--- sidebar_position: 1 title: Getting Started description: Entry point for deploying the Physical AI Toolchain author: Microsoft Robotics-AI Team ms.date: 2026-06-12 ms.topic: overview keywords: - getting-started - quickstart - deployment - onboarding --- Deploy the Physical AI Toolchain and submit your first training job. This hub guides you through setup, deployment, and verification. The default path starts on a laptop, not in the cloud. Begin with [Start Here — T0 Dev](#start-here--t0-dev), then graduate to higher tiers only when your scale demands them. ## Choose Your Tier Adoption is modeled as six graduated tiers (T0-T5). Each tier states the minimum infrastructure needed to complete the full training lifecycle: capture demonstrations on a robot, train an imitation policy, validate it, and run that policy back on the robot. Each tier is a legitimate stopping point. Start at T0 and graduate only when a concrete trigger forces it. | Tier | When to start here | Graduate when… | Quick start | |---------------------|----------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------|---------------------------------------------------------------| | **T0 — Dev** ⭐ | Default. One laptop, one robot; zero cloud and zero Kubernetes. | No local GPU; the task needs many training iterations as conditions vary; or a second person needs the data. | [Tier 0 — Dev](../recipes/tier-0-dev/README.md) | | **T1 — Lab** | One site, a few robots, a shared GPU box; first cloud storage. | Training scale or team size outgrows one GPU box; dataset governance and catalogs become necessary. | [Tier 1 — Lab](../recipes/tier-1-lab/README.md) | | **T2 — Pilot** ✅ | Recommended production. One site at scale; cloud training default. | The robot count or update cadence makes hand-updating each robot error-prone and version skew real, while everything is still at one reachable site. | [Tier 2 — Pilot](../recipes/tier-2-pilot/README.md) | | **T3 — Production** | Advanced. Single-site declarative deploy (local k3s + Flux, no Arc). | Robots span multiple sites, or sites become unreachable from a single operator network. | [Tier 3 — Production](../recipes/tier-3-production/README.md) | | **T4 — Scale** | Advanced. Multi-site **fleet delivery**; Arc reachability broker. | You explicitly want production signals to drive retraining and fleet-wide health analytics. This is a deliberate decision, not an automatic consequence of scale. | [Tier 4 — Scale](../recipes/tier-4-scale/README.md) | | **T5 — Operate** | Roadmap. **Fleet intelligence** for drift detection and retraining. | Available after the roadmap implementation lands. | [Tier 5 — Operate](../recipes/tier-5-operate/README.md) | ⭐ default · ✅ recommended production For the tier-by-tier infrastructure boundaries see the [Architecture Overview](../contributing/architecture.md). Jump to [T0 — Dev](../contributing/architecture.md#t0--dev), [T1 — Lab](../contributing/architecture.md#t1--lab), [T2 — Pilot](../contributing/architecture.md#t2--pilot), [T3 — Production](../contributing/architecture.md#t3--production), [T4 — Scale](../contributing/architecture.md#t4--scale), or [T5 — Operate](../contributing/architecture.md#t5--operate). See the canonical [Tier Model](../design/tier-model.md) for the authoritative tier table and vocabulary. > [!NOTE] > **Roadmap honesty.** T5 (Operate / fleet intelligence) is on the roadmap and not yet available. The fleet-intelligence domain is currently specified, with implementation planned. Today's shipping capability spans T0-T4. ### Start Here — T0 Dev The default starting path is **one laptop and one robot**, with zero cloud and zero Kubernetes. You close the full capture -> train -> validate -> run loop entirely on local hardware. 1. **Set up:** clone the repo and run `./setup-dev.sh` (Python 3.12 via `uv`, virtual environment, training dependencies). No Azure subscription required. 2. **Capture:** record ROS 2 bags to local disk on the robot or laptop. 3. **Move data:** `cp` or `rsync` from robot to laptop. 4. **Curate:** run the dataviewer in `local` mode on the laptop. 5. **Train:** run `lerobot-train` on the laptop (CPU or a local GPU). 6. **Track:** training outputs are written to local disk; hosted experiment tracking enters at T2. 7. **Validate:** run `run-local-lerobot-eval.py` / `play.py` locally. 8. **Run on robot:** launch the inference node as a plain process or container. No Flux, no gating, no GitOps. **Edge infra:** ROS 2 and Docker only. **Cloud infra:** none. See the [Tier 0 — Dev recipe](../recipes/tier-0-dev/README.md) for the step-by-step walkthrough. > [!TIP] > **Graduate when** you have no local GPU, the task needs many training iterations as conditions vary, or a second person needs the data. At that point, step up to [Tier 1 — Lab](../recipes/tier-1-lab/README.md) (first cloud storage) or jump straight to the recommended production path, [Tier 2 — Pilot](../recipes/tier-2-pilot/README.md) (cloud training). The [Quickstart](quickstart.md) covers the cloud (T2 — Pilot) path end to end. ## 🚀 Guides | Guide | Description | |--------------------------------------------|------------------------------------------------------------| | [Start Here — T0 Dev](#start-here--t0-dev) | Default local-first path: laptop + one robot, no cloud | | [Choose Your Tier](#choose-your-tier) | Pick a tier and see its graduation triggers | | [Quickstart](quickstart.md) | Cloud path (T2 — Pilot): clone to the first cloud training job | | Architecture Overview (coming soon) | System topology, components, and data flow | | Glossary (coming soon) | Term definitions for Azure, NVIDIA, and OSMO | ## ⏱️ Time and Cost The local default path (T0 — Dev) has **no cloud cost**. It runs entirely on your laptop. The estimates below apply to the cloud path ([Quickstart](quickstart.md), T2 — Pilot and up). | Item | Estimate | |-----------------------|--------------------| | Total deployment time | ~1.5-2 hours | | Quick validation cost | ~$25-50 | | GPU VM rate | ~$3.06/hour (A100) | > [!NOTE] > Run `terraform destroy` when finished to stop incurring costs. See [Cost Considerations](../contributing/cost-considerations.md) for detailed estimates. ## 📋 Prerequisites Summary The default path (T0 — Dev) needs only **Python ≥3.12** plus ROS 2 and Docker, with no Azure subscription and no Kubernetes tooling. The additional tools below are required only for the cloud path ([Quickstart](quickstart.md), T2 — Pilot and up). | Tool | Version | Required for | |-----------|---------|------------------------| | Python | ≥3.12 | All tiers (incl. T0) | | Terraform | ≥1.9.8 | Cloud path (T2+) | | Azure CLI | ≥2.65.0 | Cloud path (T2+) | | kubectl | ≥1.31 | Kubernetes tiers (T3+) | | Helm | ≥3.16 | Kubernetes tiers (T3+) | For the cloud path, an Azure subscription with Contributor + User Access Administrator roles, GPU quota for `Standard_NC24ads_A100_v4`, and an NVIDIA NGC account are required. See [Prerequisites](../contributing/prerequisites.md) for full details. ## 📚 Related Documentation | Resource | Description | |---------------------------------------------------------------------------------------------------|-----------------------------------------| | [Contributing Guide](../contributing/README.md) | Development workflow and code standards | | [Deployment Guide](https://github.com/microsoft/physical-ai-toolchain/blob/main/deploy/README.md) | Detailed deployment reference | | [Cost Considerations](../contributing/cost-considerations.md) | Pricing breakdown and optimization |