the SAP ecosystem to participate in the survey.

Each year's survey is composed of a mix of new topic questions and repeat select questions from past years to track trends.

Key Findings from the 2025 Survey

Generative AI is Positively Affecting our Developer Ecosystem - While these tools are not exclusively Joule-based, roughly half of our respondents report evaluating Gen AI developer tooling and, among those using it, over half report that these tools have a "profound positive impact" in their real world use.
Low Code Use is Increasing - SAP Build low code tools lead in this space. Overall use of Low Code products on production projects continues to increase.
BTP Guidance Framework - limited awareness -- this important guidance document was announced at TechEd this past November, but - as with BTP Developers Guide in 2024 - barely half of respondents were aware of this document's existence.
Work-from-Home Continues Trending Down - we have been tracking these numbers since 2021. 6% fewer folks used their home as their primary workplace when compared to last year.

These topics will be covered in more detail in the rest of the report.

General Topics

We included a number of general questions in the survey covering such topics as age, work location and status, as well as role.

Employment Status and Relation to SAP

Employment

"What is your current employment status?"

SAP Developer Community Composition

We have seen no significant change in these breakdowns from year to year.

SAP Developer Community Age Distribution

This shows increases in both the oldest age buckets for 2025. This was accompanied by similar decreases in the middle two age buckets ( ranging from 25 - 44 years of age). This merits deeper research as it potentially indicates either a gap in hiring younger employees to cover the normal aging of our community, or an outright loss of qualified younger staff. The two year trend in the 25-34 group reinforces that idea.

Where are the Developers?

We asked, "Which country are you based in?"

By country, India has the largest number of respondents and probably the densest geographic concentration, followed by Germany, and then the USA. When grouped by SAP region, however, EMEA North is largest.

The proportions remain essentially unchanged from 2024.

Part II. Narrowing the Response Set

A primary objective of our survey is to build a model of our community's usage patterns. It helps us build more effective programs. With that goal in mind, in the remainder of the report we will narrow the data we'll consider in these ways:

External Developers Only - we exclude responses from SAP employees, and others (students, analysts, etc.)
Recent Development Experience - we only include responses from those who stated that they had been part of an SAP extension or integration project within the past year

When we filter the data set in this way, we are left with a set of 628 responses that we'll work with for the remainder of the report.

Workplace and Trends

Office Location

"As of right now, where do you primarily physically work from?"

Year-to-year trend in Workplace

This shows a slowing trend of employees moving back to conventional offices from home. From a past Developer Survey, 29% of respondents reported they had primarily worked at home prior to the pandemic.

Recent SAP Project Experience?

"Have you developed integrations or extensions for any SAP products or technologies in the past 12 months?"

Extension and Integration are key scenarios for developers working with SAP applications. We wanted to focus on developers with recent experience in these areas. Roughly 70% of developers answered "Yes" here.

Popular Programming Languages

"Which programming languages have you used in your development work over the past 12 months? Please select all that apply."

This was posed as a multiple response question. ABAP dominates, as it has in every year of our survey. JavaScript is very popular as well, which could be attributed to its use in both SAPUI5 and CAP. Java and Python are roughly tied for third place. The relative positions of these top four languages hasn't changed much, year to year.

We extended our survey beyond just language use to ask about popular frameworks.

Popular Programming Frameworks

We asked respondents who are Developers to pick which common front and back-end frameworks they have worked with in the past year.

Front-Ends

Back-Ends

Application and Business Technology Platform Architectures

SAP applications can be extended using any one of several architectural approaches. These can be grouped into two major classes: on-stack (in-app) and side-by-side (essentially, BTP-based).

We were wondering which approaches are most popular.

Popular Extension Architectures

"For the most recent extension project that you were part of, what was the principal runtime architecture of the back-end elements?"

We don't distinguish between newer and older SAP applications (e.g., S/4HANA vs. ECC) -- this likely explains the dominance of "on-stack ABAP". Aside from the obvious dominance of ABAP, we can also see that CAP is relatively popular. As with last year's results on this question, we see Low Code products like Build Apps and Build Process Automation are popular -- this question is asking about real production projects, and seeing those two account for roughly 10% of all extension projects highlights the value of these tools.

ABAP and CAP both have style or language variants in their respective frameworks. ABAP can be programmed as "ABAP for Cloud Development" or "ABAP Standard". CAP is available in JavaScript or Java language variants. Where the respondent selected ABAP or CAP, we asked about the variant used on the project:

ABAP Variants

CAP Variants

The rough 50-50 breakdown of ABAP styles is consistent with responses elsewhere in the survey. It reflects that roughly half of the ABAP projects are ECC or very early forms of S/4HANA. Importantly, the group that didn't understand the difference is considerably smaller than last year (2% versus 5% in 2024).

CAP/JavaScript is plainly most popular for CAP projects. Unlike last year, we do have respondents reporting the use of CAP/Java projects, although the percentage is modest - 5%.

Popular IDEs

We are always curious about general work patterns of developers. We asked about developer's most commonly used editors. While it would be difficult in such a survey of all developers to add many questions, we were particularly curious about most popular IDE or editor.

Preferred IDE or Editor

"Which IDE or text editor do you most commonly use for development tasks?"

Consistent with past year's results, the editor of choice will depend on the language. SAP GUI and -- arguably -- Eclipse are the primary editors for ABAP, so it's maybe no surprise that they are reflected as most popular. This is the first year that SAP Build Code appeared in the survey. We maintained separate choices for BAS and Build Code to see where folks might say that they have been using BAS, but have not yet transitioned to the broader capabilities offered by SAP Build Code.

Potential / Use of Generative AI for Developers

Beginning with the 2024 survey, we asked developers about their thoughts and use of several major capabilities of Generative AI as it applies to software development.

Generative AI for Developers

"Which of these GenAI capabilities has your team explored or used? Select the choice that best describes your use of each capability. "

An brief explanation of each category was included in the survey and is shown below:

Code Generation and Autocompletion: GenAI tools can generate code snippets, complete lines of code, or suggest entire functions based on the context provided by the developer.

Bug Detection and Code Review: AI models can be trained to identify potential bugs and vulnerabilities in the code. They can also suggest improvements in code quality, best practices, and code consistency.

Automated Testing and Test Data Generation: GenAI can be used to create test cases, which can help in thoroughly testing software applications. This includes generating test data that covers edge cases, potentially reducing the time and effort in manual testing.

Documentation Generation: AI can assist in generating and updating technical documentation. It can interpret the code and create comprehensive documentation, which is crucial for maintaining and scaling software projects.

Natural Language Processing (NLP) for Code: AI models can translate natural language queries into code, allowing developers to express their intent in plain language. This can be particularly helpful for novice developers working with unfamiliar frameworks.

Customized User Experience Design: AI can help generate user interface designs to enhance user experience.

AI-Powered Development Tools: Integration of AI in IDEs for real-time assistance, such as suggesting optimizations, refactoring code, or even predicting the next steps in development.

Automated Code Refactoring: AI can assist in code refactoring by suggesting or even implementing improvements in code structure.

Software Design and Architecture: AI can help in generating software design patterns or architectural models based on specified requirements. This can speed up the initial stages of development and ensure adherence to best practices.

Enhanced Security Protocols: Generative AI can be used to develop advanced security protocols and encryption methods, providing robust security solutions in software applications.

We also believed that Generative AI has been on the scene long enough for folks to being to experiment with putting it to use in meaningful ways. So, we included a couple of questions to gauge their perceptions of its utility to to their teams.

Current Gen AI Use

"How would you characterize your team’s current use of Generative AI tools or LLMs in your day-to-day work? Consider tools from any vendor or open source. Examples might include Joule, SAP Generative AI Hub, GitHub Copilot, or open source LLMs. (select the most applicable statement)"

Gen AI Effectiveness for Developers

"Looking back on the past year, what is your own sense of how the use of Generative AI based tools or LLMs impacted the effectiveness or efficiency of your team?"

Joule in SAP Build Code

Several Joule features were introduced into SAP Build Code prior to the survey. We asked respondents about their awareness or use of these features.

Joule for CDS Modeling

Joule for Test Data Set Generation

Joule Code Generation

Development Job Roles

We asked respondents which common development-related roles they perform.

"Which roles describe your day-to-day work? (Select all that apply)"

Respondents often "wear multiple hats" in their job. This question is asked in "select all that apply" form to capture that information.
The Venn diagram on the right depicts the relative overlaps of the top seven roles reported by respondents. The area of each overlapping region reflects the count respondents performing the overlapping roles. For example, from this we can seel that all (UX) "Designers" are also "Developers". Most but not quite all "Enterprise Architects" also are "Solution Architects".

These patterns of overlap are very similar to those in past years.

It is unsurprising that the "Developer" role leads. Architecture roles as a group have increased each year.

Development Domains

For respondents selecting "Developer" as one of their roles, we were interested in the specific application Domains they work in:

"Which of the following development domains describe your day-to-day work? (select all that apply)"

Year to year, Integration came in higher than we expected. Mobile web front-end development is more common than native mobile application development. This isn't completely unexpected, but worth noting.

Developer Satisfaction with SAP

SAP Development Satisfaction

"For your recent experience developing an integration or extension, how would you describe your overall satisfaction with SAP?"

Roughly 74% of respondents report they are “very satisfied” or “satisfied” with their development work with SAP. This is back near the 75% reported two years ago and also up from the 2024 figure, 70%.

We also gave all respondents a list of potential obstacles in their use of SAP tools and technologies and asked them to select the most significant pain points.

Obstacles in SAP tool us

"Which of these potential pain points are obstacles to your development when using SAP tools or frameworks? (select up to three most significant items)"

BTP Guidance Framework

Awareness and Use of the BTP Guidance Framework

"The SAP BTP Guidance Framework was announced at SAP’s TechEd this past November. Which statement best described your level of familiarity with this document?"

Nearly half of respondents were unaware of this document's release. Raising awareness for this and other related BTP framework documents would be a good goal for 2025.

Cloud Insights

We first asked respondents if they used any cloud providers for their projects. For those that answered, "yes", we asked about the use of several major providers, including SAP BTP. We also asked if their use included SAP or non-SAP projects. This information was condensed into a chart depicting the relative use of each cloud provider.

Similar to the earlier questions around BTP Environments, these numbers do not reflect a tally of projects for each platform - instead it reflects developer exposure to each.

Using Cloud Providers?

"Are you using any cloud providers for your development projects?"

We asked, "Which cloud providers are you using for development projects?" and we asked in a way to separate SAP-related projects from more general use.

SAP Business Technology Platform

Microsoft Azure

Amazon Web Services

Google Cloud Platform

Alibaba Cloud

Low / No Code Tools

This is the third consecutive year in which we've asked questions around Low Code and No Code (LCNC) products. The list was composed of leaders in the then current Gartner LCNC Magic Quadrant Report and select SAP products. SAP Process Automation -- now Build Process Automation -- was not yet announced at the time of the 2022 survey, so it only appears in the more recent data.

For each product, we asked respondents to select the most applicable category reflecting their awareness or use. These questions were organized as a single large matrix of boxes in Qualtrics - There is no "I am unfamiliar with this product" response. Instead, the respondent simply would skip checking one of the other boxes.

It is important to point out that we are depicting response counts in the following charts rather than percentages and that the difference in total annual response counts must be considered to compare between years. The raw respondent count for 2023 was 2.4 times larger than 2022. And 2024 responses were less than half of 2023. We'll normalize for that in a later table.

These depictions can be misleading: our number of respondents each year varies. But we can normalize for that annual variation; what does this look like then? If we extract the data for production use for all products, we can use that to calculate the proportion use of each product by year (in other words, the sum of each column below totals one hundred percent). We can then visualize the relative shift in popularity over time:

Productive use of Product	2022	2023	2024	2025
Vendor A	20%	48%	15%	27%
SAP Build family (Build Apps + BPA)	2%	32%	23%	35%
Vendor B	14%	28%	7%	18%
SAP Signavio	5%	14%	5%	9%
Vendor C	7%	12%	3%	3%
Vendor D	2%	2%	1%	3%
Vendor E	3%	4%	1%	2%
Vendor F	3%	4%	1%	2%

SAP Build low code products dominate in our SAP ecosystem.

Learning and Help Resources

We were interested in preferences in the format or media type of learning resources by respondents.

Preferred Learning Resources

"When learning new skills or technical subjects over the past 12 months, what resources did you tend use the most? Please select up to three resources."

These results seem relatively unchanged year-to-year. One point of interest is the large difference in popularity between tutorials and MOOCs (53% versus 16% here). Tutorials tend to be hands-on. They are also broken into substantially smaller time chunks. This reinforces our hypothesis that different content types and required time investment have an impact in some learning scenarios. This merits investigation.

On the topic of Help, we shifted in this next question from formats to specific web sites.

Top SAP Help Resource Sites

"What is your preferred resource if you need help, or have a challenge with the SAP technology / tool you are working with? "

Almost sixty percent of respondents say they use organic search or SAP Community as their primary source of SAP help. This is consistent with past years. This continues to demonstrate good visibility of the Community site overall and likely speaks well for the usefulness of the site content.

Satisfaction with SAP Learning Hub

SAP Learning Hub Awareness

"Which of the following best describes your exposure to the SAP Learning site?"

Overall Value of Learning Hub

"Overall how valuable are the resources on that site in helping you to understand SAP developer technologies?"

Technical Certifications

"How many technology-related certification programs have you completed within the past four years?"

Survey Methodology

This report is based on a Qualtrics web-based survey of 1,492 respondents from 58 countries. The survey ran for six calendar weeks between January 22 to March 7th, 2025. Ninety percent of the respondents invested ten minutes or less with the survey.

