Loopgraph is an open-source tool designed for loop engineering in autonomous companies, transforming recurring corporate workflows into governed, visual AI loops. Moving beyond basic prompt chains and direct webhooks, Loopgraph introduces a structured approach to automation by separating reasoning from authority.

With Loopgraph, teams can interview a department to generate versioned LoopSpecs. Its routing component, Hermes, handles event routing to the appropriate loop. The system provides tools to validate, simulate, approve, trace, and thoroughly review every outcome. Crucially, Hermes acts as a proposer while a deterministic policy engine decides what is permitted to run, ensuring safety and compliance across operations.

Built to be completely open-source, local-first, and model-agnostic, Loopgraph provides developers and organizations with robust infrastructure to orchestrate autonomous workflows with enterprise-grade governance.

Pros of Loopgraph

Separates AI reasoning from authority using a deterministic policy engine
Open-source, local-first, and model-agnostic architecture
Generates versioned LoopSpecs through departmental interviews
Comprehensive capabilities for simulation, validation, tracing, and review

Cons of Loopgraph

Requires technical setup as a developer tool
May have a learning curve for defining LoopSpecs and deterministic policies
Relies on external large language models (bring-your-own-model approach)

Frequently Asked Questions

What is Loopgraph?

Loopgraph is an open-source tool that turns recurring company work into governed, visual AI loops. It separates AI reasoning from authority, allowing Hermes to propose actions while a deterministic policy decides what may run.

Is Loopgraph open-source?

Yes, Loopgraph is open-source, local-first, and model-agnostic, available via GitHub for developers and autonomous companies.

How does Loopgraph handle security and governance?

Loopgraph separates reasoning from authority. While Hermes (the event router) proposes actions, a deterministic policy engine strictly decides what processes are allowed to run.