Petal Labs builds modular, composable tools for agentic AI systems. Our mission is to empower developers to orchestrate, evaluate, and refine intelligence across models, providers, and workflows — with clarity, transparency, and control.
We believe AI should feel organic: systems that grow, adapt, and interconnect rather than lock you into rigid pipelines. Each tool in the Petal ecosystem is designed like a petal in a larger bloom — independent, purposeful, and capable of forming something extraordinary when combined.
A perception and evaluation engine for multi‑LLM workflows.
- Compare and rank model outputs
- Run structured evaluations
- Build feedback loops for refinement
A lightweight orchestration layer for agentic systems.
- Compose multi‑step flows
- Fan‑out across providers
- Route intelligently based on context
Every tool stands alone — and works even better together.
Use the models you want. Mix and match freely.
No black boxes. Every decision can be inspected, measured, and improved.
Clear APIs, thoughtful defaults, and tools that stay out of your way.
Each project includes its own documentation, examples, and quickstart guides.
Explore the repos to begin building with Iris and PetalFlow.
Petal Labs is building the foundation for a new era of agentic development — one where intelligence is not a monolith but a garden. A place where ideas bloom, tools interconnect, and creators shape their own ecosystems of AI.
If you’re building the future of agentic systems, orchestration, or multi‑model workflows, you’re in the right place.
✨ Build intelligence, petal by petal.