Delegate the outcome.
Keep the judgment.
Henry is an autonomous operations agent that plans, acts, sleeps, asks and proves—without turning your day into another chat thread.
Outcomes arrive messy.
Henry makes them executable.
A natural-language objective becomes an authority envelope: what may happen, what must never happen, when you must be asked, and how much the attempt may cost.
It sees the work,
not just the request.
Henry reads constraints against the calendar, records evidence and chooses a route. Every decision remains observable, timestamped and reversible.
Watch one mission
move through time.
The performance below is deterministic and self-contained. It will begin when it enters the frame.
Constraints understood
30 minutes · before Thursday · protect Tuesday focus
Autonomy stops
where judgment begins.
Henry suspends without polling, then asks once at the consequential boundary. Rejection activates a safe fallback—not a hidden retry.
Completion is a claim.
Verification is evidence.
Henry reads the result back against the original constraints before declaring success. The event, approval, cost and full execution history become durable proof.