.png)
Harvey
AI Agents
← Websites
CTGT
Reliable frontier AI for regulated domains, built on interpretability research
Visit site ↗
5.0MB · 7,111px tall · uses Switzer Variable, webflow-icons, Helvetica Neue
Industry
AI Infrastructure, Security
Style
Monochrome, Minimal, Typographic
Stage
Seed
Built with
Webflow
About the site
CTGT is an applied interpretability lab building enterprise AI for regulated industries like finance, with a policy engine that constrains any model to organisational policy at output. The site uses black-and-white landscape photography as its visual language: a misty mountain ridge fills the hero behind a black-and-grey headline, with clean benchmark charts and a dark shoreline carrying the rest. Restrained Swiss-style type keeps it serious. A strong reference for AI research companies selling to cautious enterprise buyers.
How this page sells
For risk and compliance leaders at regulated firms: press, research, big-name endorsements and exact benchmarks borrow a lab's credibility.
Primary navigation
- About us
- Research
- Finance
- Request access
ClickTap a section to see it on the page
'Reliable frontier intelligence for regulated domains' with 'frontier' and 'intelligence' in grey, over a misty black-and-white mountain ridge, with 'View the research' as the main button.
Why it converts: 'Regulated domains' filters the audience in the headline itself, and leading with research instead of a demo signals rigour, the trait a compliance buyer screens for.
Four dated press cards: Forbes (CES Top 12), WSJ, Forrester ('explainable AI shines bright') and Business Insider ('raised $7 million').
Why it converts: For a seed-stage company selling to banks, third-party coverage borrows the trust a long customer list would otherwise provide.
One large statement: 'CTGT's research is built upon a deeper understanding of how AI models internally represent and process information', with the company name in grey.
Why it converts: Interpretability is the mechanism behind the reliability claim. Naming it separates CTGT from wrappers that simply prompt a model harder.
'From the creator of Keras to the leadership of the Fortune 500': a carousel led by CTGT at JP Morgan's IMF summit, plus quotes from a Brex CTO and a U.S. Cyber Command deputy general counsel.
Why it converts: Wall Street, a CTO and national security in one row cover the three people who can block a regulated AI deal: the business, engineering and legal.
Bar charts pitting CTGT against Gemini 3 Pro, Claude 4.5 and RAG. In one, RAG dropped legal-reasoning accuracy to 39% and CTGT's Policy Engine doubled it to 78%.
Why it converts: Head-to-head numbers against models the buyer already knows turn 'reliable' from an adjective into something they can check.
'Total SEC, FINRA, and organization-wide compliance. Instantly applied. No manual setup required.' over a mountain photo, with 30% better policy adherence, 20-40% lower TCO, 97% faster policy updates.
Why it converts: Naming the regulators speaks the finance buyer's language, and 'no manual setup' removes the implementation fear that stalls most compliance tooling.
'Interested in partnering with CTGT?' and one small Contact us button over a dark, reflective shoreline.
Why it converts: 'Partnering' rather than 'buying' suits a lab selling to institutions, and the quiet close keeps the premium, research-first tone to the end.
