Evidence mounts.
See first.

We build the intelligence infrastructure that reads AI research, code, policy, and talent in concert, so your team sees the direction before the news does.

Explore HiddenState Request a Demo
Who we are

The intelligence layer between AI research and what comes next.

Mosaic Theory was founded on a simple observation: every major shift in AI is preceded by public evidence. Research breakthroughs, infrastructure acquisitions, talent migrations, policy shifts. The evidence is just scattered across thousands of sources, published in different formats, at different times, in different communities.


We build the systems that find, weight, and connect this evidence into a single, actionable intelligence layer. Venture investors, AI-lab strategy teams, consulting practices, policy desks, enterprise CTOs, and tech analysts use Mosaic Theory to read the direction of the AI frontier early.

24/7
Continuous monitoring
4x
Ingestion cycles per day
W-Index
Proprietary scoring
8.3 days
Average research-to-announcement lead
Flagship product

HiddenState

A daily intelligence engine that reads the ML and AI landscape across research, code, policy, and talent — and shows what's becoming real before the news does.

HiddenState

Evidence stacks.
Direction shows.

HiddenState reads independent sources across research, code, policy, and talent, groups findings by technical theme using proprietary ML, and delivers a ranked daily briefing scored by our W-Index — a composite of evidence weight, maturity, and engagement.

Request a Demo
Find

Catch patterns forming across research, code, and community

Connect

Weight the evidence when research, code, and hiring all point the same way

Confirm

Track every read through to announcement with documented foresight

Deliver

One briefing every morning. Ranked, contextualised, ready to act on

HiddenState daily briefing — 2026-03-20
Leadership

Built by researchers.
For decision-makers.

Our team combines deep machine learning research, operational scale, and strategic experience across technology, capital, and policy.

Cosmo Santoni
Cosmo
Founder & CEO

Cosmo is a PhD candidate in Foundational Machine Learning and Applied Mathematic at Imperial College London, where he specialises in state-of-the-art sequence models and real-time inference. His research and software engineering collaborations span Imperial, Cambridge, and DFKI in academia, along Google and the World Health Organisation in industry, with published work in ICML, NeurIPS, Nature and The Lancet. He built HiddenState from the ground up and provides the technical vision, ML architecture, and research depth behind the platform's signal detection and scoring engine.

GitHub Scholar LinkedIn
Nick Kaufmann
Nick
Senior Strategic Adviser

Having begun his career in the UK Treasury, Nick is Managing Director, Head of Equity Capital Markets and Syndicate, and Head of the London Office at WOOD & Company. He served as the independent equity advisor to TomTom on its €539m IPO and previously led Emerging Markets ECM at UniCredit/Creditanstalt and Citi/Schroders. He currently serves as a non-executive director of Pollen Technologies. Nick brings over three decades of institutional capital markets experience, buy-side relationships, and strategic positioning to Mosaic Theory.

LinkedIn
Get started

The signals are already there.
We give your team
foresight.

Explore HiddenState Request a Demo