The survey was promoted via the SAP Community website and the SAP Developer Center, sap.com pop-up intercepts, social media posts.

Other References

The annual Stack Overflow Developer Survey is an excellent (and no-cost) reference for industry-wide habits of developers.

Highlights from SAP Inside Track Bengaluru – 14th June 2025

2026-02-10T23:30:00.035000+01:00

Highlights from SAP Inside Track Bengaluru – 14th June 2025

#sitBLR | #SAPInsideTrack | #CommunityMatters | #SAPDevelopers

Another phenomenal day at SAP Whitefield as the Bengaluru chapter of SAP Inside Track (SITBLR) brought together technologists, thought leaders, and passionate community members under one roof!

A vibrant Saturday filled with knowledge-sharing, innovation, networking, and celebration of community spirit — here are the moments that stood out!

🌟 Keynote That Set the Tone

The event kicked off with an inspiring keynote by Sindhu Gangadharan, MD & SVP, SAP Labs India. Her powerful message on “The Future of Tech Leadership and Community Innovation” resonated deeply with the audience. She highlighted the role of community-led growth, the potential of Agentic AI + BDC, and how grassroots innovation can transform enterprises.

🔵 “When the community thrives, innovation accelerates.” – Sindhu

💡 Spotlight Session – Devraj from IBM

One of the most awaited sessions was by Devraj from IBM, who delivered an energetic and insightful talk on “Agentic AI with Joule studio and Agent builder ”. His storytelling at demo pod 9, and practical use cases sparked amazing conversations during and after the session.

🎉 What’s more — Devraj’s SAP Press book “S/4HANA” 5th edition was the talk of the day. Attendees were excited to get selife with the author!

🤝 Community Meet-up – IBM Team & Sindhu

A special moment unfolded when IBM SAP community members met with Sindhu. It was a powerful reminder of how cross-company collaboration drives the SAP ecosystem forward.

Meaningful conversations. New connections. Shared purpose.

📸 Event Gallery – Capturing the Energ

Sindhu’s Keynote Moments
Devraj’s S/4HANA BOOK
Devraj Live Session
IBM Team with Sindhu
Demo Pods & Hands-on Sessions
SitBlr Moments & Networking
Speaker at SIT
Volunteers at SIT

The spirit of collaboration, the depth of tech conversations, and the warmth of the SAP community made this day unforgettable. A huge shoutout to the organizers, volunteers, speakers, and every attendee who made #sitBLR2025 a success!

🧩 Let’s Keep the Momentum Going!

If you attended a session that inspired you or had a great hallway chat, share your stories on SAP Community with #sitBLR and let the ripple continue. 🫶

🔁 See you at the next SAP Inside Track! Until then — keep building, keep sharing, and keep growing! 💙

Bienvenidos al grupo de SAP Community Mexico

2026-02-12T05:27:48.032000+01:00

Bienvenidos a la Comunidad SAP de México 🇲🇽🎉

Nos emociona abrir este espacio para la Comunidad SAP en México. Un lugar para conectar, compartir experiencias reales y fortalecer los lazos entre personas de todo el país, incluyendo encuentros y eventos presenciales.

Esta comunidad es organizada por María Antonieta López, Sandra Solís y Edgar Martinez.

Creemos firmemente en el valor de la Comunidad SAP y en el trabajo que realizan todos los representantes y contribuidores de la comunidad día con día. Este espacio existe para apoyar ese esfuerzo, aportar de regreso y ayudar a fortalecer el ecosistema SAP en México a través de la participación, la colaboración y el aprendizaje compartido.

En el centro de esta comunidad hay una creencia simple que compartimos:

SAP se aprende mejor a través de experiencias reales, conversaciones abiertas y personas ayudándose a crecer unas a otras.

Esto significa aprender de lo que realmente sucede en los proyectos, hacer preguntas honestas, compartir lecciones aprendidas y apoyarnos entre distintos roles, módulos y niveles de experiencia.

También tenemos una aspiración clara. Queremos ayudar a que la Comunidad SAP en México crezca con la misma fuerza, energía e impacto que vemos en otros países de Latinoamérica. Sabemos que ese tipo de comunidad se construye con el tiempo, gracias a personas que participan de forma constante y se apoyan mutuamente.

Qué puedes esperar aquí

Conversaciones prácticas de SAP basadas en el trabajo real
Intercambio abierto de conocimiento, tanto de éxitos como de retos
Una comunidad que va más allá de las conversaciones en línea y llega a eventos en México

Tono de la comunidad

Para mantener este espacio valioso y acogedor para todos:

Sé respetuoso, amable y constructivo
Comparte experiencias y haz preguntas honestas
Nada de spam, ventas agresivas ni competencias de ego
La comunidad y el aprendizaje siempre van primero

Preséntate 👋

Si te parece bien, compártenos en los comentarios:

¿En qué parte de México te encuentras o planeas visitar?
¿En qué área o módulo de SAP estás trabajando actualmente?
¿Qué estás aprendiendo o qué te genera curiosidad en este momento?
¿Algún interés fuera de SAP?

Esta comunidad será moldeada por las personas que participan en ella. Nos alegra que estés aquí y estamos muy entusiasmados por construir juntos algo significativo para la Comunidad SAP en México y para la Comunidad SAP en Latinoamérica.

María Antonieta, Sandra y Edgar

👉Usa el botón Comentar o Publicar comentario para presentarte.

------------------------------------------------------------------------------------------------------------------------------------------------

English version below 👇

------------------------------------------------------------------------------------------------------------------------------------------------

Welcome to the SAP Community for México 🇲🇽🎉

We’re excited to open this space for the SAP Community in México. A place to connect, exchange real experiences, and build stronger connections across the country, including in-person meetups and events.

This community is hosted by María Antonieta López, Sandra Solís, and Edgar Martinez.

We strongly believe in the value of the SAP Community and in the work that SAP Community representatives and contributors do every day. This space exists to support that effort, contribute back, and help strengthen the SAP ecosystem in México through participation, collaboration, and shared learning.

At the core of this community is a simple belief we all share:

SAP is best learned through real experiences, open conversations, and people helping each other grow.

That means learning from what actually happens in projects, asking honest questions, sharing lessons learned, and supporting each other across roles, modules, and levels of experience.

We also have a clear aspiration. We want to help grow the SAP Community in México to the same scale, energy, and impact we see in other countries across Latin America. We know that kind of community is built over time by people who consistently show up and support each other.

What you can expect here

Practical SAP conversations grounded in real work
Open knowledge sharing from both successes and challenges
A community that grows beyond online discussions and into events in México

Community tone

To keep this space valuable and welcoming for everyone:

Be respectful, kind, and constructive
Share experiences and ask honest questions
No spam, no aggressive selling, no ego games
Community and learning always come first

Say hello 👋

If you’d like, introduce yourself in the comments:

Where in México are you based, or traveling to?
What SAP area are you working with today?
What are you currently learning or curious about?
One interest outside of SAP?

This community will be shaped by the people who participate in it. We’re glad you’re here, and we’re looking forward to building something meaningful together for the SAP Community in México and the broader SAP Community in Latin America.

María Antonieta, Sandra, and Edgar

👉Use the Comment or Post Your Comment button below to introduce yourself.

Bangalore Stammtisch - Jan 29

2026-02-12T14:17:32.340000+01:00

I was happy to be part of the 1st 2026 SAP Event at Bangalore and it was this SAP Stammtish at the Mindset consulting office. I reached there by 5.00 PM and the venue was already set and people started walking in. For many if you are not aware of SAP Stammtisch, it is informal gathering , Short event where people Share, Learn and Network especially with SAP background.

we were greeted by the dyanmic team of Mindset which includes @Parvathi shankar ( ParUu and @Kamalika. The team helped to settle down with initial quick introduction.

The event started with a quick Quiz in mentimeter hosted by @Vara Prasad Thatikonda ( From Daimler). This was intresting and the fastest fingers with correct answers will be the winner. It was kind of a icebreaker but there were some hard questions also .

Post the quiz we started our discussion on the using VS code for ABAP , then to AI application, SAP Joule, SAP BDC, Integration suite. These were great dsicussion. There were discussion on the real use case how the customer was using the BTP services especially the SAP Translation Hub and how they saved lot of efforts. These were great insights from @vara prasad. We also heard from @Abhishek pal on the usage of the Joule in the SAP sucess factor .

The discussion really opened up some venues and it was a food for the thoughts. The time flew fast and it was already 8 Pm and many wanted to head back to home challengin the Bangalore traffic.

we ended with our ritual, that is the group photo !! We also plan to meet at the SAP Inside track 2026 at Bangalore.

Once again thank for Mindset consulting team for hosting us and being a great supporter of the SAP community.

Mini Session 02 - SAP COMMUNITY KOLKATA - 2026

2026-02-14T15:46:23.811000+01:00

🌐Unlock SAP’s Core Tech – From ABAP to AI

From Basics to Core Technology: Discover the Engineering Behind SAP

We’re excited to invite you to our upcoming session where we move beyond the fundamentals and dive into the real engineering behind SAP. This session is designed to be technical yet easy to grasp, ensuring you gain practical insights into SAP’s backend and the latest in enterprise AI.

🔹Topic 1: Diving into SAP’s Backend World with ABAP

Introduction to SAP ABAP (Advanced Business Application Programming)
How backend development powers large-scale enterprise systems
The role of coding in driving business applications

🔹Topic 2: Trends and Research in Enterprise Artificial Intelligence

Overview of Enterprise AI and GenAI and why it matters
Latest trends and cutting-edge research in AI
How modern AI technologies are being adopted in business

👥Who Should Attend?

Students from Computer Science, IT, or engineering backgrounds
Coders curious about backend development
Anyone interested in Artificial Intelligence and research
Students eager to learn the technical skills needed for an SAP career

👉No prior SAP knowledge required!

🌟Why You Should Join

Understand the backend technology of SAP
Learn about the latest AI trends in the corporate world
Network with industry experts and peers

📢Stay Connected

Follow SAP Community Kolkata on our social channels to stay updated on future events and initiatives:

📞Contact for Assistance

For queries before or on the day of the session, please feel free to comment.

📍Event Details

Venue: Godrej Genesis, Street Number 18, Block EP&GP, Sector V, Bidhannagar, Kolkata, West Bengal 700091, India – 15th Floor
Date: Saturday, 21 February
Time: 11:00 AM – 1:00 PM
Registration Link - https://luma.com/dkatd349

Building an AI-Powered Chemical Quality Management System: Part 1 - SAP CAP Backend Setup

2026-02-16T05:51:02.854000+01:00

Introduction

In this three-part series, we'll build an intelligent chemical quality management system that combines SAP Cloud Application Programming (CAP) Model with AI agents powered by LangChain and SAP AI Core. This system will automatically detect quality issues, recommend mitigation plans, and execute corrective actions.

What You'll Build:

A robust backend using SAP CAP to manage chemical production batches
AI agents that can reason about quality issues and suggest solutions
Automated workflows that integrate with your production systems

Who This Is For: Even if you're new to Python or AI development, this guide will walk you through every step with clear explanations and examples.

Part 1 Overview

In this first part, we'll set up:

SAP AI Core and AI Launchpad environment
SAP CAP backend service for chemical quality management
OData API endpoints for data access

Prerequisites

Before we begin, ensure you have:

SAP Business Technology Platform (BTP) account with AI Core entitlement
Node.js (version 18 or higher) installed
SAP Cloud Application Programming Model CLI: npm install -g @sap/cds-dk
Basic understanding of databases and REST APIs

Step 1: Setting Up SAP AI Core and AI Launchpad

1.1 Create SAP AI Core Service Instance

SAP AI Core is a managed service that allows you to execute AI workloads. Here's how to set it up:

In SAP BTP Cockpit:

Navigate to your subaccount
Go to Services → Service Marketplace
Search for "SAP AI Core"
Click Create and choose:

Plan: Extended (or Standard based on your needs)
Instance Name: my-ai-core-instance

Click Create and wait for provisioning

1.2 Create Service Key

A service key contains credentials needed to connect to AI Core:

Go to Instances and Subscriptions
Find your AI Core instance
Click the three dots (•••) → Create Service Key
Name it: ai-core-key
Click Create
Click View to see the credentials
Copy the entire JSON and save it as ai_core.json in your project root

Your ai_core.json should look like this:

{

"clientid": "sb-xxxxx",

"clientsecret": "xxxxx=",

"url": "https://your-subdomain.authentication.eu10.hana.ondemand.com",

"serviceurls": {

"AI_API_URL": "https://api.ai.prod.eu-central-1.aws.ml.hana.ondemand.com"

"appname": "xxxxx"

}

1.3 Subscribe to AI Launchpad

AI Launchpad provides a user interface to manage your AI scenarios:

In BTP Cockpit, go to Service Marketplace
Search for "SAP AI Launchpad"
Click Subscribe
Once subscribed, go to Instances and Subscriptions
Click "Go to Application" next to AI Launchpad
This opens the AI Launchpad interface where you can monitor deployments later

Step 2: Understanding the Data Model

Our chemical quality management system tracks:

Core Entities:

ProductBatch: Each production run with production details and status
SensorLog: Real-time sensor readings (temperature, pH, pressure, etc.)
LabTest: Laboratory test results for quality verification
QualityIssue: Detected quality problems with severity levels
Recommendation: AI-generated suggestions for fixing issues
RCAReport: Root cause analysis documentation

Relationships:

One ProductBatch has many SensorLogs and LabTests
One QualityIssue has many Recommendations
Issues link to RCAReports for detailed analysis

Step 3: Creating the SAP CAP Service

3.1 Initialize CAP Project

Open your terminal and run:

