map-the-terrain

Map the Terrain

In hard mode the binding constraint is the approval clock, not the tech. AI collapsed the cost of building and did nothing to the cost of waiting, so the part you can now compress was never the part that decided whether you lived. Map the clock and the wedge before you write a line of code, or you will ship a beautiful product years late into a market someone else already framed.

The method

Four moves, in order. The first reorders everything after it. Full framework, the wedge rubric, GEO versus SEO, the EU timeline cautions, and a worked terrain map: references/terrain-method.md.

Step 1: Map the approval clock

Find the longest queue between you and a paying customer, and measure it in months: EFSA opinion, Novel Foods dossier, clinical validation, CE mark. Approval time, not build time, sets your real timeline, because it sits in a process you do not own. Put that number at the top of the page and sequence entry around it. Start the dossier first; build in the shadow of the queue. If you cannot name the number, you are not ready to build. You are ready to research.

Step 2: Find the wedge

Name the one gap worth attacking. Where do incumbents solve half the problem and leave the rest to the customer? Where is the manual cost so high that a slow, expensive human process is begging to be undercut? Write it in one sentence that names what the incumbent fails to do. If it comes out as "we will be better", that is a slogan, not a wedge. Keep digging until it names a hole.

Step 3: Point AI at the terrain

The regulatory record is enormous and largely text. Use a long-context model to digest dossiers, prior authorisations, opinions, and competitor filings into one regulation-aware map: who is in which queue, what data requirements sank earlier applicants, where the gap between filings is wide enough to walk through. The machine reads. You decide which wedge, which risk, which sequence. Confuse the two and you outsource the one job that is yours.