Microsoft gives it away free. Here is what you still build and run.
The Agent Governance Toolkit is a real, well-built open-source project and a good way to learn the shape of the problem. It is a public-preview toolkit you wire into your own agents - which means the production pieces, state, policy lifecycle, and operations, are yours to stand up. Swiftward is the finished on-prem product.
Library you assemble, versus product you deploy
- It is a library and SDK, not a deployable product. You integrate the toolkit into your agent code and operate it. Swiftward is a single binary plus Postgres with the console, gateways, and audit already in place.
- Its kernel is stateless by design, so you bring and run your own state store (Redis, Kafka, and the like) for counters, rate limits, and trust scores. Swiftward ships that state in Postgres, ACID and replay-safe, with nothing to wire up.
- Versioned policy-as-code is the part we add: immutable versions, one-click rollback, shadow mode and backtesting against real traffic, and replay of a past decision on the exact version that was live. The toolkit governs at runtime; it is not a policy-lifecycle product.
- Human-in-the-loop in Swiftward is a durable, audited workflow: a review case persists, and its resolution re-enters the same audited pipeline as a recorded event.
What the toolkit does well
It is free, MIT-licensed, has framework adapters and SDKs, real runtime enforcement, and Microsoft's attention validates that this is a real category. If your need is a prototype or a single team standing up governance from parts, it may be exactly right. We will tell you that on the call rather than pretend otherwise.
The honest line
Free software is not free to run in production. The toolkit gives you the map and the building blocks; Swiftward is the finished engine that ships state, policy versioning, replay, and a tamper-evident audit trail, so you deploy them rather than assemble and operate them yourself. See the capability comparison.