# Create project directory

mkdir chemical-quality-system

cd chemical-quality-system

# Initialize CAP project

cds init

# Create folder structure

mkdir db srv

3.2 Define the Data Model

Create db/schema.cds with the following content:

namespace sap.capire.CAP_Chemical_Service;

entity ProductBatch {

key ID: UUID @title: 'BatchID';

productID: String(50);

productionDate: DateTime;

shift: String(10);

operator: String(100);

status: String(30); // "Completed", "Held", "Rejected"

qualityStatus: String(30); // "Pass", "Fail", "Pending"

logs: Composition of many SensorLog on logs.batch = $self;

tests: Composition of many LabTest on tests.batch = $self;

}

entity SensorLog {

key ID: UUID;

batch: Association to ProductBatch;

timestamp: DateTime;

temperature: Decimal(5,2);

pressure: Decimal(5,2);

pH: Decimal(4,2);

viscosity: Decimal(6,2);

moisture: Decimal(5,2);

anomalyDetected: Boolean default false;

}

entity LabTest {

key ID: UUID;

batch: Association to ProductBatch;

testType: String(50); // "Viscosity Test", "Purity Test"

result: String(100);

resultValue: Decimal(10,3);

unit: String(20);

testDate: DateTime;

isCompliant: Boolean;

statusVal: Integer;

}

entity QualityIssue {

key ID: UUID;

batch: Association to ProductBatch;

detectedDate: DateTime;

issueType: String(100); // "Out-of-Spec pH"

severity: String(20); // "Low", "Medium", "High"

detectedBy: String(100); // AI Agent, Manual, Audit

status: String(30); // "Open", "Resolved", "Escalated"

recommendations: Composition of many Recommendation

on recommendations.issue = $self;

linkedRCA: Association to RCAReport;

statusVal: Integer;

}

entity Recommendation {

key ID: UUID;

issue: Association to QualityIssue;

suggestion: String(500);

suggestedBy: String(100); // AI Agent, Engineer

actionTaken: String(500);

actionStatus: String(30); // "Pending", "Completed"

triggeredWorkflowID: String(100);

statusVal: Integer;

}

entity RCAReport {

key ID: UUID;

relatedIssue: Association to QualityIssue;

rootCause: String(1000);

analysisDate: DateTime;

engineer: String(100);

documentReference: String(255);

summaryVector: LargeBinary; // For RAG embeddings

}

entity CalibrationLog {

key ID: UUID;

instrumentID: String(100);

calibrationDate: DateTime;

calibratedBy: String(100);

deviationObserved: String(255);

batchImpacted: Association to ProductBatch;

}

entity AuditLog {

key ID: UUID;

eventDate: DateTime;

user: String(100);

action: String(255);

relatedBatch: Association to ProductBatch;

issue: Association to QualityIssue;

notes: String(500);

}

3.3 Create Service Definition

Create srv/chemical-service.cds:

using { sap.capire.CAP_Chemical_Service as db } from '../db/schema';

service ChemicalService {

entity ProductBatch as projection on db.ProductBatch;

entity SensorLog as projection on db.SensorLog;

entity LabTest as projection on db.LabTest;

entity QualityIssue as projection on db.QualityIssue;

entity Recommendation as projection on db.Recommendation;

entity RCAReport as projection on db.RCAReport;

entity CalibrationLog as projection on db.CalibrationLog;

entity AuditLog as projection on db.AuditLog;

// Custom view for AI agent

@readonly

entity RecommendationView as SELECT from db.Recommendation {

issue.batch.productID as productID,

issue.severity as severity,

issue.status as status

};

// Action to update recommendations

action updateRecommendations(

productID: String,

workflowID: String,

actionStatus: String,

issueStatus: String

) returns String;

}

3.4 Implement Custom Logic

Create srv/chemical-service.js:

const cds = require('@sap/cds');

module.exports = cds.service.impl(async function() {

const { Recommendation, QualityIssue } = this.entities;

// Custom action to update recommendations

this.on('updateRecommendations', async (req) => {

const { productID, workflowID, actionStatus, issueStatus } = req.data;

try {

// Find recommendations for this product and workflow

const recommendations = await SELECT.from(Recommendation)

.columns('ID', 'issue_ID')

.where({ triggeredWorkflowID: workflowID });

if (recommendations.length === 0) {

return `No recommendations found for workflow ${workflowID}`;

}

// Update recommendations

await UPDATE(Recommendation)

.set({ actionStatus: actionStatus })

.where({ triggeredWorkflowID: workflowID });

// Update related quality issues

const issueIDs = [...new Set(recommendations.map(r => r.issue_ID))];

await UPDATE(QualityIssue)

.set({ status: issueStatus })

.where({ ID: { in: issueIDs } });

return `Successfully updated ${recommendations.length} recommendations and ${issueIDs.length} quality issues`;

} catch (error) {

req.error(500, `Error updating recommendations: ${error.message}`);

}

});

3.5 Add Sample Data

Create db/data/sap.capire.CAP_Chemical_Service-ProductBatch.csv:

ID;productID;productionDate;shift;operator;status;qualityStatus

550e8400-e29b-41d4-a716-446655440001;CHEM-A-2024;2024-01-15T08:00:00Z;Morning;John Doe;Completed;Pass

550e8400-e29b-41d4-a716-446655440002;CHEM-B-2024;2024-01-16T14:00:00Z;Afternoon;Jane Smith;Held;Fail

Create db/data/sap.capire.CAP_Chemical_Service-QualityIssue.csv:

ID;batch_ID;detectedDate;issueType;severity;detectedBy;status;statusVal

650e8400-e29b-41d4-a716-446655440001;550e8400-e29b-41d4-a716-446655440002;2024-01-16T15:30:00Z;Out-of-Spec pH;High;AI Agent;Open;1

Create db/data/sap.capire.CAP_Chemical_Service-Recommendation.csv:

ID;issue_ID;suggestion;suggestedBy;actionStatus;triggeredWorkflowID;statusVal

750e8400-e29b-41d4-a716-446655440001;650e8400-e29b-41d4-a716-446655440001;Adjust pH levels using buffer solution;AI Agent;Pending;WF-001;1

Step 4: Running the CAP Service

4.1 Install Dependencies

npm install

4.2 Deploy to SQLite (Local Development)

cds deploy --to sqlite

4.3 Start the Service

cds watch

You should see output like:

[cds] - connect to db > sqlite { database: 'sqlite.db' }

[cds] - serving ChemicalService { path: '/odata/v4/chemical' }

[cds] - server listening on { url: 'http://localhost:4004' }

4.4 Test the API

Open your browser to http://localhost:4004 to see the service index page.

Test these endpoints:

GET http://localhost:4004/odata/v4/chemical/QualityIssue

GET http://localhost:4004/odata/v4/chemical/RecommendationView

GET http://localhost:4004/odata/v4/chemical/ProductBatch?$expand=logs,tests

What We've Accomplished

In Part 1, you've successfully:

✅ Set up SAP AI Core and AI Launchpad environment
✅ Created a comprehensive data model for chemical quality management
✅ Built a fully functional CAP backend service
✅ Exposed OData APIs for data access
✅ Added sample data for testing

Next Steps

In Part 2, we'll dive into Python, FastAPI, and LangChain to understand how AI agents work and how to structure our agent service.

Resources

Building an AI-Powered Chemical Quality Management System: Part 2 - Python, FastAPI & LangChain Fund

2026-02-16T05:55:28.972000+01:00

Introduction

Welcome to Part 2! In Part 1, we built a robust SAP CAP backend for managing chemical production quality. Now, we'll explore the AI layer that will make our system intelligent.

In this part, you'll learn:

What LangChain is and why it's perfect for AI agents
How FastAPI creates modern, fast web services
The concepts behind AI agents and tool calling
How to structure a Python project for AI applications

Don't worry if you're new to Python! We'll explain every concept clearly with practical examples.

Understanding the Technology Stack

What is Python?

Python is a beginner-friendly programming language that's become the standard for AI and machine learning. It's:

Easy to read: Code looks almost like plain English
Powerful: Used by companies like Google, Netflix, and NASA
Rich ecosystem: Thousands of libraries for AI, data science, and web development

What is FastAPI?

FastAPI is a modern Python framework for building APIs (Application Programming Interfaces). Think of it as a bridge that lets different software talk to each other.

Why FastAPI?

Fast: One of the fastest Python frameworks available
Easy: Automatic API documentation and validation
Modern: Built-in support for async operations and type hints

Simple Example:

from fastapi import FastAPI

app = FastAPI()

@app.get("/hello")

def say_hello():

return {"message": "Hello, World!"}

This creates an API endpoint at http://localhost:8000/hello that returns a greeting!

What is LangChain?

LangChain is a framework for building applications powered by Large Language Models (LLMs) like GPT-4. It helps you:

Connect LLMs to external data and tools
Create AI agents that can reason and take actions
Build complex workflows with multiple AI steps

Key Concept: AI Agents

An AI agent is like a smart assistant that can:

Understand your request in natural language
Decide which tools to use to answer
Execute those tools
Reason about the results
Respond with a helpful answer

Example:

You ask: "Which products have quality issues?"
Agent thinks: "I need to check the database for open issues"
Agent does: Calls the get_products_with_quality_issues tool
Agent responds: "Product CHEM-B-2024 has a high-severity pH issue"

Setting Up Your Python Environment

Step 1: Install Python

Windows:

Download Python from python.org
Run installer and check "Add Python to PATH"
Click "Install Now"

Mac/Linux:

# Mac (using Homebrew)

brew install python@3.11

# Ubuntu/Debian

sudo apt update

sudo apt install python3.11 python3-pip

Verify installation:

python --version # Should show Python 3.11 or higher

Step 2: Create a Virtual Environment

A virtual environment keeps your project dependencies isolated. Think of it as a separate workspace for this project.

# Navigate to your project directory

cd chemical-quality-system

# Create virtual environment

python -m venv venv

# Activate it

# Windows:

venv\Scripts\activate

# Mac/Linux:

source venv/bin/activate

You'll see (venv) in your terminal prompt when activated.

Step 3: Install Required Libraries

Create a file called requirements.txt:

fastapi==0.109.0

uvicorn[standard]==0.27.0

requests==2.31.0

langchain==0.1.0

langchain-openai==0.0.2

generative-ai-hub-sdk==1.0.0

python-dotenv==1.0.0

Install everything:

pip install -r requirements.txt

What each library does:

fastapi: Web framework for building APIs
uvicorn: Server to run FastAPI apps
requests: Makes HTTP calls to our CAP service
langchain: Framework for building AI agents
langchain-openai: LangChain integration with OpenAI models
generative-ai-hub-sdk: SAP AI Core connectivity
python-dotenv: Loads environment variables from files

Understanding LangChain Concepts

Language Models (LLMs)

The "brain" of your agent. LangChain supports various models:

from langchain_openai import ChatOpenAI

# Create an LLM instance

llm = ChatOpenAI(

model="gpt-4o-mini",

temperature=0.0 # 0 = deterministic, 1 = creative

)

# Use it

response = llm.invoke("What is quality control?")

print(response.content)

Temperature explained:

0.0: Consistent, factual responses (best for our use case)
0.5: Balanced creativity and consistency
1.0: More creative, varied responses

Tools

Tools are Python functions that your agent can call. They must:

Have a clear description (tells the agent when to use them)
Accept inputs and return outputs
Be decorated with @tool

Example:

from langchain.tools import tool

@tool

def calculate_average_ph(batch_id: str) -> str:

"""

Calculate the average pH for a given batch.

Use this when asked about pH levels or averages.

"""

# Your logic here

return f"Average pH for batch {batch_id}: 7.2"

The agent reads the docstring to understand when to use this tool!

Prompts

Prompts give your agent personality and instructions:

from langchain.prompts import ChatPromptTemplate

prompt = ChatPromptTemplate.from_messages([

("system", "You are a chemical quality expert assistant."),

("human", "{input}"),

("placeholder", "{agent_scratchpad}") # For agent's reasoning

])

Prompt components:

System message: Sets the agent's role and behavior
Human message: The user's input
Agent scratchpad: Space for the agent's internal reasoning

Agent Executor

The executor manages the agent's decision-making loop:

from langchain.agents import AgentExecutor, create_openai_functions_agent

# Create agent

agent = create_openai_functions_agent(

llm=llm,

tools=[tool1, tool2, tool3],

prompt=prompt

)

# Create executor

agent_executor = AgentExecutor(

agent=agent,

tools=[tool1, tool2, tool3],

verbose=True, # Shows agent's thinking process

handle_parsing_errors=True # Gracefully handles errors

)

# Use it

result = agent_executor.invoke({

"input": "What quality issues exist?"

})

Building Your First Simple Agent

Let's build a minimal agent to understand the flow:

File: simple_agent.py

from langchain.agents import AgentExecutor, create_openai_functions_agent

from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder

from langchain.tools import tool

from langchain_openai import ChatOpenAI

# Step 1: Create a simple tool

@tool

def get_batch_status(batch_id: str) -> str:

"""Get the status of a production batch."""

# Simulated data

statuses = {

"BATCH-001": "Completed - Quality Pass",

"BATCH-002": "Held - Quality Issue Detected"

}

return statuses.get(batch_id, "Batch not found")

# Step 2: Initialize LLM

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0.0)

# Step 3: Create prompt

prompt = ChatPromptTemplate.from_messages([

("system", "You help manage chemical production batches."),

("human", "{input}"),

MessagesPlaceholder(variable_name="agent_scratchpad")

])

# Step 4: Create agent

agent = create_openai_functions_agent(

llm=llm,

tools=[get_batch_status],

prompt=prompt

)

# Step 5: Create executor

agent_executor = AgentExecutor(

agent=agent,

tools=[get_batch_status],

verbose=True

)

# Step 6: Test it

if __name__ == "__main__":

result = agent_executor.invoke({

"input": "What's the status of BATCH-002?"

