Discovery
Follow one real piece of work. Find the hidden handoffs, workarounds, judgment, repeated questions, and disappearing details.
How it works
Discovery first. Software only when it earns its place.
We do not begin with a feature list. We begin by following the work, listening to the people, and preserving what is actually happening.
Blueprint
Define sources, roles, boundaries, decisions, proof requirements, outputs, and the smallest useful loop.
Pilot
Build the bounded loop. Test it with real work. Keep the promise narrow enough to verify.
Care
Repair what fails, preserve what works, and add capability only after the operation has earned it.
The governing rule
AI proposes. Humans confirm.
The Hovelkin may prepare a record, recommendation, draft, route, or next action. Human Save controls what becomes official or consequential.
You should leave Discovery knowing:
What the real problem is. Where responsibility breaks. Which source must be preserved. What a useful system should do. What it must never do. Whether building it is justified.
You do not need to arrive knowing what software you need.
Bring the mess. We will help determine whether the right answer is a Hovelkin, a simpler process change, or no software at all.
Start Discovery →