Poindexter Labs has raised GBP £2 million in a seed round led by Episode 1, with backing from Evertrue Capital and Octopus Ventures’ First Cheque Fund.
The London-based AI data company said the oversubscribed financing also included angel investors and some of its own contributors, including mathematicians and scientists who use the platform and invested personal funds in the business.
The fund-raising comes as Poindexter signs its first direct contract with a major frontier AI lab, moving from intermediary work to operating as a named supplier. The company reached USD $1.6 million in revenue in its first six months before taking external investment.
Founded by Jocelyn D’Arcy, Poindexter produces training data for advanced AI models that require expert reasoning rather than simple labelling tasks. Many existing data platforms were built for image tagging and text categorisation, not step-by-step work in fields such as mathematics, science, law and engineering, it says.
That distinction has become more important as AI developers seek datasets that can teach models to reason through complex problems. Poindexter argues that much of the training data used across the sector is either inaccurate or discarded because review systems reward rejection rather than improvement.
The company has built its model around a network of Olympiad medallists, PhDs and professors from institutions including Oxbridge, the Ivy League and MIT. Contributors are admitted through live subject-knowledge interviews and work within a collaborative peer-review process that Poindexter says has been refined across millions of tasks.
That process delivers more than 95% of contracted data and achieves a 99.5% acceptance rate on completed work, according to the company. It contrasts that with an industry average of 5% to 60%.
Poindexter has also developed a software platform, now in beta, that supports its own data-production service and is intended for licensing to government departments and companies building AI systems internally. The platform is designed to let internal specialists work with the same collaborative tools used by its external contributor network.
Data bottleneck
Poindexter is entering a market shaped by rising demand for specialised datasets as developers push beyond general-purpose language models towards systems expected to show stronger reasoning. Next-generation models from groups such as OpenAI, DeepMind and Anthropic rely on worked solutions and expert judgement, increasing pressure on suppliers that can source and review material with a high level of technical accuracy.
Its argument is that this part of the AI supply chain has remained underdeveloped. Rather than treating data work as a mass task completed in isolation, Poindexter says it has borrowed methods from academic research, with contributors reviewing each other’s work and improving submissions rather than discarding them.
D’Arcy founded the company after working in education and academic administration, including roles as a maths teacher and Chief of Staff at the National Mathematics and Science College. Her academic background includes degrees from MIT, Cambridge and Oxford.
In comments accompanying the announcement, D’Arcy set out the company’s view of the market problem.
“The workflow that has defined the industry since its inception was designed for a factory, not for knowledge creation. As a result, a huge chunk of training data is discarded not because it is wrong, but because adversarial review processes actively incentivise discarding tasks rather than improving them. This is a workflow problem, and we built Poindexter the way academics build knowledge: collaboratively, transparently, with peer review at every step. We have an accepted ACL paper, a growing research pipeline, and a commitment to building the kind of high-quality sovereign datasets the UK’s AI future depends on. Our proof point is our revenue. If we weren’t 5x more efficient than the platforms we replaced, we simply wouldn’t exist,” said Jocelyn D’Arcy, Founder and CEO, Poindexter Labs.
Episode 1 said the company had already shown both technical and commercial progress before the seed round. Its backing adds to a small but growing group of UK funds supporting infrastructure businesses linked to AI development rather than model makers themselves.
“Poindexter has done something rare – built a business that is both technically differentiated and commercially validated from day one. The traction speaks for itself. We backed them because the problem is real, the solution is working, and Jocelyn is unlike any founder we have ever met,” said Adam Shuaib, General Partner, Episode 1.