})

print("\n=== Result ===")

print(result["output"])

Run it:

python simple_agent.py

You'll see output like:

> Entering new AgentExecutor chain...

Invoking: `get_batch_status` with `{'batch_id': 'BATCH-002'}`

Held - Quality Issue Detected

The status of BATCH-002 is "Held - Quality Issue Detected".

> Finished chain.

=== Result ===

The status of BATCH-002 is "Held - Quality Issue Detected".

What happened?

Agent received your question
Decided to use get_batch_status tool
Called the tool with batch_id='BATCH-002'
Got the result
Formulated a natural language response

Understanding FastAPI Basics

FastAPI turns your Python functions into web API endpoints.

Basic FastAPI App

File: basic_api.py

from fastapi import FastAPI, Query

app = FastAPI(title="Chemical Quality API")

# Simple endpoint

@app.get("/")

def home():

return {"message": "Welcome to Chemical Quality API"}

# Endpoint with parameter

@app.get("/batch/{batch_id}")

def get_batch(batch_id: str):

return {

"batch_id": batch_id,

"status": "Completed",

"quality": "Pass"

}

# Endpoint with query parameter

@app.get("/search")

def search_batches(

status: str = Query(..., description="Batch status to filter by")

😞

return {

"query": status,

"results": ["BATCH-001", "BATCH-003"]

}

Run it:

uvicorn basic_api:app --reload

Test in browser:

http://localhost:8000 → Home message
http://localhost:8000/batch/BATCH-001 → Batch info
http://localhost:8000/search?status=Completed → Search results
http://localhost:8000/docs → Automatic interactive API documentation!

FastAPI Path Decorators Explained

@app.get("/path") # GET request - retrieve data

@app.post("/path") # POST request - create data

@app.put("/path") # PUT request - update data

@app.delete("/path") # DELETE request - remove data

For our agent API, we'll use GET requests with query parameters.

Connecting to SAP CAP Service

To call our CAP service from Python, we use the requests library:

File: cap_client.py

import requests

BASE_URL = "http://localhost:4004/odata/v4/chemical"

def get_quality_issues():

"""Fetch open quality issues from CAP service."""

url = f"{BASE_URL}/QualityIssue"

params = {

"$filter": "status eq 'Open' or status eq 'Escalated'",

"$expand": "batch"

}

try:

response = requests.get(url, params=params)

response.raise_for_status() # Raises error for bad status

data = response.json()

issues = data.get("value", [])

return issues

except requests.exceptions.RequestException as e:

print(f"Error calling CAP service: {e}")

return []

# Test it

if __name__ == "__main__":

issues = get_quality_issues()

print(f"Found {len(issues)} quality issues")

for issue in issues:

print(f"- {issue['issueType']} (Severity: {issue['severity']})")

OData Query Parameters:

$filter: Filter results (like SQL WHERE clause)
$expand: Include related entities
$select: Choose specific fields
$top: Limit number of results
$skip: Pagination

Project Structure

Here's how to organize your Python agent service:

chemical-quality-system/

├── venv/ # Virtual environment

├── ai_core.json # SAP AI Core credentials

├── requirements.txt # Python dependencies

├── simple_agent.py # Learning example

├── basic_api.py # Learning example

├── cap_client.py # Learning example

Key Concepts Recap

AI Agent Flow:

User Question → Agent → Selects Tool → Executes Tool →

Gets Result → Reasons → Returns Answer

LangChain Components:

LLM: The AI brain (GPT-4, etc.)
Tools: Functions the agent can call
Prompt: Instructions for the agent
Agent: Decision-maker that uses tools
Executor: Runs the agent loop

FastAPI Flow:

HTTP Request → FastAPI Route → Python Function →

Process → Return Response

Common Pitfalls to Avoid

Forgetting to activate virtual environment

Always see (venv) in your terminal

Tool descriptions too vague

Bad: "Gets data"
Good: "Retrieves quality issues for batches with Open or Escalated status"

Not handling errors

Always use try-except when calling external APIs

Running wrong Python version

Use python --version to check

What You've Learned

In Part 2, you now understand:

✅ What Python, FastAPI, and LangChain are
✅ How AI agents make decisions and use tools
✅ How to structure a Python project
✅ How to connect to the SAP CAP backend
✅ The basics of building FastAPI endpoints

Next Steps

In Part 3, we'll combine everything to build the complete AI agent service that:

Detects quality issues automatically
Generates mitigation plans
Executes corrective actions
Integrates with SAP AI Core

Resources for Learning More

Python Basics:

FastAPI:

LangChain:

Continue to Part 3: Building the Complete AI Agent Service

AI-Powered Chemical Quality Management System: Part 3 - Creating AI Agent with LangChain & FastAPI

2026-02-16T05:57:58.395000+01:00

Introduction

Welcome to Part 3! You've built the SAP CAP backend (Part 1) and learned Python, FastAPI, and LangChain fundamentals (Part 2). Now we'll implement the complete AI agent using the exact code structure from your project.

What we'll build:

Three specialized tools that interact with the CAP backend
An intelligent LangChain agent powered by SAP AI Core
FastAPI endpoints for natural language queries
Complete integration between all components

By the end, your system will:

Detect quality issues through natural language queries
Generate detailed mitigation plans
Execute corrective actions automatically

Architecture Overview

User Query (Natural Language)

↓

FastAPI Endpoint (/qualityIssue, /mitigationPlan, /executeMitigationPlan)

↓

LangChain Agent (llm_agent.py)

↓

Agent Executor decides which tool to use

↓

Tool (chemical_service.py) calls CAP Backend via OData

↓

Process results and return to Agent

↓

Agent formulates natural language response

↓

Return JSON response to user

Project Structure

Let's organize our Python agent service properly:

chemical-quality-agent/

├── venv/ # Python virtual environment

├── ai_core.json # SAP AI Core credentials

├── requirements.txt # Python dependencies

├── chemical_service.py # LangChain tools for CAP backend

├── llm_agent.py # Agent configuration with AI Core

└── main.py # FastAPI application

Step 1: Understanding the Complete Code

Before we dive into implementation, let's understand what each file does:

chemical_service.py: Contains three LangChain tools (decorated with @tool) that make HTTP requests to our CAP backend OData service.

llm_agent.py: Configures the connection to SAP AI Core, initializes the LLM (GPT-4o-mini), creates the agent with tools, and sets up the agent executor.

main.py: Creates FastAPI endpoints that accept natural language queries and pass them to the agent executor.

Step 2: Creating the Tools (chemical_service.py)

Tools are Python functions that the AI agent can call. Each tool must have a clear description (docstring) that tells the agent when to use it.

File: chemical_service.py

from langchain.tools import tool

import requests

BASE_URL = "http://localhost:4004/odata/v4/chemical"

@tool

def get_products_with_quality_issues(input: str) -> str:

"""

Get all product batches with quality issues (Open or Escalated).

"""

url = f"{BASE_URL}/QualityIssue?$expand=batch&$filter=status eq 'Open' or status eq 'Escalated'"

try:

response = requests.get(url)

if response.ok:

issues = response.json().get("value", [])

if not issues:

return "No open or escalated quality issues found."

result = "\n".join([

f"Product ID: {issue['batch']['productID']}, Issue: {issue['issueType']}, Severity: {issue.get('severity', 'N/A')}"

for issue in issues

])

return result

return f"Error fetching issues: {response.status_code} - {response.text}"

except Exception as e:

return f"Exception occurred: {e}"

@tool

def get_mitigation_plan(input: str) -> str:

"""

Fetch mitigation plans for product batches with quality issues.

Filters based on 'Open' or 'Escalated' statuses and provides structured suggestions.

"""

try:

response = requests.get(f"{BASE_URL}/RecommendationView")

if response.ok:

recommendations = response.json().get("value", [])

filtered = [rec for rec in recommendations if rec["status"] in ["Open", "Escalated"]]

if not filtered:

return "No open or escalated mitigation plans found."

result_lines = []

grouped = {}

for rec in filtered:

product = rec["productID"]

if product not in grouped:

grouped[product] = {

"severity": rec["severity"],

"suggestions": []

}

grouped[product]["suggestions"].append({

"suggestion": rec["suggestion"],

"workflow": rec["triggeredWorkflowID"],

"status": rec["status"]

})

for product, details in grouped.items():

result_lines.append(f"### {product} (Severity: {details['severity']})")

for sug in details["suggestions"]:

result_lines.append(f"- Suggestion: {sug['suggestion']}")

result_lines.append(f" - Workflow: {sug['workflow']}, Status: {sug['status']}")

result_lines.append("") # Spacer

return "\n".join(result_lines)

else:

return f"Error fetching mitigation plans: {response.status_code} - {response.text}"

except Exception as e:

return f"API Error: {e}"

@tool

def update_chemical_recommendation(product_name: str, workflowID: str, actionStatus: str , issueStatus: str) -> str:

"""

Executes the mitigation plan for a given product.

Marks recommendations as 'Completed' and quality issues as 'Resolved'.

"""

actionStatus="Completed"

issueStatus="Resolved"

headers = {

"Content-Type": "application/json"

}

payload = {

"productID": product_name,

"workflowID": workflowID,

"actionStatus": actionStatus,

"issueStatus": issueStatus

}

post_url = f"{BASE_URL}/updateRecommendations"

try:

response = requests.post(post_url, headers=headers, json=payload)

if response.ok:

data = response.json()

return f"✅ Mitigation plan executed for product '{product_name}' (Workflow: {workflowID})\n{data.get('value') or data.get('message')}"

else:

return f"❌ Failed to update recommendation: {response.status_code} - {response.text}"

except Exception as e:

return f"❌ Error invoking mitigation update: {str(e)}"

Understanding the Tools:

@tool decorator: This tells LangChain that this function is a tool the agent can use
Docstring: The text inside triple quotes tells the agent WHEN to use this tool. The agent reads this to decide which tool to call.
get_products_with_quality_issues:

Makes a GET request to /QualityIssue with OData filter
$expand=batch includes related batch information
$filter gets only Open or Escalated issues
Returns formatted text describing each issue

get_mitigation_plan:

Calls the custom RecommendationView we created in CAP
Groups recommendations by product
Returns structured mitigation plans with workflow IDs

update_chemical_recommendation:

Posts to the custom updateRecommendations action
Updates recommendation status to "Completed"
Updates quality issue status to "Resolved"
Returns success/failure message

Key Design Pattern: All tools return strings (not objects) because the LLM needs text to understand the results.

Step 3: Creating the AI Agent (llm_agent.py)

Now let's set up the agent that uses these tools intelligently.

File: llm_agent.py

from langchain.agents import AgentExecutor, create_openai_functions_agent

from langchain.prompts import ChatPromptTemplate, SystemMessagePromptTemplate, HumanMessagePromptTemplate, MessagesPlaceholder

from gen_ai_hub.proxy.core.proxy_clients import get_proxy_client

from gen_ai_hub.proxy.langchain.openai import ChatOpenAI

import json

import os

# Load AI Core credentials

with open("ai_core.json", "r") as key_file:

svcKey = json.load(key_file)

authUrl = svcKey["url"]

clientid = svcKey["clientid"]

clientsecret = svcKey["clientsecret"]

apiUrl = svcKey["serviceurls"]["AI_API_URL"]

# Set required environment variables

env_vars = {

"AICORE_AUTH_URL": authUrl,

"AICORE_CLIENT_ID": clientid,

"AICORE_CLIENT_SECRET": clientsecret,

"AICORE_RESOURCE_GROUP": "default",

"AICORE_BASE_URL": apiUrl

}

os.environ.update(env_vars)

# Import tools from your service

from chemical_service import (

get_products_with_quality_issues,

get_mitigation_plan,

update_chemical_recommendation,

)

# Get AI Core proxy client and LLM

proxy_client = get_proxy_client('gen-ai-hub')

llm = ChatOpenAI(proxy_model_name='gpt-4o-mini', proxy_client=proxy_client, temperature=0.0)

# Register tools

tools = [

get_products_with_quality_issues,

get_mitigation_plan,

update_chemical_recommendation,

]

prompt = ChatPromptTemplate.from_messages([

SystemMessagePromptTemplate.from_template(

"You are a helpful assistant that uses tools to reason about chemical product batches, quality issues, and mitigation plans."

HumanMessagePromptTemplate.from_template("{input}"),

MessagesPlaceholder(variable_name="agent_scratchpad") # Required by agent

])

# Create agent with proper prompt

agent = create_openai_functions_agent(

llm=llm,

tools=tools,

prompt=prompt

)

# Setup agent executor

agent_executor = AgentExecutor(

agent=agent,

tools=tools,

verbose=True,

return_intermediate_steps=True,

handle_parsing_errors=True

)

Understanding the Agent Setup:

Loading AI Core Credentials:

with open("ai_core.json", "r") as key_file:
svcKey = json.load(key_file)

Reads your SAP AI Core service key that you created in Part 1

Setting Environment Variables:

os.environ.update(env_vars)

The SAP AI Core SDK needs these environment variables to authenticate

Getting the Proxy Client:

proxy_client = get_proxy_client('gen-ai-hub')

This creates a connection to SAP AI Core's Generative AI Hub

Initializing the LLM:

llm = ChatOpenAI(proxy_model_name='gpt-4o-mini', proxy_client=proxy_client, temperature=0.0)

proxy_model_name='gpt-4o-mini': Uses GPT-4o-mini model (fast and cost-effective)
temperature=0.0: Makes responses deterministic (consistent)

Creating the Prompt:

prompt = ChatPromptTemplate.from_messages([...])

