Exhibit F · Market mapping
681 UK business-model subsectors, each scored individually on nine anchored criteria across two independent axes, with live company counts from the register.
Click any bubble, row or quadrant to see the nine criteria behind it, with the anchor sentence for the level scored. Filter by section, by roll-up fit, or by minimum pool size. Bubble area is the count of active UK companies aged five years or more.
"Which industries will AI eat" is usually answered at the level of a category. That is the wrong altitude, because AI acts on tasks, and every category is a bundle of tasks with wildly different exposure.
An earlier version scored two axes per archetype and let subsectors inherit their archetype's position. It had no company counts and a roll-up fit column I could not define precisely. Someone took it apart in about four questions, and every one was fair. This version scores each subsector individually, on nine criteria with written anchors for what each 0 to 4 level means, and puts every weight, anchor and threshold in one assumptions.json so a challenge can be re-run rather than argued about.
The cut lines are the universe median, 1.625 on displacement and 2.5 on tailwind. Not a line fitted to make the answer look tidy.
To test whether the scoring was reproducible, 50 subsectors were deliberately scored twice by two different models.
Exact agreement was 33%. Mean difference 1.04 levels on a 0 to 4 scale, and 1.39 apart on tailwind. The disagreement was systematic, not noise: the smaller model scored electricity transmission tailwind at 1.12 where the defensible answer is around 3.75.
When a quadrant boundary is a median, mixing judges silently sorts one judge's rows into different quadrants than the other's. The map looks finished and is quietly incoherent. All 681 were re-scored by a single judge, and both passes were kept so the comparison stays inspectable. If you use an LLM as a rater, measure inter-rater agreement before trusting a single pass.
Counts come live from the Companies House advanced-search API per five-digit SIC code, across all 731 codes. But 251 of the 701 mapped codes are claimed by more than one subsector, so those counts are upper bounds. Every row carries a flag saying so and a lower-bound column counting only exclusive codes.
Deliberately missing: no financials by subsector, because the UK profit-and-loss exemption means small companies do not file one; no ownership filter; and these are register counts, not qualified targets.