When a brief arrives, the first thing we do is read it twice, once for what it asks, once for what it assumes. Most briefs ask for a tool (“we need a chatbot”) and assume a diagnosis (“because our response times are slow”). The assumption is where the engagement actually lives, so that is what our reply engages with.
Within 48 hours the sender gets a written response with three parts: what we understood the operational problem to be, in our words; an architecture sketch of how we would approach it; and what the sketch would need to be true in order to work, the assumptions we would test first.
The sketch is supposed to be wrong
We know the sketch is wrong when we send it. We have spent two days with your brief and zero days with your operation. But a concrete, wrong sketch is more useful than a cautious, vague one: it gives you something specific to correct, and your corrections teach us more about the operation in one reply than a month of discovery calls.
“A specific wrong answer invites correction. A vague right answer invites a meeting.
We publish this process because it filters both ways. Clients who enjoy correcting a sketch tend to be the ones who stay engaged through delivery, and the ones we do our best work for.