Norvia Labs is a collection of focused experiments for people building, testing, and operating AI agents.

Each project tackles one practical problem:

• forge runs agent tools, loops, and policies with deterministic replays.
• guard adds governance, identity, and execution safety to agent traffic.
• contract turns existing APIs into tool schemas agents can use.
• route reduces prompt cost and latency by pruning unused tool schemas.
• audit records agent actions in a structured, cryptographic ledger.

Projects are built in the open, with MIT licensing as they ship, and documented as they develop.

Whether you are prototyping an agent, connecting it to existing systems, or adding controls before it reaches production, start with the experiment portfolio.

The GitHub organization has the source code, issues, and project updates.

• Explore the experiments
• Read the thesis
• Get in touch