Human Agent Interface / Setup Inflow

Where does your agent workflow break?

A short setup intake for one real workflow. One block at a time. Mostly choices, a few short answers, no account, no backend.

Back to products
Step 1 of 6 / Current Work

01 / Current Work

What are you working on?

Start with the actual task or workflow. One field is enough for this first screen.

short context

02 / Tools And Agent Stack

Which agents are already in the loop?

Top choices first. Open the rest only if needed.

Which tools or agents do you currently use?
required
optional

03 / Failure Pattern

What currently breaks?

Pick the top three. This should stay diagnostic, not exhausting.

Failure pattern

Pick max 3 failure patterns.

optional

04 / Human vs Agent Ownership

Who owns which part of the work?

Start with the obvious work types. Reveal advanced ones only if they matter.

Which parts do you currently keep yourself?
Which parts do agents currently do?
What is missing between you and the agents?
optional

05 / Autonomy Boundary

How much control should move to agents?

More autonomy is only useful when the boundary is explicit.

required
1 sentence
What would you like to stop approving manually if safer?

06 / Verification Maturity

How do you know the result is real?

If there is no verifier rule yet, that is useful signal.

How do you currently check agent work?
required
What evidence would make you trust the result?
optional

Output / Copyable

Owner Intake Packet

Nothing is submitted. This is the copyable context for Samuel or the first call.

Generate the packet after finishing the intake.
One block visible