Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Fixpoint: Stateful AI Workflows
Learn how Fixpoint enables building recoverable, multi‑step AI workflows with explicit state management, agent coordination, caching, and a human‑in‑the‑loop UI.
We’ve been making an open-source project for building stateful, multi-step AI workflows. https://github.com/gofixpoint/fixpoint/
The use-case: you have LLM workflows that have multiple states/steps, and you want to make your workflow recoverable and easier to control. We let devs connect multiple LLM agents (or just prompts) and keep track of agent + workflow state (memory, cache past steps). We just added a human-in-the-loop UI.
Fixpoint focuses on giving the programmer tight control over what each AI is doing at each step of the workflow (compared to frameworks like Crew AI, which are a bit more autonomous). You can think of us kind of like someone took SQS or Temporal, and super-charged it with AI-specific features.
Fixpoint automates remote knowledge worker sourcing and screening via flexible APIs.