Insights · Field Notes
Notes from inside the machine.
What we learn building AI into B2B revenue engines, written down while it is still specific. Opinionated, practical, and useful without a call. No list to join: the feed is RSS, and the concierge answers questions about any of it.
Latest
- The State of AEO in B2B SaaS, 2026We ran 274 well-known B2B software companies through the same reader a buyer's AI assistant uses. 61% now publish an llms.txt, 76% carry structured data, only 4% block an AI crawler, and one in twelve could not be read at all. The numbers, the category gaps, what a model actually sees in the first 5,000 characters, and what to fix first.
- What a Buyer's AI Sees When It Reads Your HomepagePaste your homepage into an assistant with no context and ask what the company does. The answer is what buyers hear when they ask ChatGPT, Claude or Perplexity about you. Here is how to run that cold read in one click, what the common gaps look like, and how to close them.
- AEO for B2B: The Four Checks That Decide Whether AI Can Find YouBuyers ask ChatGPT, Claude, and Perplexity before they search. Four deterministic checks (llms.txt, AI crawler access, structured data, sitemap hygiene) decide whether those assistants can read your company, and what they say when they cannot.
- The Five Questions Deal Teams Skip in GTM Due DiligenceData rooms answer the questions they were built to answer. The five that decide whether a revenue engine is real (invoice-level retention, aged pipeline, founder gravity, paid-spend dependence, CRM-to-billing reconciliation) are the ones deal teams routinely skip. Here is why, and what to ask for instead.
Short, direct answers to common questions live in Answers. Follow along by RSS, or ask the concierge in the nav about anything you read here. The free tools are the practical half of these notes.