Two-Phase Authored Discovery

Map what’s there. Then ask whether it should be.

01The foundation

Your org chart is scar tissue.

The meeting nobody remembers creating. The approval step that exists because someone got burned once and built a fence. The role that exists because two systems don’t talk and somebody has to carry the data between them in their head. None of this was architected. It accumulated. Each layer made sense at the time — a response to a constraint. And the constraint was almost always the same thing: the right information wasn’t where it needed to be when it needed to be there.

When those constraints dissolve, the scar tissue stays. You have to go looking for it. It will not volunteer.

Most AI deployment accelerates the scar tissue. AESOP looks for it. And asks whether it still belongs.

02Two approaches to discovery

Phase 1 — Inherited where most platforms stop	Phase 2 — Authored what makes AESOP different
Maps the current process, step by step Captures the roles, approvals, and meetings as they are Asks: “How does this work today?” Produces a brief shaped by the existing org chart	Pulls at the seams. Which roles exist because of old constraints? Maps decisions. Not departments. Separates real constraints from inherited ones Asks: “What would you build if you were starting today?” Enriches the brief. The assumptions get challenged.

03How it works

Phase 1Map what’s thereA form interview or brain dump. Captures process, pain points, roles, constraints. The current state, honest and unvarnished.

Phase 2Pull the threadsSeven sections of questioning. Each references what Phase 1 captured and asks: should this exist? Or is it just scar tissue?

OutputOne brief, two voicesThe inherited map. The authored reframing. Side by side in one document. Design and Build consume it unchanged.

04Why this matters

The replication trap: everyone posts their agent count. “Fourteen agents. I get updates before I wake up.” That isn’t transformation. It’s a status meeting with better latency — a faster horse with a dashboard. Nobody posts about the approval layer they deleted. Nobody measures the roles they dissolved. Agent count is the vanity metric of the AI era. The question that matters: what changes when judgment and context stop being gated by headcount?

01Structure as output, not inputInherited takes the org chart as given. Authored treats it as the thing under investigation. One designs around the structure. The other designs the structure.

02Replication vs. dissolutionTwo projects hide under the word “transformation.” If removing the agent sends you back to the old way, you automated. If the agent makes you realize the role had no reason to exist, you transformed. Same technology. Opposite outcomes.

03Legibility biasAutomating a standup produces a demo you can screenshot. Asking why the standup exists produces a redesign you can’t. The first is fundable. The second is necessary. Guess which one everyone picks.

04The only questionDid this agent replace a person? Or did it dissolve the reason the role existed at all? If you don’t know the difference, you’re not transforming — you’re pouring concrete over the old foundation and calling it new construction.

“Does this agent replace a person, or does it dissolve the reason the role existed?”

Answer the first and you’re automating. Answer the second and you’re authoring. AESOP forces the question before anything gets built.