W12: Data Ingestion

Eric Hubbell

Overview
This week we migrated our ingestion pipeline to an AI-driven approach. It is mostly an internal change, but it matters because ingestion sits near the front of everything else Playbooks does after a project is connected.
If we can understand a play more accurately and with less overhead, we can make the rest of the system faster, more flexible, and easier to scale.
What's New
- Replaced the older synchronous mapping flow with an AI-driven ingestion pipeline
- Added parallel processing for more of the ingestion workload
- Improved how plays are mapped into the Playbooks taxonomy
- Improved how deployment defaults are generated for incoming projects
- Reduced manual and template-heavy internal processes
Better Project Understanding
The new pipeline uses frontier models to analyze a connected project and map it into the Playbooks ecosystem. That includes assigning the frameworks, languages, platforms, packages, tools, and topics that help make a play more discoverable and better understood across the product.
This gives us a more flexible ingestion layer than the older rule-heavy approach, especially as the range of supported projects continues to expand.
Better Deployment Defaults
We also moved beyond a narrower template-based system for generating deployment configuration. The new ingestion flow can produce more lightweight, structured defaults for getting a play ready to run on our infrastructure.
That should translate into:
- fewer manual steps on our side
- better default accuracy
- a cleaner path from connected source to runnable demo
Why It Matters
Most users will not notice this change directly, but they should feel it indirectly over time. Smarter ingestion means less friction when publishing, fewer edge cases to handle manually, and a stronger foundation for scale as more projects move through the platform.
That's all for now
This is one of those updates that is more foundational than flashy. We’re excited about it because it simplifies our internal systems while making Playbooks more capable behind the scenes.
Related Articles
View
Changelog
W14: Introducing MCP
This week we introduced `@playbooks/mcp`, a new MCP server for Playbooks. It lets frontier models and coding agents connect to the same Playbooks workflows already available through the CLI.

Changelog
W10: Introducing Widgets
This week we introduced Widgets, a new way to embed Playbooks outside the core app. Widgets make it easier to place plays across websites, documentation, and READMEs with lightweight, copy-pasteable components.

Changelog
W8: Introducing Partials
This week we introduced Partials, a new way to package and deploy smaller units of code on Playbooks. Partials make it easier to share components, sections, and other reusable pieces without wrapping everything in a full starter or stack.