Open foundations for trustworthy AI
The trust layer beneath our products is open. Machinise K.K. commercializes the open-source research of QMI Lab — so how our agents reason, and how you verify it, are things you can read.
What we build in the open
Lodestar
SHIPPEDThe open-source reasoning-chain engine for AI agents. Lodestar records how an agent reaches a decision — the chain of reasoning, the tools it called, and the evidence behind every action — so it can be inspected and verified. It powers Machinaut and integrates with MachineCraft.
Asterism
SOONQMI Lab's second open-source project. Commercialization through Machinise K.K. is coming soon — follow the work on GitHub.
Lodestar and Asterism are open-source projects by QMI Lab
Open research, in production
Powers Machinaut
Machinaut's audit and reasoning trail runs on Lodestar. The engine that records what an agent did is open, so the evidence is something you can verify.
Integrates with MachineCraft
Agents and workflows built in MachineCraft plug into Lodestar's reasoning chain, carrying the same trail from design through to deployment.
Build on open foundations
Use the open source, or talk to us about putting it into production with Machinaut and MachineCraft.