← back
A Piece of Pi: Embedding The OpenClaw Coding Agent In Your Product — Matthias Luebken, Tavon
Takeaway
Coding agents become product primitives when you reuse their tool-loop + shell runtime with thin extension APIs for UI and session events.
Summary
- Speaker explores Pi (minimal coding agent from OpenClaw, Mario joining Arendelle) as a template for embedding coding agents into products beyond dev tools.
- Examples: Claude Desktop's Excel skill wraps pandas/openpyxl/LibreOffice CLIs to manipulate Excel — showing 'small tools, big skill' composition.
- Demos a CRM lead-qualifier agent: TypeScript class, before-tool-call hooks for RBAC/authorization, event subscriptions for tool-call observation.
- Extends to coding-agent territory with bash runtime, session-event and UI-interaction APIs, and a /pipeline slash command using context.ui.select to mix backend tools with user UI selection.
coding-agentsopenclawsdk
Original description
OpenClaw feels like it's learning: it discovers capabilities, stitches tools together, builds solutions it wasn't explicitly taught. The reality is simpler — it's an LLM calling tools in a loop, powered by Pi, a minimal coding agent SDK. This talk is about what you can build once you understand that. Matthias Luebken walks through embedding Pi in a real product: a B2B sales pipeline where incoming RFP emails route to customer-specific agent sessions, CLIs expose CRM and ERP data in a form the agent can use cleanly, and the only output a human sees is a draft in their inbox. The architectural principle running through it: don't fight the coding agent, make things easy for it. Design your data access and tool interfaces so the agent can work naturally rather than having to compensate for complexity. Speaker info: - https://x.com/luebken -https://github.com/luebken