AI isn't gonna teach you this.

A free course, a method, and an ongoing practice for the patterns AI uses on everyone — including the people you'd think were safest.

The argument

Lawyers, finance professionals, people running real businesses in the City and on Wall Street are using AI every day and sometimes getting caught. Not because they're careless. Because the patterns work on everyone unless someone has been trained to notice them.

AI Clarity is the training. It started as field notes — what happens when you sit with these tools long enough to see what they're actually doing, instead of what they appear to be doing. It became a methodology with seven named patterns on the machine side, ten failures on the human side, and the interventions that break each one.

Built by EverythingThreads as an independent research company. Not a vendor, not an AI lab, not a consultancy with something to upsell on the back end.

AI isn't gonna teach you this. That's what we built.

The three tiers, briefly

AI Clarity Foundation

Free

Two hours · 6 segments · 11 languages · No card required

The three patterns most people will recognise the moment they're named, plus the four questions that interrupt them. A certificate when you finish.

Start free

AI Clarity Practitioner

£19.99 /month

Or £179/year · cancel any time · pause up to 60 days/year

The working library — 1,120 slides across developer and architecture practice. New slides added every month. The canary set running on your behalf. A member community.

See Practitioner

What's underneath the method

The taxonomy used in the courses — seven machine patterns (M1–M7), ten human failure patterns (F1–F10) — was developed through working with these tools intensively over a period of weeks, then formalised against the existing academic literature on RLHF, sycophancy, and model evaluation. It is preregistered with the Open Science Framework and is the basis for ongoing independent research.

See the methodology →

The books

The work behind AI Clarity is documented in two books.

The Agreement Trap — a 74,000-word account of what happens when you take generative AI seriously enough to notice what it's actually doing. Literary nonfiction; reads like a memoir; functions like a field report.

AI Clarity — the method book. The taxonomy, the interventions, the case studies, the business case for treating AI literacy as a real skill rather than a vague aspiration.

See both books →

For teams

If you're responsible for AI literacy at a company — whether for compliance under Article 4 of the EU AI Act, or because your team is using AI tools daily without a shared framework — AI Clarity Professional ships with a manager dashboard, per-seat licensing, and Article 50-compliant export.

See AI Clarity for teams →