SystemMessage: Tells the agent its role and capabilities
HumanMessage: The user's input (filled in at runtime)
agent_scratchpad: Where the agent writes its reasoning steps

Creating the Agent:

agent = create_openai_functions_agent(llm=llm, tools=tools, prompt=prompt)

This creates an agent that can call functions (tools)

Agent Executor:

agent_executor = AgentExecutor(...)

verbose=True: Shows the agent's thinking in console (great for debugging!)
return_intermediate_steps=True: Returns which tools were called
handle_parsing_errors=True: Gracefully handles errors

Step 4: Creating the FastAPI Service (main.py)

Finally, let's create the web API that makes the agent accessible.

File: main.py

from fastapi import FastAPI, Query

from llm_agent import agent_executor

app = FastAPI()

@app.get("/")

def read_root():

return {"message": "LangChain + CAP OData Integration API is live!"}

@app.get("/qualityIssue")

def chemical_quality_issues(prompt: str = Query(..., description="Natural language prompt")):

try:

result = agent_executor.invoke({"input": prompt})

return {"response": result}

except Exception as e:

return {"error": str(e)}

@app.get("/mitigationPlan")

def quality_issue_mitigation_plan(prompt: str = Query(..., description="Natural language prompt")):

try:

result = agent_executor.invoke({"input": prompt})

return {"response": result}

except Exception as e:

return {"error": str(e)}

@app.get("/executeMitigationPlan")

def execute_mitigation_plan(prompt: str = Query(..., description="Natural language prompt")):

try:

result = agent_executor.invoke({"input": prompt})

return {"response": result}

except Exception as e:

return {"error": str(e)}

Understanding the FastAPI Endpoints:

Root Endpoint (/😞

@app.get("/")
def read_root():

Simple health check to verify the API is running

Quality Issue Endpoint (/qualityIssue😞

@app.get("/qualityIssue")
def chemical_quality_issues(prompt: str = Query(...)):

Accepts a natural language query as a parameter
Query(...) means this parameter is required
Passes the prompt to the agent executor
Returns the agent's response

Mitigation Plan Endpoint (/mitigationPlan😞 Same pattern but for getting mitigation recommendations
Execute Mitigation Endpoint (/executeMitigationPlan😞 For executing corrective actions through natural language

The Magic:

result = agent_executor.invoke({"input": prompt})

This single line:

Takes the user's natural language query
Sends it to the agent
Agent reads available tools
Agent decides which tool(s) to call
Agent calls the tools
Agent formulates a response
Returns structured results

Step 5: Installing Dependencies

Create requirements.txt:

fastapi==0.109.0

uvicorn[standard]==0.27.0

requests==2.31.0

langchain==0.1.0

langchain-openai==0.0.2

generative-ai-hub-sdk==1.0.0

Install everything:

# Make sure you're in your virtual environment

source venv/bin/activate # Mac/Linux

# or

venv\Scripts\activate # Windows

# Install dependencies

pip install -r requirements.txt

Step 6: Testing the Complete System

6.1 Start the CAP Backend

First terminal:

cd chemical-quality-system

cds watch

You should see:

[cds] - serving ChemicalService { path: '/odata/v4/chemical' }

[cds] - server listening on { url: 'http://localhost:4004' }

6.2 Start the FastAPI Service

Second terminal:

cd chemical-quality-agent

source venv/bin/activate

uvicorn main:app --reload --port 8000

You should see:

INFO: Uvicorn running on http://127.0.0.1:8000

INFO: Application startup complete.

6.3 Test with Your Browser

Open: http://localhost:8000/docs

This opens the interactive API documentation (automatically generated by FastAPI!).

Test Query 1: Check Quality Issues

Click on /qualityIssue endpoint
Click "Try it out"
Enter prompt: What quality issues do we currently have?
Click "Execute"

You'll see the agent:

Call get_products_with_quality_issues tool
Return formatted results

Test Query 2: Get Mitigation Plans

Click on /mitigationPlan endpoint
Click "Try it out"
Enter prompt: Show me the mitigation plans for products with quality issues
Click "Execute"

Test Query 3: Execute Mitigation

Click on /executeMitigationPlan endpoint
Click "Try it out"
Enter prompt: Execute the mitigation plan for CHEM-B-2024 using workflow WF-001
Click "Execute"

6.4 Watch the Agent Think (Verbose Mode)

In your second terminal (where FastAPI is running), you'll see the agent's reasoning:

> Entering new AgentExecutor chain...

Invoking: `get_products_with_quality_issues` with `{'input': ''}`

Product ID: CHEM-B-2024, Issue: Out-of-Spec pH, Severity: High

The current quality issues include:

- **Product ID:** CHEM-B-2024

- **Issue:** Out-of-Spec pH

- **Severity:** High

> Finished chain.

This shows you exactly what the agent is doing!

6.5 Test with cURL

# Check quality issues

curl -X GET "http://localhost:8000/qualityIssue?prompt=What%20quality%20issues%20exist%3F"

# Get mitigation plans

curl -X GET "http://localhost:8000/mitigationPlan?prompt=Show%20mitigation%20plans"

# Execute mitigation

curl -X GET "http://localhost:8000/executeMitigationPlan?prompt=Execute%20mitigation%20for%20CHEM-B-2024%20workflow%20WF-001"

6.6 Test with Python

import requests

# Check quality issues

response = requests.get(

"http://localhost:8000/qualityIssue",

params={"prompt": "What quality issues do we have?"}

)

print(response.json())

# Get mitigation plans

response = requests.get(

"http://localhost:8000/mitigationPlan",

params={"prompt": "Show me mitigation plans"}

)

print(response.json())

Understanding the Agent's Decision-Making

Let's trace what happens when you ask: "What quality issues do we have?"

Step 1: User Query Arrives

GET /qualityIssue?prompt=What quality issues do we have?

Step 2: FastAPI Receives Request

def chemical_quality_issues(prompt: str = Query(...)):

result = agent_executor.invoke({"input": prompt})

Step 3: Agent Analyzes Query The agent (GPT-4o-mini) reads:

System prompt: "You help with chemical quality management"
User question: "What quality issues do we have?"
Available tools and their descriptions

Step 4: Agent Decides "This question is asking about quality issues. I should use get_products_with_quality_issues tool."

Step 5: Agent Calls Tool

get_products_with_quality_issues.invoke({"input": ""})

Step 6: Tool Executes

# Makes HTTP request

requests.get("http://localhost:4004/odata/v4/chemical/QualityIssue?...")

Step 7: Tool Returns Results

"Product ID: CHEM-B-2024, Issue: Out-of-Spec pH, Severity: High"

Step 8: Agent Formulates Response Agent reads the tool output and creates a natural language response:

"Currently, we have one quality issue:

Product CHEM-B-2024 has an Out-of-Spec pH issue with High severity."

Step 9: FastAPI Returns JSON

{

"response": {

"input": "What quality issues do we have?",

"output": "Currently, we have one quality issue...",

"intermediate_steps": [...]

}

Customizing the System

Add a New Tool

Want to check sensor data? Add to chemical_service.py:

@tool

def get_sensor_readings(batch_id: str) -> str:

"""Get the latest sensor readings for a specific batch."""

url = f"{BASE_URL}/SensorLog"

params = {

"$filter": f"batch/productID eq '{batch_id}'",

"$orderby": "timestamp desc",

"$top": 5

}

try:

response = requests.get(url, params=params)

if response.ok:

logs = response.json().get("value", [])

# Format and return

return format_sensor_data(logs)

except Exception as e:

return f"Error: {e}"

Then add it to the tools list in llm_agent.py:

from chemical_service import (

get_products_with_quality_issues,

get_mitigation_plan,

update_chemical_recommendation,

get_sensor_readings # Add this

)

tools = [

get_products_with_quality_issues,

get_mitigation_plan,

update_chemical_recommendation,

get_sensor_readings # Add this

]

Customize the System Prompt

Make the agent more specific by editing the prompt in llm_agent.py:

SystemMessagePromptTemplate.from_template(

"You are a chemical quality expert assistant specializing in pharmaceutical manufacturing. "

"You help identify quality issues, suggest corrective actions, and execute mitigation plans. "

"Always prioritize safety and compliance with GMP regulations."

)

Add More Endpoints

Add specialized endpoints in main.py:

@app.get("/dailyReport")

def daily_quality_report(date: str = Query(...)):

prompt = f"Generate a quality report for {date} including all issues and their status"

result = agent_executor.invoke({"input": prompt})

return {"response": result}

Troubleshooting Common Issues

Issue 1: "Module not found: gen_ai_hub"

Solution:

pip install generative-ai-hub-sdk

Issue 2: "Connection refused to localhost:4004"

Solution: Make sure CAP service is running:

cd chemical-quality-system

cds watch

Issue 3: "Invalid AI Core credentials"

Solution: Verify your ai_core.json file exists and has correct format:

{

"clientid": "...",

"clientsecret": "...",

"url": "...",

"serviceurls": {

"AI_API_URL": "..."

}

Issue 4: Agent keeps calling wrong tool

Solution: Improve the tool's docstring to be more specific:

@tool

def get_products_with_quality_issues(input: str) -> str:

"""

Use this tool ONLY when the user asks about:

- Current quality problems

- Which products have issues

- Open or escalated quality concerns

Do NOT use for: mitigation plans, sensor data, or historical issues

"""

Issue 5: "verbose" output not showing

Solution: Check your terminal where you ran uvicorn main:app. The verbose output appears there, not in the API response.

Performance Tips

Use temperature=0.0 for consistent, deterministic responses
Keep tool descriptions clear so the agent makes good decisions
Cache frequent queries in your tools if needed
Set max_iterations in AgentExecutor to prevent infinite loops

What We've Accomplished

In Part 3, you've successfully:

✅ Created three LangChain tools that interact with CAP backend
✅ Set up an AI agent connected to SAP AI Core
✅ Built FastAPI endpoints for natural language queries
✅ Tested the complete system end-to-end
✅ Learned how agents make decisions

Next Steps

In Part 4, we'll deploy this FastAPI service to SAP BTP Cloud Foundry so it's accessible in production, complete with:

Manifest.yaml configuration
Environment variable management
Cloud deployment steps
Production best practices

Key Takeaways

The Power of LangChain:

Tools let AI agents interact with external systems
Agents intelligently decide which tools to use
Natural language becomes the interface

The Beauty of This Architecture:

CAP Backend: Handles data and business logic
FastAPI: Provides web API
LangChain: Enables AI reasoning
SAP AI Core: Provides the LLM

Everything works together seamlessly!

Continue to Part 4: Deploying to SAP BTP Cloud Foundry

Building an AI-Powered Chemical Quality Management System: Part 4 - Deploying to BTP Cloud Foundry

2026-02-16T06:01:04.648000+01:00

Introduction

Welcome to the final part of our series! You've built a complete AI-powered quality management system:

Part 1: SAP CAP Backend
Part 2: Python, FastAPI & LangChain Fundamentals
Part 3: Complete AI Agent Implementation

Now it's time to deploy your FastAPI agent service to SAP BTP Cloud Foundry so it's accessible in production. No more "localhost" - your AI agent will be running in the cloud!

What you'll learn:

How Cloud Foundry works on SAP BTP
Creating a production-ready manifest.yaml
Managing environment variables securely
Deploying and monitoring your application
Best practices for production deployments

What is SAP BTP Cloud Foundry?

Cloud Foundry is a Platform-as-a-Service (PaaS) that runs on SAP Business Technology Platform. Think of it as:

A managed environment where you just push your code and Cloud Foundry handles the rest
Automatic scaling based on your application's needs
Built-in monitoring and logging
Integration with SAP services (like AI Core)

Why use Cloud Foundry for our agent?

Easy deployment: One command to deploy
Automatic restarts: If your app crashes, Cloud Foundry restarts it
Environment management: Easy dev/test/prod separation
Service binding: Automatic connection to SAP AI Core
Scalability: Scale up/down based on demand

Prerequisites

Before deployment, ensure you have:

✅ SAP BTP account with Cloud Foundry enabled
✅ Cloud Foundry CLI installed
✅ Your FastAPI application from Part 3
✅ SAP AI Core service instance and credentials

Installing Cloud Foundry CLI

Windows:

# Download installer from:

# https://github.com/cloudfoundry/cli/releases

# Run the .exe installer

Mac:

brew install cloudfoundry/tap/cf-cli@8

Linux:

wget -q -O - https://packages.cloudfoundry.org/debian/cli.cloudfoundry.org.key | sudo apt-key add -

echo "deb https://packages.cloudfoundry.org/debian stable main" | sudo tee /etc/apt/sources.list.d/cloudfoundry-cli.list

sudo apt-get update

sudo apt-get install cf8-cli

Verify installation:

cf --version

Step 1: Preparing Your Application for Cloud Foundry

1.1 Restructure for Cloud Deployment

Cloud Foundry needs a slightly different structure. Let's reorganize:

├── Chemical_Service_agent/ # NEW: Our Python agent for Cloud Foundry

│ ├── chemical_service.py

│ ├── llm_agent.py

│ ├── btpaiagent.py # NEW: Renamed from main.py

│ ├── requirements.txt

│ ├── manifest.yml # NEW: Cloud Foundry configuration

│ ├── runtime.txt # NEW: Python version

│ └── Procfile # NEW: Start command

└── ai_core.json

1.2 Create btpaiagent.py

Rename and modify main.py to btpaiagent.py with Cloud Foundry compatibility:

Key Changes for Cloud Foundry:

PORT environment variable: Cloud Foundry assigns a random port

port = int(os.getenv("PORT", 8080))

Health endpoint: Cloud Foundry checks if app is healthy

@app.get("/health")

VCAP_APPLICATION: Cloud Foundry provides app metadata

os.getenv("VCAP_APPLICATION", "local")

1.3 Update llm_agent.py for Cloud Environment

File: agent/llm_agent.py

from langchain.agents import AgentExecutor, create_openai_functions_agent

from langchain.prompts import ChatPromptTemplate, SystemMessagePromptTemplate, HumanMessagePromptTemplate, MessagesPlaceholder

from gen_ai_hub.proxy.core.proxy_clients import get_proxy_client

from gen_ai_hub.proxy.langchain.openai import ChatOpenAI

import os

# Load AI Core credentials from environment variables (Cloud Foundry)

authUrl = os.getenv("AICORE_AUTH_URL")

clientid = os.getenv("AICORE_CLIENT_ID")

clientsecret = os.getenv("AICORE_CLIENT_SECRET")

apiUrl = os.getenv("AICORE_BASE_URL")

resourceGroup = os.getenv("AICORE_RESOURCE_GROUP", "default")

# Validate credentials

if not all([authUrl, clientid, clientsecret, apiUrl]):

raise ValueError("Missing required AI Core environment variables")

# Set environment variables for SDK

env_vars = {

"AICORE_AUTH_URL": authUrl,

"AICORE_CLIENT_ID": clientid,

"AICORE_CLIENT_SECRET": clientsecret,

"AICORE_RESOURCE_GROUP": resourceGroup,

"AICORE_BASE_URL": apiUrl

}

os.environ.update(env_vars)

# Import tools

from chemical_service import (

get_products_with_quality_issues,

get_mitigation_plan,

update_chemical_recommendation,

)

# Get AI Core proxy client and LLM

proxy_client = get_proxy_client('gen-ai-hub')

llm = ChatOpenAI(proxy_model_name='gpt-4o-mini', proxy_client=proxy_client, temperature=0.0)

# Register tools

tools = [

get_products_with_quality_issues,

get_mitigation_plan,

update_chemical_recommendation,

]

prompt = ChatPromptTemplate.from_messages([

SystemMessagePromptTemplate.from_template(

"You are a helpful assistant that uses tools to reason about chemical product batches, quality issues, and mitigation plans."

