How AI Search Decides What to Recommend
An open framework for working that out, the articles that explain it, and the code from two tools I built and then switched off.
- Open Methodology
- 149 Factors, 8 Pillars
- Built by a Practitioner
The Three-Layer AI Visibility Gap
Not Found
AI models cannot recommend what they cannot see. If your content is not structured for AI crawlers, you are invisible to a channel that is still growing.
Not Cited
Even when AI finds your content, it may not reference you. Without clear authority signals and structured claims, you are data without attribution.
Not Chosen
Being found and cited is not enough. AI picks the clearest, most trustworthy answer. If that is not you, a competitor gets the recommendation.
The Amplifier Effect
AI lowers the build barrier, which means more competitors in every market. When anyone can build, discoverability decides who wins.
An Open Framework for AI Search
MASTERY-AI v3.2.0: 8 weighted pillars and 149 atomic factors, published in full. No black box. A methodology you can read, argue with and build on.
The Tools, and What Happened to Them
I built two tools on this framework and ran them as paid products. Both were switched off in August 2026. The code is open, so the work is still useful even though the services are not there to use.
AImpactScanner
Scored a page against the MASTERY-AI framework and returned the gaps in priority order. Retired in August 2026. The source is public under the MIT licence.
LLM.txt Mastery
Crawled a site and generated a spec-compliant llms.txt file, with a validator for checking one you already had. Retired in August 2026. The source is public under the MIT licence.
AImpactMonitor
A monitoring layer that was built and never opened to anyone. No public release, nothing to sign up for, and it is not coming back. It is here because the site used to advertise it.
Built by a Practitioner
I built AI Search Mastery because I needed it myself. As a programmer building AI-powered products, I found my own tools were invisible to AI search. The MASTERY-AI framework came out of solving that, in practice rather than in theory.
The tools that came after it did not find enough users to justify running them, so I shut them down and opened the code. That is the part most sites leave out, so it is on the front page of this one.
A practitioner sharing what works, not a guru selling certainty.