TL;DR
A former senior Merseyside Police detective has raised money from Silicon Valley investors for a stealth start-up building AI tools for police forces, valued at close to £300m ($400m). The Financial Times reported the funding, and Silicon UK summarised it on 28 October. The company, 22one, is the latest police technology venture from Steven Richardson, who previously co-founded and sold the redaction tool Riven.
Who is behind it
Richardson left Merseyside Police in 2021 as a detective chief superintendent. He then led UK policing work at Amazon Web Services before co-founding Riven, which uses AI to redact documents and scan data and is used by UK forces. That company has since been sold.
According to the FT, as relayed by Silicon UK, 22one’s backers are Creandum, Lux Capital, Sequoia and General Catalyst. UK company filings show it was incorporated in December 2025. Beyond that, little is public: its website shows only the word “INVESTIGATE”, flashing. People familiar with the business told the FT it will build AI tools to help forces work more efficiently.
The policing context
The timing fits government policy. Ministers set up PoliceAI in June to develop, test and scale AI tools for forces.
It also fits the argument over Palantir. Silicon UK notes growing interest in home-grown alternatives to the US company, whose public sector contracts include work for the NHS. The Eastern Region unit for special operations, serving five East Midlands forces, ended a Palantir pilot. Sadiq Khan blocked a proposed £50m Metropolitan Police contract with Palantir in June, and Palantir later dropped its legal challenge to that decision.
Looking Forward
A valuation near £300m for a company with no public product is a bet on the founder and the market. In our view, the more useful signal is that top-tier US investors now see UK and wider policing as a market worth a stealth start-up, at a time when forces are under pressure from government to adopt AI and from politicians to avoid dependence on Palantir. 22one will need to show what it builds, and how it handles the bias and data-protection questions that follow police AI, before that valuation can be judged.