Guides¶
Deep dives into FastroAI's core components.
Each guide explains what a component does, when you'd use it, and how to get it working. Start with whatever problem you're trying to solve.
-
FastroAgent
Wrap PydanticAI agents with automatic cost tracking. Stateless, production-ready, with consistent response formats.
-
Cost Calculator
Track token costs in microcents for exact billing. Override pricing for volume discounts or custom models.
-
Pipelines
Chain multiple AI steps with automatic parallelization. Track costs across entire workflows.
-
Safe Tools
Timeout, retry, and graceful error handling for AI tools. Keep requests alive when external services fail.
-
Tracing
Correlate AI calls with the rest of your request flow. Integrate with any observability platform.
Where to Start¶
Not sure which guide to read?
Building an AI feature? Start with FastroAgent. It gives you cost tracking on every AI call.
Multiple AI steps? Read Pipelines for execution order, parallelization, and aggregated costs.
Tools calling external APIs? Check Safe Tools for graceful degradation when services fail.
Going to production? Set up Tracing to debug slow or expensive requests.
Most projects start with FastroAgent - it gives you cost tracking on every AI call, which you'll want in production. As your application grows, add pipelines for multi-step workflows and safe tools for external service calls.