Field notes on how AI decides who to name.
Guides and notes on the mechanics of AI-mediated discovery, the market structure underneath it, and how to read your own category.
Grouped by the question they answer, not by date. Start anywhere — each note links to the next, to the underlying framework, and to the Index data it draws on.
How AI-mediated discovery works
The mechanics underneath the chat box — the retrieval layer, query fan-out, the entity record, and the agent buyer. See the framework →
The retrieval layer nobody optimizes for
The search infrastructure beneath the chatbots — Exa, Brave, Tavily, and the retired Bing API — and why it now decides who AI can find. Meet your Retrieval Surface.
You ask one question. AI asks twelve more.
Query fan-out, with the data: Google AI Mode fans out to 5–11 sub-queries, ChatGPT to ~2–3 — and the queries are getting sharper. What it means for getting found.
How AI systems see your company
A machine that never visited your site is describing you to buyers — confidently, often wrongly. What it sees, why every engine sees you differently, and how to find out.
When the buyer is an AI agent
Agentic commerce, grounded: the protocols (ACP, UCP, MCP), the adoption data, and why Gartner expects agents to intermediate $15T in B2B purchases by 2028 — and what it takes to be buyable by a machine.
Who gets named, and why
The market structure of the answer layer — who holds the territory around a buyer's question, and what moves it. See it in the Index data →
When AI engines disagree: frozen, fragmented, and molten markets
Category temperature names the shape of AI agreement. Three top-level types, a two-axis frame inside fragmented (polar vs diffuse), measured across five industries. A working frame for reading the structure of engine disagreement.
Citation territory: mapping the AI answer layer
Someone already owns the answer to your buyer's question. AI-mediated discovery has a market structure — and the right unit of analysis isn't your visibility score but the territory around each question: who holds it, where the open field is, and how the ground shifts.
Why AI recommends some companies — and ignores others
AI doesn't rank you; it asks around. Why the engines disagree about who exists, and why brand mentions beat backlinks roughly 3× in getting you named.
Why AI recommends your competitor's software
AI builds software shortlists from G2, Capterra, docs, and Reddit — and reads your own comparison pages for the competitor. Why, and what actually moves it.
Check your own category
Practical reads for a specific situation — a self-check, a vertical, and the "who do we hire" question. When you want it measured properly, that's the Diagnostic.
Will ChatGPT recommend your firm?
How to check in 20 minutes whether AI names your firm — the manual audit, why the engines pick who they pick, and why professional-services firms are uniquely exposed.
How AI picks financial advisors — and why the bar is higher
Finance is a "Your Money or Your Life" category: AI holds it to the highest trust bar, prefers magazines over advisors, and the usual playbook collides with SEC/FINRA rules.
We asked Grok who to hire for AI search — and weren't on the list
A dogfooding note: we ran our own diagnostic on ourselves, for the exact question our buyers ask. Grok recommended a list of agencies and left us off — then explained why, in our own words. The diagnosis, on our framework, and what we changed.
Comparisons
Where a diagnostic fits next to the tools and agencies in the category — an analyst's read, not a sales pitch.
AI-visibility tools → AI-search agencies →Forthcoming
The recurring cross-engine study lives in the Answerability Index — the same prompt sets, re-run and published by sector. Vertical notes for specialty SaaS and financial services are in preparation. Pieces are published when the underlying work is done, not on a schedule.