Build, deploy, and monitor AI workflows as DAGs. Built in Rust for production reliability.
A complete platform for building and running AI agent workflows
Define workflows as directed acyclic graphs. Parallel execution, conditional branching, and human-in-the-loop all built in.
No memory leaks, no GC pauses. Run workflows for hours or days without degradation. Same foundation as Cloudflare Workers.
Execution traces for every step. See exactly what happened, when, and why. Debug failures with rich context, not stack traces.
Pause and resume workflows at any step. Survive crashes, handle long-running tasks, add human approvals.
Organizations, team members, API keys, usage tracking. Everything you need for a production platform.
Create workflows with JSON or the visual builder. Run them via REST API. No 9-page concept guides.
Use the visual builder or JSON to create your workflow as a DAG. Add LLM calls, tool executions, parallel branches, and conditional logic.
Save your workflow and run it via API. Get an execution ID to track progress. Workflows checkpoint automatically.
See execution traces for every step. View inputs, outputs, timing, and costs. When things fail, get the context you need to fix it.