Launched this week
Midfleet
Govern autonomous agent work from intent to evidence
9 followers
Govern autonomous agent work from intent to evidence
9 followers
Give agents a mandate, not a script. Midfleet keeps intent, scope, rules, ownership, approvals, blockers and evidence attached to autonomous agent work while agents decide how to get it done. Now self-serve with a free plan and no credit card.
BetterClaw
@smashadani Congratulations on your launch, here to support you!!
@better_shaya Thanks Shaya!
I have read through a lot of threads lately on PH and I see that we keep using “agent governance” to mean permissions, approval gates and audit logs.
Sure these matter, but they govern individual actions. They do not govern the work.
A workflow can prescribe every step when the path is known. Complex work is different and thats something I discovered when Raju and I were building Midfleet.
The agent needs to investigate, make trade-offs, recover from dead ends and sometimes change its approach.
If we sequence every decision, we have built automation with an LLM inside it. If we simply tell the agent to achieve a goal, we have delegated without boundaries.
The middle ground is a form of mandate:
the intent behind the work
what is in and out of scope
rules that must remain true
who owns the current state
when the agent must escalate
the outcome and evidence required to finish
The agent remains needs to be free to reason inside that space. Governance does not tell it how to think. It makes the authority under which it acts explicit and durable across agents, humans and handoffs.
A workflow says what happens next.
Governance says what the agent is allowed to decide.
How are you encoding that boundary today without turning autonomous work back into a rigid sequence? Keen to hear thoughts on this :)