HumanMessagePromptTemplate.from_template("{input}"),

MessagesPlaceholder(variable_name="agent_scratchpad")

])

# Create agent

agent = create_openai_functions_agent(

llm=llm,

tools=tools,

prompt=prompt

)

# Setup agent executor

agent_executor = AgentExecutor(

agent=agent,

tools=tools,

verbose=True,

return_intermediate_steps=True,

handle_parsing_errors=True

)

Key Change: Uses environment variables instead of ai_core.json file.

1.4 Update chemical_service.py for Production

File: agent/chemical_service.py

from langchain.tools import tool

import requests

import os

# Get CAP backend URL from environment or use default

BASE_URL = os.getenv("CAP_BACKEND_URL", "http://localhost:4004/odata/v4/chemical")

@tool

def get_products_with_quality_issues(input: str) -> str:

"""

Get all product batches with quality issues (Open or Escalated).

"""

url = f"{BASE_URL}/QualityIssue?$expand=batch&$filter=status eq 'Open' or status eq 'Escalated'"

try:

response = requests.get(url)

if response.ok:

issues = response.json().get("value", [])

if not issues:

return "No open or escalated quality issues found."

result = "\n".join([

f"Product ID: {issue['batch']['productID']}, Issue: {issue['issueType']}, Severity: {issue.get('severity', 'N/A')}"

for issue in issues

])

return result

return f"Error fetching issues: {response.status_code} - {response.text}"

except Exception as e:

return f"Exception occurred: {e}"

@tool

def get_mitigation_plan(input: str) -> str:

"""

Fetch mitigation plans for product batches with quality issues.

Filters based on 'Open' or 'Escalated' statuses and provides structured suggestions.

"""

try:

response = requests.get(f"{BASE_URL}/RecommendationView")

if response.ok:

recommendations = response.json().get("value", [])

filtered = [rec for rec in recommendations if rec["status"] in ["Open", "Escalated"]]

if not filtered:

return "No open or escalated mitigation plans found."

result_lines = []

grouped = {}

for rec in filtered:

product = rec["productID"]

if product not in grouped:

grouped[product] = {

"severity": rec["severity"],

"suggestions": []

}

grouped[product]["suggestions"].append({

"suggestion": rec["suggestion"],

"workflow": rec["triggeredWorkflowID"],

"status": rec["status"]

})

for product, details in grouped.items():

result_lines.append(f"### {product} (Severity: {details['severity']})")

for sug in details["suggestions"]:

result_lines.append(f"- Suggestion: {sug['suggestion']}")

result_lines.append(f" - Workflow: {sug['workflow']}, Status: {sug['status']}")

result_lines.append("")

return "\n".join(result_lines)

else:

return f"Error fetching mitigation plans: {response.status_code} - {response.text}"

except Exception as e:

return f"API Error: {e}"

@tool

def update_chemical_recommendation(product_name: str, workflowID: str, actionStatus: str, issueStatus: str) -> str:

"""

Executes the mitigation plan for a given product.

Marks recommendations as 'Completed' and quality issues as 'Resolved'.

"""

actionStatus = "Completed"

issueStatus = "Resolved"

headers = {

"Content-Type": "application/json"

}

payload = {

"productID": product_name,

"workflowID": workflowID,

"actionStatus": actionStatus,

"issueStatus": issueStatus

}

post_url = f"{BASE_URL}/updateRecommendations"

try:

response = requests.post(post_url, headers=headers, json=payload)

if response.ok:

data = response.json()

return f"✅ Mitigation plan executed for product '{product_name}' (Workflow: {workflowID})\n{data.get('value') or data.get('message')}"

else:

return f"❌ Failed to update recommendation: {response.status_code} - {response.text}"

except Exception as e:

return f"❌ Error invoking mitigation update: {str(e)}"

Key Addition: Configurable BASE_URL from environment variable.

Step 2: Creating Cloud Foundry Configuration Files

2.1 Create manifest.yml

This is the most important file - it tells Cloud Foundry how to run your app.

File: agent/manifest.yml

---

applications:

- name: btpaiagent

memory: 256M

disk_quota: 1G

instances: 1

buildpacks:

- https://github.com/cloudfoundry/python-buildpack.git

command: python btpaiagent.py

random-route: false

health-check-type: http

health-check-http-endpoint: /health

timeout: 180

env:

AICORE_BASE_URL: https://api.ai.prod.eu-central-1.aws.ml.hana.ondemand.com

AICORE_CLIENT_ID: your-client-id

AICORE_CLIENT_SECRET: your-client-secret

AICORE_AUTH_URL: https://your-subdomain.authentication.eu10.hana.ondemand.com

AICORE_RESOURCE_GROUP: default

CAP_BACKEND_URL: https://your-cap-backend.cfapps.eu10.hana.ondemand.com/odata/v4/chemical

Understanding manifest.yml:

name: btpaiagent

Your application name in Cloud Foundry
Will be used in the URL: btpaiagent.cfapps.eu10.hana.ondemand.com

memory: 256M

disk_quota: 1G

Memory: RAM allocated to your app (256MB is enough for FastAPI)
Disk: Storage space for your app and dependencies

instances: 1

Number of app instances to run
Start with 1, scale up later if needed

buildpacks:

- https://github.com/cloudfoundry/python-buildpack.git

Tells Cloud Foundry this is a Python application
Buildpack installs Python and dependencies automatically

command: python btpaiagent.py

The command to start your application
Runs btpaiagent.py which starts uvicorn

health-check-type: http

health-check-http-endpoint: /health

Cloud Foundry checks /health endpoint to verify app is running
If unhealthy, Cloud Foundry restarts the app

timeout: 180

Maximum time (seconds) for app to start
AI Core connection can take time, so 180s is safe

env:

Environment variables for your application
DO NOT commit real credentials to git!

2.2 Create requirements.txt

File: agent/requirements.txt

fastapi==0.109.0

uvicorn[standard]==0.27.0

requests==2.31.0

langchain==0.1.0

langchain-openai==0.0.2

generative-ai-hub-sdk==1.0.0

2.3 Create runtime.txt

Specifies Python version for Cloud Foundry.

File: agent/runtime.txt

python-3.11.x

2.4 Create Procfile (Optional)

Alternative way to specify start command.

File: agent/Procfile

web: python btpaiagent.py

Step 3: Securing Your Credentials

NEVER commit real credentials to git! Let's use secure practices.

3.1 Create .gitignore

File: agent/.gitignore

# Credentials

ai_core.json

manifest.yml

# Python

__pycache__/

*.pyc

*.pyo

venv/

.env

# IDE

.vscode/

.idea/

3.2 Create manifest-template.yml

File: agent/manifest-template.yml

---

applications:

- name: btpaiagent

memory: 256M

disk_quota: 1G

instances: 1

buildpacks:

- https://github.com/cloudfoundry/python-buildpack.git

command: python btpaiagent.py

random-route: false

health-check-type: http

health-check-http-endpoint: /health

timeout: 180

env:

AICORE_BASE_URL: <YOUR_AI_CORE_BASE_URL>

AICORE_CLIENT_ID: <YOUR_CLIENT_ID>

AICORE_CLIENT_SECRET: <YOUR_CLIENT_SECRET>

AICORE_AUTH_URL: <YOUR_AUTH_URL>

AICORE_RESOURCE_GROUP: default

CAP_BACKEND_URL: <YOUR_CAP_BACKEND_URL>

3.3 Fill in Your Credentials

Copy the template and add real values:

cd agent

cp manifest-template.yml manifest.yml

# Now edit manifest.yml with your actual credentials

Get your credentials from ai_core.json:

{

"clientid": "sb-xxxxx", → AICORE_CLIENT_ID

"clientsecret": "xxxxx=", → AICORE_CLIENT_SECRET

"url": "https://...", → AICORE_AUTH_URL

"serviceurls": {

"AI_API_URL": "https://..." → AICORE_BASE_URL

}

Step 4: Deploying to Cloud Foundry

4.1 Login to Cloud Foundry

# Login to SAP BTP Cloud Foundry

cf login -a https://api.cf.eu10.hana.ondemand.com

# You'll be prompted for:

# - Email

# - Password

# - Org (organization)

# - Space (dev, test, prod)

Example:

API endpoint: https://api.cf.eu10.hana.ondemand.com

Email: your.email@company.com

Password: ********

Select an org:

my-company-org

Org: 1

Select a space:

dev
test
prod

Space: 1

Targeted org my-company-org

Targeted space dev

4.2 Verify Your Target

cf target

You should see:

API endpoint: https://api.cf.eu10.hana.ondemand.com

API version: 3.x.x

User: your.email@company.com

Org: my-company-org

Space: dev

4.3 Push Your Application

# Navigate to agent directory

cd agent

# Deploy to Cloud Foundry

cf push

What happens during cf push:

Upload: Your code is uploaded to Cloud Foundry
Staging: Python buildpack is downloaded and installed
Dependencies: pip install -r requirements.txt runs
Start: Your app starts with python btpaiagent.py
Health Check: Cloud Foundry checks /health endpoint
Route: A URL is created for your app

You'll see output like:

Pushing app btpaiagent to org my-company-org / space dev as your.email@company.com...

Uploading files...

8.52 MiB / 8.52 MiB [======================] 100.00% 2s

Staging app...

Cell 1234 creating container for instance 5678

Creating app with these attributes...

+ name: btpaiagent

path: /path/to/agent

+ memory: 256M

+ disk: 1G

+ instances: 1

Staging process started ...

...

Staging complete

Uploading droplet...

Waiting for app to start...

name: btpaiagent

requested state: started

instances: 1/1

usage: 256M x 1 instances

routes: btpaiagent.cfapps.eu10.hana.ondemand.com

last uploaded: Thu 06 Nov 14:32:15 UTC 2025

stack: cflinuxfs3

buildpacks: python

state since cpu memory disk

#0 running 2025-11-06T14:32:30Z 0.0% 45M of 256M 512M of 1G

Success! Your app is running at: https://btpaiagent.cfapps.eu10.hana.ondemand.com

4.4 Test Your Deployed App

# Test health endpoint

curl https://btpaiagent.cfapps.eu10.hana.ondemand.com/health

# Test quality issues

curl "https://btpaiagent.cfapps.eu10.hana.ondemand.com/qualityIssue?prompt=What%20quality%20issues%20exist?"

# View in browser

# Open: https://btpaiagent.cfapps.eu10.hana.ondemand.com/docs

Step 5: Troubleshooting Cloud Foundry Deployments

Issue 1: App Fails to Start

Symptoms:

App instance crashed

Start unsuccessful

Solution:

# Check logs

cf logs btpaiagent --recent

# Common causes:

# 1. Missing dependencies - check requirements.txt

# 2. Wrong Python version - check runtime.txt

# 3. Port binding error - ensure app uses PORT env variable

# 4. Missing environment variables - check manifest.yml

Fix port binding:

# In btpaiagent.py

port = int(os.getenv("PORT", 8080)) # Cloud Foundry sets PORT

uvicorn.run(app, host="0.0.0.0", port=port)

Issue 2: Health Check Failing

Symptoms:

App has crashed

Health check never passed

Solution:

# Increase timeout in manifest.yml

timeout: 240 # Increase from 180 to 240 seconds

# Check health endpoint locally

cf logs btpaiagent --recent | grep health

Verify health endpoint:

@app.get("/health")

def health_check():

return {"status": "UP"} # Must return 200 status

Issue 3: Environment Variables Not Working

Symptoms:

Missing required AI Core environment variables

KeyError: 'AICORE_BASE_URL'

Solution:

# Check environment variables

cf env btpaiagent

# Set individual variable

cf set-env btpaiagent AICORE_BASE_URL "https://..."

