Thomson Reuters made its first pre-revenue acquisition in 174 years when it bought a legal AI startup founded by a lawyer who taught himself planning law as a teenager to save his family home.
Safe Sign Technologies, created by Alexander Kardos-Nyheim while he trained as a solicitor at Allen & Overy, caught the attention of the global information giant before generating a single dollar from subscriptions. The deal, announced 21 August 2024, signals how specialist artificial intelligence research can command serious value ahead of commercial traction.
Kardos-Nyheim’s path started with a planning dispute in Rollins House, a dilapidated Art Deco factory in south-east London where his mother lived. A developer wanted to convert the site into flats. The family couldn’t afford lawyers, so while still in school, he studied planning law between exam sessions. He built a case from overlooked records and precedents, then presented it himself before a council panel. The family won against an experienced law firm.
Years later, he juggled solicitor training at Allen & Overy with building Safe Sign. His days began at 4:30 a.m. He worked on the startup before heading to the firm, then returned to it each evening. He later admitted the pace was unsustainable. The company also nearly ran out of money multiple times.
British investors repeatedly rejected him, so he flew to New York with £200 to his name. North American backers eventually provided nearly all the capital. His team recruited researchers from Cambridge, MIT, Harvard, and DeepMind to build AI models specifically for legal reasoning, not just another chatbot bolted onto existing software.
A consumer-facing legal product failed to attract interest, forcing a pivot toward the underlying technology. That bet on specialised models, trained for rigorous legal work, ultimately became the company’s core asset.
Thomson Reuters has said Safe Sign’s models showed industry-leading performance in domain-specific evaluations. The acquisition aims to accelerate development of its CoCounsel AI assistant. The price was not disclosed.
For legal professionals, the deal underscores a growing demand for AI systems that combine technical power with domain expertise and accuracy. For founders, it proves that deep research and defensible technology can outweigh polished pitch decks and early sales figures.
