How Old School AI actually works.

Most self-description fails because people describe the communicator they aspire to be, not the one they actually are. The correlation between how people describe their communication style and how they actually communicate is about 0.29 out of 1.0. Psychology settled this decades ago, and the research behind our product says so plainly. So our interview never asks you to describe yourself. It asks for incidents, choices, and real writing: the evidence methods researchers use to draw out what people actually know but can't say.

We took the identity frameworks consultancies apply to companies, the elicitation instruments psychologists apply to experts, and the language science that reads who you are from how you actually write, and rebuilt them as one system for a single person, across five layers. Every question traces to a validated research instrument: the interview method the American military developed to learn why its best pilots succeeded, the forced-choice measurement tradition built to get honest answers where flattery was cheap, and the behavioral anchoring technique that ties every rating to a real observed moment. The names and dates are below. Every answer feeds a specific part of your encoded profile. Nothing decorative survives.

There's no trick in any of this. We applied rigorous research, rigorous methodology, and rigorous application. That's the whole recipe.

Our methodology is backed by decades of experience: tested, tried, and true. We didn't skim summaries of any of it. We acquired the primary sources, the original papers and the original editions, we read them, and we can tell you which page each design decision came from.

And the tradition is alive.

The living researchers whose published work we studied, named the way a bibliography names them.

The brand identity frameworks we rebuilt made the world's most valuable brands, and companies still pay consultancies to apply them today. They're the same brand-equity frameworks behind the annual most-valuable-brands rankings.

Our influence questions draw on the same peer-reviewed persuasion research the UK government built a policy unit on.

The differentiation framework in our positioning work is the one Nintendo's late president publicly credited for the Wii.

And the adoption thinking behind our migration work follows the playbook Silicon Valley companies from Salesforce to Google hired its author to run.

So the extraction is run the way the McKinseys, Bains, and Boston Consulting Groups of the world do it for companies: structured, documented, checked against the frameworks it came from. That class of rigor, pointed at one person.

We invite the scrutiny, because specificity is the defense: the moment anyone makes a specific research claim, it's checkable. Old school, in the most literal sense of our name.

Seventy years of elicitation science, pointed at one person: you.

25+

Built from primary sources, read in full: twenty-five plus books and papers across five research traditions, from elicitation science to positioning to direct response.

Your identity lives in your Old School AI profile, and Claude reaches it live through the Old School AI connector. Most tools in this category hand you a file and wish you luck. Ours is one skill upload plus the connector: you set it up once on your desktop, about ten minutes, and from then on it's present in every conversation, everywhere you use Claude, on every Claude plan including the free plan.

Update an answer tonight and Claude knows it tomorrow. No need to search for or install additional skills: the repository connects automatically through the connector and keeps growing, and every skill you have updates with it. No re-uploading. No reinstalls. No ZIP files to shepherd.

And it travels with you: installed connectors are active in the Claude mobile app too. Only adding a new custom connector requires desktop.

And everything you encode stays yours. Always downloadable, never locked in.