Jidoka Labs · The lab behind Jido

Building reliable AI systems on the BEAM.

Jidoka Labs is the AI systems lab behind Jido, the open-source agent framework for Elixir and the BEAM. We build tools, patterns, and reference systems for agents that must execute useful work, recover from failure, and remain under human control.

Project
Open source
Language
Elixir
Runtime
BEAM · OTP
Focus
Reliability
JIDO / RUNTIME
Supervised

Execution trace

RUN · 0042

  1. receive_signal 12ms
  2. plan_actions 38ms
  3. execute_action Running
  4. request_review Queued

State

Durable

Events

Visible

Failure

Recoverable

Built for Elixir · OTP Human control enabled

Flagship open-source project

Jido

View the project on GitHub

The agent framework at the center of the lab.

Jido gives Elixir engineers a framework for agent behavior, actions, signals, and runtime coordination. It brings the strengths of the BEAM to systems that must keep working when real conditions change.

  • Open source
  • BEAM native
  • Observable

Our engineering position

Reliability is not a feature added at the end.

It is the shape of the system. Jidoka Labs works on the runtime concerns that turn model output into dependable execution.

Clear execution

Agents need defined actions, durable state, and visible progress. Reliable work starts with a system you can inspect.

01

Recovery by design

Failure is part of the runtime. Supervision, retries, and recovery paths belong in the architecture from the start.

02

People in command

Human review is a system capability. People set the rules, inspect the work, and make the decisions that matter.

03

Work from the lab

Build in the open. Prove the work.

The lab turns its point of view into public software, working systems, and useful technical guidance.

Open source

Jido

The agent framework for Elixir and the BEAM. Jido is the foundation for the lab's work on supervised, observable, and recoverable agents.

View Jido on GitHub
Applied engineering

Reference systems

Focused implementations that turn runtime ideas into patterns engineers can understand, test, and use in real systems.

In development

Shared learning

Research and field notes

Practical writing about agent architecture, evaluations, context, supervision, and the work required to reach production.

Read from the lab

Selected collaboration

Start with the technical fit.

Jidoka Labs works with a small number of teams whose needs align with Jido and the lab's technical direction.

  • Jido adoption and architecture
  • Agent-system production readiness
  • Reference systems and applied research
  • Open-source sponsorship

Mike Hostetler is the founder of Jidoka Labs and the creator of Jido. He has spent more than three decades building software, companies, open-source communities, and engineering teams.

Tell us what you are building.

Share the system, technical constraint, or Jido question you want to discuss.