Meryem Arik, Dr Jamie Dborin, and Dr Fergus Finn founded Doubleword (originally TitanML) in London in 2021 to solve one of the biggest barriers to enterprise AI adoption: self-hosted inference. During their postdoctoral research at UCL, Jamie and Fergus discovered that techniques from their quantum machine learning work could dramatically improve AI inference performance. Together with Meryem, who studied theoretical physics and philosophy at Oxford, they set out to make self-hosting AI models as straightforward as using third-party APIs.

Inference is where AI delivers real-world value, transforming trained models into business outcomes. As enterprises increasingly need to own and control this capability for reasons of cost, security, and performance, the challenge of building and maintaining the underlying infrastructure has become a major bottleneck. Doubleword's platform enables one-click deployment of any open-source, proprietary, or fine-tuned model across cloud and on-premise environments, with built-in monitoring, logging, and GPU resource management. The company has already secured partnerships with Snowflake, Dataiku, and Nvidia, and counts major enterprises in pharma, manufacturing, and finance among its customers.

We led Doubleword's $12 million Series A in May 2025. The company is one of the earliest movers in the inference space, having been founded before ChatGPT brought large language models into the mainstream. Doubleword is now expanding its team globally and deepening its platform capabilities as enterprise demand for self-hosted AI accelerates.