Vetting tells you what a repo can do.
Governance tells you what happens next.
The Vetted Registry is static analysis -- what a repo exposes, and whether it's reachable. IAIso is the runtime layer: what actually happens once that repo is running as part of an agent.
Vetting + Governance
Two layers, not one bigger one
A vetted repo tells you what's possible. A governed runtime tells you what's actually allowed to happen -- and stops it automatically if it doesn't.
- The challenge
A 12-person team drowning in status updates, scattered docs, and manual handoffs.
- With SmartTasks
One workspace: whiteboard the plan, let personas draft, and a workflow ships the busywork.
- The result
The team meets less and ships more — every update writes itself.
Hours saved / week
0.0
Whiteboards · Personas · Workflows
IAIso's five invariants
Bounded pressure
Compute, tokens, and tool calls accumulate against a measured ceiling -- enforced at the infrastructure level, not requested of the model.
No learning across resets
Every safety release is a lossy wipe. Anything that needs to persist across it requires explicit, scoped consent.
Clocked evaluation only
Safety checks run at discrete intervals. There's no continuous, ungoverned loop for pressure to hide inside.
Consent-bounded expansion
Every tool call carries a cryptographically signed, scoped, expiring token. Nothing runs on implied permission.
No proxy optimization
Pressure is computed at the OS level, outside the model's own reasoning -- so it can't be reasoned around.
What actually happens when an agent's pressure crosses the safety threshold?
Want this on a repo you're already running?
We can help you wrap an existing agent stack with governance, or vet a repo you're considering deploying -- as a one-off engagement or ongoing work.
IAIso is open -- the framework and its Python SDK are publicly available. What we offer is the expertise to calibrate and deploy it correctly for your actual workload.