# Restart to apply

cf restart btpaiagent

Issue 4: Out of Memory

Symptoms:

Instance exceeded memory limit

App crashed with exit status 137

Solution:

# Increase memory allocation

cf scale btpaiagent -m 512M

# Check memory usage

cf app btpaiagent

Issue 5: Connection to CAP Backend Fails

Symptoms:

Error fetching issues: Connection refused

API Error: timeout

Solution:

# Check CAP backend URL

cf env btpaiagent | grep CAP_BACKEND_URL

# Update if wrong

cf set-env btpaiagent CAP_BACKEND_URL "https://your-cap-backend.cfapps.eu10.hana.ondemand.com/odata/v4/chemical"

# Restart

cf restart btpaiagent

Test connectivity:

# Add to btpaiagent.py

@app.get("/test-connection")

def test_cap_connection():

try:

import requests

url = os.getenv("CAP_BACKEND_URL")

response = requests.get(f"{url}/ProductBatch", timeout=5)

return {

"cap_backend": url,

"status": response.status_code,

"reachable": response.ok

}

except Exception as e:

return {

"cap_backend": url,

"error": str(e),

"reachable": False

}

Issue 6: AI Core Authentication Fails

Symptoms:

401 Unauthorized

Invalid client credentials

Solution:

# Verify credentials

cf env btpaiagent | grep AICORE

# Update credentials from ai_core.json

cf set-env btpaiagent AICORE_CLIENT_ID "sb-xxxxx"

cf set-env btpaiagent AICORE_CLIENT_SECRET "xxxxx="

cf set-env btpaiagent AICORE_AUTH_URL "https://..."

# Restart

cf restart btpaiagent

Issue 7: Slow Performance

Symptoms:

Requests taking >30 seconds
Timeouts

Solution:

# Scale up instances for load balancing

cf scale btpaiagent -i 3

# Increase memory for better performance

cf scale btpaiagent -m 512M

# Check app metrics

cf app btpaiagent

Add caching to tools:

from functools import lru_cache

import time

cache_timestamp = {}

@tool

def get_products_with_quality_issues(input: str) -> str:

"""Get all product batches with quality issues (Open or Escalated)."""

cache_key = "quality_issues"

current_time = time.time()

# Use cached result if less than 5 minutes old

if cache_key in cache_timestamp:

if current_time - cache_timestamp[cache_key] < 300:

return cached_get_issues()

# Fetch fresh data

result = fetch_quality_issues()

cache_timestamp[cache_key] = current_time

return result

Step 6: Example Production Manifest

Here's a production-ready manifest.yml with all best practices:

File: agent/manifest-production.yml

---

applications:

- name: btpaiagent-prod

memory: 512M

disk_quota: 1G

instances: 2

buildpacks:

- https://github.com/cloudfoundry/python-buildpack.git

command: python btpaiagent.py

random-route: false

health-check-type: http

health-check-http-endpoint: /health

timeout: 180

env:

PYTHONUNBUFFERED: true

LOG_LEVEL: INFO

AICORE_BASE_URL: https://api.ai.prod.eu-central-1.aws.ml.hana.ondemand.com

AICORE_CLIENT_ID: your-production-client-id

AICORE_CLIENT_SECRET: your-production-client-secret

AICORE_AUTH_URL: https://your-subdomain.authentication.eu10.hana.ondemand.com

AICORE_RESOURCE_GROUP: production

CAP_BACKEND_URL: https://cap-prod.cfapps.eu10.hana.ondemand.com/odata/v4/chemical

services:

- my-app-logs

- my-autoscaler

routes:

- route: btpaiagent-prod.cfapps.eu10.hana.ondemand.com

Key production settings:

instances: 2 - High availability with 2 instances
memory: 512M - More memory for production load
services - Bound to logging and autoscaling
PYTHONUNBUFFERED: true - See logs in real-time
Separate resource group - "production" for AI Core

What You've Accomplished

Congratulations! 🎉 You've successfully:

✅ Part 1: Built a robust SAP CAP backend for chemical quality management
✅ Part 2: Learned Python, FastAPI, and LangChain fundamentals
✅ Part 3: Created an intelligent AI agent with custom tools
✅ Part 4: Deployed the agent to SAP BTP Cloud Foundry

Your complete system:

SAP CAP Backend managing production data
AI Agent powered by LangChain and SAP AI Core
FastAPI Service deployed on Cloud Foundry
Natural Language Interface for quality management

Resources

SAP BTP Cloud Foundry:

SAP AI Core:

Python & FastAPI:

Community:

Conclusion

You've completed a comprehensive journey from concept to production deployment! Your AI-powered chemical quality management system is now running in the cloud, accessible to users, and ready for real-world use.

Remember:

Always test in dev/test before production
Monitor your app regularly
Keep dependencies updated
Document your deployment process
Use proper security practices

Thank you for following this series! You now have the skills to build and deploy AI-powered applications on SAP BTP.

Happy deploying! 🚀

Запис вебінару: Технології та інновації як рушійні сили зростання бізнесу

2026-02-16T16:29:26.203000+01:00

Не встигли потрапити на міжнародну онлайн-лекцію "Технології та інновації як рушійні сили зростання бізнесу"?

— не біда!

На каналі Європейської Бізнес Асоціації (EBA) на вас чекає запис вебінару, де експерт Йожек Грушковняк аналізує технології як головний інструмент розвитку бізнесу; розглядає глобальні виклики, зокрема демографічні зміни, економічну фрагментацію та екологічні ризики, що змушують компанії шукати нові шляхи для зростання. Особлива увага приділяється цифровій трансформації та впровадженню систем ERP, які дозволяють оптимізувати ланцюги постачання та фінансові операції.

Щоб дізнатися про це більше — переходьте за посиланням.

Comunidade SAP Joinville em movimento 🌍

2026-02-17T03:32:13.732000+01:00

A Comunidade SAP Joinville segue crescendo!
Se você está chegando agora, seja muito bem-vindo.
Se já faz parte, agradecemos por continuar contribuindo e fortalecendo este espaço.

A SAP Community é aberta a profissionais, estudantes, parceiros e a todos que desejam aprender, compartilhar conhecimento e se desenvolver no ecossistema SAP.

Aqui você pode:

• compartilhar experiências
• publicar conteúdos técnicos
• tirar dúvidas
• aprender com outros membros
• ampliar sua rede de conexões

Se você está visualizando este post, mas ainda não é membro oficial da comunidade, lembre-se:

⚠️É importante criar seu usuário na SAP Community e se inscrever no grupo Joinville para participar ativamente.

🌐Acesse:
https://community.sap.com/t5/joinville/gh-p/joinville

E fique à vontade para divulgar conteúdos e convidar outras pessoas interessadas em SAP a participarem.

💼LinkedIn:
https://www.linkedin.com/company/sap-inside-track-joinville

📸Instagram:
https://www.instagram.com/sitjoinville/

Seguimos fortalecendo nossa comunidade por meio da participação ativa de cada membro.

✅Compartilhe, nos comentários, seus temas de interesse e áreas de atuação no ecossistema SAP.
Se desejar, relate também experiências ou aprendizados que possam contribuir com a comunidade.

I Asked 50 SAP Professionals What Scares Them About AI. Here's What They Said.

2026-02-17T08:55:54.082000+01:00

How This All Started

Picture this. I am sitting in a room full of SAP consultants, project managers, functional leads, and business analysts at a community meetup last quarter. Someone casually drops the words “SAP AI” into the conversation and the entire energy in the room shifts.

Not with excitement. With something closer to anxiety.

That moment stayed with me. So I did something about it. Over the next few weeks I reached out to 50 SAP professionals across industries, experience levels, and geographies. I asked them one simple question:

What honestly scares you about AI in SAP?

No filters. No corporate answers. Just real people telling me real things.

What came back was fascinating, uncomfortable, validating, and hopeful all at once. Let me share it with you exactly as I heard it.

Before We Get Into It, A Quick Question For You

Poll: What is your gut feeling when you hear “SAP AI” right now?

Honestly a little scared
Curious but uncertain
Indifferent, still watching
Genuinely excited

Pick one. We will revisit your answer at the end. You might surprise yourself.

Fear #1: Am I Going to Become Irrelevant?

This came up more than anything else. 31 out of 50 professionals mentioned job security in some form. Not always directly. Sometimes it sounded like this:

“I have spent 14 years mastering SAP FI. If AI can do what I do in seconds, what exactly am I bringing to the table?”

Or this:

“My client literally asked me last month if they still needed my team once Joule was fully live. That question kept me up at night.”

If you felt this way and said it out loud, that took courage. The fear is not irrational. But history shows that every major enterprise technology shift created more specialized roles than it eliminated. SAP itself was once feared as a job killer. Today it supports millions of professionals globally.

AI will not replace the SAP professional who understands the business deeply. It will replace the SAP professional who refuses to evolve.

The question is not whether AI will change your role. It already is. The question is whether you will shape that change or be shaped by it.

A Moment to Reflect

Poll: How worried are you personally about AI affecting your SAP career?

Very worried
A little worried
Not worried, I see the opportunity
Not sure yet

Whatever you picked is valid. You are not alone.

Fear #2: What Happens When AI Gets It Wrong?

This concern came mostly from technical professionals. 24 out of 50 raised accuracy and accountability risks. One senior SAP architect said:

“My worry is not that AI will be wrong. My worry is that it will be wrong confidently. And nobody will check it because everyone assumes the machine is right.”

We have all seen what happens when a financial posting goes wrong in SAP. Month end chaos. Audit questions. Stressed finance teams. Now imagine that at AI speed and scale.

This is a real concern. The answer is not to avoid AI. The answer is governance inside your SAP landscape. Human checkpoints. Confidence thresholds. Exception workflows. These are foundational controls.

Fear #3: Our Data Is a Mess and Nobody Wants to Say It

22 out of 50 professionals mentioned data quality as their hidden fear. One project manager said:

“Everyone in leadership is excited about SAP AI. Nobody wants to be the person who says our master data is a disaster.”

AI amplifies whatever already exists in your SAP system. Clean data gets smarter. Messy data gets messier faster. Years of inconsistent master records cannot be covered by an AI layer.

Before AI deployment, do the foundational work. Run data quality audits. Fix master data governance. Strengthen controls. It is not glamorous, but it works.

Quick Check In

Poll: How would you describe your SAP data quality today?

Clean and governed
Mixed quality
Messy
Not sure

Fear #4: Change Management Will Kill This Before It Starts

19 out of 50 professionals focused on people risk, not technology risk. One change consultant shared:

“I have seen strong SAP programs fail because users never trusted the system. AI is even more personal. If users feel replaced, adoption drops quietly.”

Resistance is often silent. Users override suggestions. Avoid tools. Work around the system. Adoption drops and leadership asks why.

If you are leading SAP AI initiatives, your first meeting should be with end users, not vendors.

Fear #5: Leadership Thinks AI Is a Magic Button

17 out of 50 professionals expressed leadership expectation gaps. One said:

“Our CEO declared we would be AI first this year. We are not even fully on S/4HANA yet.”

There is a gap between boardroom AI excitement and SAP reality. Urgency without readiness creates burnout. Realism without excitement creates inertia. Strong programs balance both.

Implementers must educate upward with clarity and courage.

The Other Side of Fear

After asking about fears, I asked what gives them hope. The tone shifted:

“If AI removes manual reconciliations, I can focus on analysis.”

“Junior consultants adapt quickly. That gives me confidence.”

“I resisted Fiori at first too. Now I cannot imagine working without it.”

Every SAP disruption felt uncomfortable before it felt normal. This one is no different.

Improving the SAP Community together, one comment at a time

2026-02-17T16:21:05.951000+01:00

As a Developer Advocate, I have the privilege of working with some amazing colleagues in that team, but also in the wider Developer and Community Relations organisation. For the size of the community, it's a small group, and the folks are working hard to keep the wheels on the bus and rolling in the right direction. But they can only do so much. In this post, I want to highlight a challenge for us all, and how we can, again, collectively, tackle it together.

The challenge

I'm talking about content quality, which to a greater or lesser degree has for a long time been a topic of debate both in the questions area and also the blog post area. My relationship with the SAP Community platform goes back to day 1, with SDN.

Actually, it goes back to well before that - creating and running the first online SAP community in 1995, and co-creating the SAP Developer Network has caused me to never stop caring for and worrying about those platforms, the people, and the content.

Blog post quality

The quality of some blog posts on the SAP Community platform could be better. I know it. You know it. We all know it. The number of people like you and me who worry about this sort of thing is far outweighed by the larger number of folks that make up the entire community.

Over time various folks have attempted to help educate folks so that they create better content. Here are some posts from me on this topic:

These attempts have worked, but only to a certain extent. Not only that, but we have AI content generation to contend with too. While used appropriately, and with thought, this approach can produce great results. But it can also produce poor content, especially when the human component in this equation is not entirely engaged with the purpose and process. Moreover, the ease with which poor content can be produced is astounding. And so the problem has got worse over time, and I can only see that progression accelerating.

The SAP Community belongs to us

Like you, seeing blog posts that don't belong makes me sad and frustrated. So, what should we do in this situation?

Report important issues to the moderators

Well, first, there's a decision to be made. Does the blog post break the important parts of the Rules Of Engagement? What are the important parts? Well, a bit of common sense will help. Does it defame, injure or harass? Is the content illegal? Report it to the moderators.

Feel empowered to call other issues out yourself

In other cases, don't report it. Use your voice publicly, by attaching a comment to the post, explaining briefly what you consider inappropriate, or just plain incorrect. Encourage your fellow SAP Community members to upvote your comment if they agree.

Use the power of the collective, the power of the shared heart, to highlight, educate, and yes, if you want to think of it in this way, to shame the author into adjusting or removing the post and changing their ways.

