# Core Concepts ## What is GoalfyData? GoalfyData is a runtime platform for AI Agents to build data assets. The Agent creates the assets; GoalfyData keeps them running. ## Three-Layer Architecture: Build -> Run -> Share ### 1. Build -- Enable Any Agent to Build Data Assets GoalfyData does not replace your AI Agent -- it enhances it. Through Skill / MCP / CLI, GoalfyData integrates into your existing Agent workflow: - Claude Code - Codex - Manus - ChatGPT / Custom GPTs Your Agent can naturally produce: - **Datasets** -- Structured data generated from APIs, files, or databases - **Update Scripts** -- Python / TypeScript scripts that automatically refresh data - **Templates** -- Reusable data structures and configurations - **Data Apps** -- Visual data applications built by the Agent **Key Insight**: Users don't need to learn a complex platform first. They build assets directly within the Agent they already use. ### 2. Run -- Keep Assets Running and Updated This is the core differentiator between GoalfyData and every other tool. Agent outputs are typically ephemeral: - Scripts that only run locally - CSVs that expire within a week - Dashboards that stop working when you close the tab GoalfyData transforms them into continuously running services: - **Scheduled Sync** -- Cron-based automatic data refresh - **Sandbox Execution** -- Update scripts run in isolated environments - **Version Control** -- Track changes and roll back when needed - **Atomic Updates** -- full_replace mode ensures no empty-table intermediate states Agents can build anything, and GoalfyData keeps it running. ### 3. Share -- Secure, Controlled, Cross-Agent Reuse Data assets only generate real value when they can be shared and reused. GoalfyData transforms scattered files into manageable, shareable assets: - **Share with Colleagues** -- Role-based access control - **Share with Clients** -- Read-only or preview permissions - **Team Sharing** -- Organization-level data assets - **Share with Other Agents** -- Agent-to-Agent data flows Permission levels: View / Preview / Query / Update / Manage **Problem Solved**: No more emailing CSVs, no more repeatedly explaining data definitions, and no more rebuilding the same dataset in every new Agent conversation. ## What GoalfyData Is Not Understanding boundaries is just as important as understanding value. | GoalfyData Is Not | It Is | |---|---| | An AI Agent | A runtime platform for data assets built by Agents | | A BI Tool (Tableau/Looker) | A platform where Agents build data apps, not humans manually | | A Lightweight dbt | A hosting layer for Agent-generated data pipelines | | An MCP Tool/Server | A persistence layer for data after MCP connects it | | A General App Host (Vercel) | A hosting platform designed specifically for Agent-built data assets | GoalfyData is the infrastructure layer that every AI Agent needs, but no single Agent provides. ## Why It Matters Now Three major trends are converging: 1. **Agents Are in Production** -- 57.3% of organizations are already running Agents (LangChain Survey 2026) 2. **MCP Is Becoming a Standard** -- 28% of Fortune 500 companies have adopted MCP as infrastructure 3. **Output Persistence Is the Gap** -- Everyone talks about agent memory, but no one solves the problem of making agent outputs persist GoalfyData fills exactly this gap. ## Architecture Overview ``` AI Agents (Claude / Codex / Manus) | | Build: Skill / MCP / CLI v +----------------------------------+ | GoalfyData | | | | +-----------+ +--------------+ | | | Datasets | | Data Apps | | | +-----+-----+ +------+------+ | | | | | | +-----v-----+ +------v------+ | | | Scheduled | | Version | | | | Sync | | Control | | | +-----+-----+ +------+------+ | | | | | | +-----v---------------v------+ | | | Sandbox Execution | | | +-------------+---------------+ | | | | | +-------------v---------------+ | | | Sharing & Access Control | | | +-----------------------------+ | +----------------------------------+ | v Teams / Clients / Other Agents ```