This is how society works. And the SAP Community platform is an online society. So feel empowered to treat it as one. Call the person out. Gently, politely, firmly, however you see fit. You be the judge - we're all adults here. Don't treat it as a kindergarten and go to the teacher whenever you dislike something.

Why? Because the teachers are in short supply. The moderators behind the scenes are working flat out on an ever increasing load of reports. Some that I would definitely class as important, but many, many others that I'd class as trivial and inappropriate.

Celebrate good posts

Calling out bad posts like this is important. Upvoting such call-out comments is also important.

But as well as calling out bad posts, remember to celebrate good ones by adding comments of praise or encouragement. Or just upvote them. That is the least you can do, it only takes a second.

The power of filtering and natural selection

Calling out poor content by leaving a comment helps educate, even gently admonish, if needed. But upvoting and adding multiple positive comments helps even more, not only to highlight good content, but also to feed the filter mechanisms that can surface content that you want to see, controlled in part by you.

Final thoughts

Thanks for reading this far. Here's a TL;DR (yes, I know it should go at the top, but think of it as the call to action instead).

Stop reporting poor quality content, and similar transgressions. Call the authors out yourself. Everyone is a member, and everyone is a moderator. And if we moderate ourselves with respect, honesty and firmness, we just might turn this giant sea-going cargo ship towards a better quality future for the community platform that belongs to all of us.

Originally published on qmacro.org

Close Offboarding Processes automatically using Integration Center

2026-02-18T06:09:52.081000+01:00

Introduction

In this blog post, we will walk through how to automatically close offboarding processes in SAP SuccessFactors using Integration Center. Specifically, we will configure a scheduled integration that updates processes based on the employee’s termination date. By the end of this guide, you will understand how to set up the integration, map the required fields, and ensure the process status is correctly updated.

This configuration is useful for us, in the sense that often required tasks are not all completed on the platform and therefore the job responsible for closing the processes does not work, requiring manual cancellation.

1. Choose the Integration Type

The first step is to choose the integration type. In this case, we will use the Scheduled Trigger type and select SuccessFactors as both the source and destination system.

2. Select the Start Entity

For the start entity, search for ONB2Process.

3. Map the Process ID and the default values

In my case, I set the default values for Process Status and Task Status to "Completed". However, this configuration also works with other status values, depending on your requirement.

Important note: The operation must be set to Update/Merge.

Next, we need to map the Process ID. To do this, click on "Set as associated field" and search for Process ID from Process.

4. Define the Logic

In our scenario, we needed to close offboarding processes when the termination date was 30 days in the past, so we used the target date from process that corresponds to termination date :

5. Schedule and Test

After completing the configuration, you only need to schedule the job and test it. 🙂

The expected result is:

Process Status: Completed
Task Status: Closed

Optional: Cancelling Instead of Completing

If instead of completing the process you want to cancel it, you must also map the Offboarding Cancellation Reason field:

Conclusion

In this post, we demonstrated how to configure a scheduled integration in SAP SuccessFactors to automatically complete or cancel offboarding processes. By selecting the correct integration type, mapping the required fields, and applying appropriate date logic, you can fully automate offboarding closure when the required tasks have not been completed. This approach helps ensure data accuracy, reduces manual effort, and keeps your offboarding workflows aligned with business requirements.

SAP Inside Track Joinville - 14 Março

2026-02-18T15:41:04.386000+01:00

Video de apresentação do SIT JLLE

https://youtube.com/shorts/K3M-YkRuqrM?si=NMTSjnWC7huehdIr

A Practical Solution for Multiple Counter Maintenance Plans During SAP PM Data Migration

2026-02-19T06:52:00.857000+01:00

INTRODUCTION

This blog explains how Multiple Counter Maintenance Plans can be uploaded in mass during data migration. We all know that standard tools like LTMC and LSMW is not suitable for Multiple Counter Plans. The upload program is created using a Function Module MPLAN_CREATE. The FM internally creates data in tables such as MPLA, MPOS, and MMPT. The main focus is that we developed in a way that it derives cycle values directly from the cycle set (strategy) and dynamically populating the cycle structure in this FM. This approach made the creation automatic of this complex maintenance plan upload and an alignment with configuration.

BUSINESS REQUIREMENT AND CHALLENGE

We had a business requirement which was to create multiple counter maintenance plans based on cycle sets. During the migration phase, we faced the issue that this requirement proved to be more complex than expected because there is no standard LTMC template available for uploading multiple counter maintenance plans, and LSMW was also not feasible, because it cannot be recorded if there is a scroll as the cycle data becomes difficult to manage when a strategy contains more than four cycle lines. Manual creation was not an option due to the volume of data and the complexity of the cycle definitions.

To address this limitation, we developed a Custom Upload Program using a standard Function Module MPLAN_CREATE. This blog describes how this FM was structured and, how cycle values were automatically derived from the cycle set and populated into the cycle structure of the FM.

MAIN STRUCTURES USED

This Function Module contains the following structures, and we used those structures to develop the upload program.

HEADER – Maintenance plan header
ITEM – Maintenance plan items
CYCLE – Cycle definition
RETURN – Messages

WHY MPLAN_CREATE?

We opted for MPLAN_CREATE because this FM provides full control over Maintenance plan header, Maintenance plan items and Cycle definitions. Instead of hardcoding the cycles, we wanted the system to always read cycle values directly from IP11Z. That way, the upload program would remain stable even if cycle sets change in the future.

OVERALL DESIGN IDEA

The program follows a simple logical flow. That is

First, the maintenance plan header is prepared.
Next, the maintenance plan items are prepared.
Finally, the cycle data is derived from the cycle set and filled into the Function module cycle table.

INPUT PARAMETERS FOR HEADER

The header structure defines how the plan behaves. We defined only the required scheduling-related fields.

Key fields used in our program:

WPTXT – Maintenance plan text
ABRHO – Scheduling period
MPTYP – Maintenance plan category
HUNIT – Unit for Scheduling Period
PLAN_SORT – Sort field for maintenance plans

INPUT PARAMETERS FOR MAINTENANCE PLAN ITEMS

For every maintenance plan there are multiple items linked to the same equipment, but with different task list counters. For every item, we filled the item number, equipment, task list type, task list group, and group counter. Order type, maintenance activity type, and priority were also passed.

One fascinating part here was the work center. The FM does not accept the work center name directly. It accepts the Object ID (OBJID). So, we derived it dynamically by passing the work center and plant into table CRHD and fetching the corresponding OBJID. This value was then assigned to the item structure. The below fields are defined in the item

WPPOS – Item number
PSTXT – Item text
EQUNR – Equipment
PLNTY, PLNNR, PLNAL – Task list assignment
IWERK – Plant
AUART – Order type
ILART – Maintenance activity type
PRIOK – Priority

WORK CENTER DERIVATION

The work center (GEWRK) cannot be passed directly by its name. So, the work center required by the FM was derived dynamically by using the below logic,

Input: Pass Work center (CRHD-ARBL) and Plant (CRHD-WERKS)
Output: Fetch the Object ID (CRHD-OBJID)

This OBJID was passed to the MPLAN_CREATE item structure.

READ AND INPUT THE CYCLE SET (STRATEGY)

Our upload program was developed in a way that cycle set is automatically fetched instead of entering cycle values manually. We passed the Cycle set name to table T351P. This table stores the parameters of each strategy, including the cycle values and units.

Input: Cycle Set Name which was created using IP11Z

Table used: T351P (Maintenance Packages)

Data Fetched from T351P

ZAEHL – Cycle number
ZYKZT – Cycle value
ZEIEH – Unit of measure
KTEX1 – Short text

By doing this, we ensured that the program always uses the same cycle definition as maintained in IP11Z. For every line found in T351P, one line was created in the MPLAN_CREATE cycle table CYCLE.

The fields were filled as follows:

The cycle number was taken from the cycle set sequence.
The cycle value was taken from the cycle definition in the cycle set.
The unit of measure was taken from the cycle set unit.
The cycle text was taken from the cycle set short text.

This mapping paved us to create multiple cycles for the same maintenance plan without manual entry.

Fields populated:

NUMBER → From T351P-ZAEHL
ZYKL1→ From T351P-KTEX1(fetch the first word, as the value is entered in the first word of the text in cycle set)
ZEIEH → From T351P-ZEIEH
PAK_TEXT → From T351P-KTEX1

COUNTER POINT DETERMINATION

As this was a counter-based maintenance plan, we had to determine the counter point dynamically. The equipment number was first passed into EQUI table to fetch the object number. That object number was then passed into IMPTT table to fetch the counter point. We selected the counter by filtering the characteristics name which represented the operating hours. This counter point was then assigned to the relevant cycle entries in CYCLE. This step ensured that the cycle was linked to the correct measurement counter of the equipment.

Pass EQUNR → table EQUI → Fetch OBJNR
Pass OBJNR → table IMPTT → Fetch POINT → Filter where ATINN = 'HOURS'

This POINT is passed into CYCLE-POINT.

SEQUENCE AND REPETITION LOGIC

To support multiple counter groups, we used:

CYCLESEQIND to define the cycle group
SETREPEATIND to indicate repetition

OUTCOME

The outcome of this approach is,

Multiple counter maintenance plans were created automatically
Cycle data was always aligned with strategy configuration
No dependency on LTMC or LSMW was required
Manual creation effort was eliminated

The solution proved to be stable, reusable, and fully controlled.

CONCLUSION

MPLAN_CREATE becomes a practical and powerful alternative when standard migration tools do not support complex PM scenarios such as Multiple Counter Maintenance Plans with cycle sets. By deriving cycle values from the table T351P and counter points from the table IMPTT, we were able to automate the creation of upload program for Multiple Counter Maintenance Plan, while keeping the logic aligned with SAP configuration. This approach helped us to bridge the gap between functional configuration and technical automation in a clean and maintainable way.

SAP Community - SAP Community

Український бізнес у режимі змін: інсайти від керівної директорки SAP Ukraine

Процеси та гігієна даних

Оборонка, кібербезпека та ментальне здоров’я

Переходьте до перегляду повних інтерв'ю за посиланням:

SAP PO to SAP CI Migration - Simplifying File-Based Scenarios with SMB Adapter in SAP CI

Solution :

High-Level File Based Architecture Explanation:

Section A – Preparing the SMB Share

Section B – Cloud Connector Configuration

Section C – Setup SMB Adapter in SAP CI Iflow

Section D - Demo:

Conclusion:

Будуй майбутнє з SAP: стажування в команді Sales

Ти ще встигаєш, стажування в SAP Ukraine чекає на тебе!

SAP Community Voice: SAP BW News, Ask Anything to SAP Champions, and Events from Tokyo to Brazil!

The 2025 Developer Insights Survey: The Report

Part I: Introduction and General Information​

Key Findings from the 2025 Survey​

General Topics​

Employment Status and Relation to SAP​

Employment

SAP Developer Community Age Distribution

Where are the Developers?​

Part II. Narrowing the Response Set

Workplace and Trends

Recent SAP Project Experience?​

Popular Programming Languages​

Popular Programming Frameworks​

Application and Business Technology Platform Architectures​

Popular IDEs​

Potential / Use of Generative AI for Developers​

Joule in SAP Build Code

Development Job Roles

Development Domains​

Developer Satisfaction with SAP​

BTP Guidance Framework

Cloud Insights​

Low / No Code Tools​

Learning and Help Resources​

Satisfaction with SAP Learning Hub

Survey Methodology

Other References​

Highlights from SAP Inside Track Bengaluru – 14th June 2025

Highlights from SAP Inside Track Bengaluru – 14th June 2025

🌟 Keynote That Set the Tone

💡 Spotlight Session – Devraj from IBM

🤝 Community Meet-up – IBM Team & Sindhu

📸 Event Gallery – Capturing the Energ

🧩 Let’s Keep the Momentum Going!

Bienvenidos al grupo de SAP Community Mexico

Qué puedes esperar aquí

Tono de la comunidad

Preséntate 👋

What you can expect here

Community tone

Say hello 👋

Bangalore Stammtisch - Jan 29

Mini Session 02 - SAP COMMUNITY KOLKATA - 2026

🌐Unlock SAP’s Core Tech – From ABAP to AI

🔹Topic 1: Diving into SAP’s Backend World with ABAP

🔹Topic 2: Trends and Research in Enterprise Artificial Intelligence

👥Who Should Attend?

🌟Why You Should Join

📢Stay Connected

📞Contact for Assistance

📍Event Details

Building an AI-Powered Chemical Quality Management System: Part 1 - SAP CAP Backend Setup

Building an AI-Powered Chemical Quality Management System: Part 2 - Python, FastAPI & LangChain Fund

AI-Powered Chemical Quality Management System: Part 3 - Creating AI Agent with LangChain & FastAPI

Building an AI-Powered Chemical Quality Management System: Part 4 - Deploying to BTP Cloud Foundry

Запис вебінару: Технології та інновації як рушійні сили зростання бізнесу

Comunidade SAP Joinville em movimento 🌍

I Asked 50 SAP Professionals What Scares Them About AI. Here's What They Said.

How This All Started

Before We Get Into It, A Quick Question For You

Fear #1: Am I Going to Become Irrelevant?

A Moment to Reflect

Fear #2: What Happens When AI Gets It Wrong?

Fear #3: Our Data Is a Mess and Nobody Wants to Say It

Part I: Introduction and General Information

Key Findings from the 2025 Survey

General Topics

Employment Status and Relation to SAP

Where are the Developers?

Recent SAP Project Experience?

Popular Programming Languages

Popular Programming Frameworks

Application and Business Technology Platform Architectures

Popular IDEs

Potential / Use of Generative AI for Developers

Development Domains

Developer Satisfaction with SAP

Cloud Insights

Low / No Code Tools

Learning and Help Resources

Other References