TORONTO — Cohere has acquired Reliant AI, a Montreal-based firm developing technology for the pharmaceutical industry, as it looks to grow in more specialized markets.
The deal gives Cohere several global biotech customers, as well as a team of researchers in Canada and Germany. It follows the firm’s recent announcement that it plans to merge with German AI firm Aleph Alpha as part of its sovereign AI push.
Talking Points
- Cohere is acquiring Montreal-based Reliant AI to help power a version of its North platform for the pharmaceutical industry
- Reliant made tools to help speed up literature review and other tasks in the drug discovery process. Cohere is buying its capabilities in technical areas, like reinforcement learning, and a team of researchers in Canada and Germany as it tries to sell to governments and regulated industries.
Cohere has successfully rolled out AI tools for common functions found at most companies, like human resources, finances and communications, said chief AI officer Joelle Pineau. She added that there was now “a ton of opportunity in the verticalization of that product,” and that Reliant was an attractive acquisition because its team has experience building both AI models and products.
Google DeepMind researchers Karl Moritz Hermann and Marc Bellemare founded Reliant in June 2023. Pharmaceutical firms used the startup’s AI research tools to conduct reviews of published medical studies, analyze past clinical trials and manage their portfolio of drug candidates.
Reliant developed specialized algorithms and data to try to ensure AI models give users the right answers—crucial in an expensive and highly-regulated domain like biotech. The firm claims its tools get drugs to market faster. Clients included British pharmaceutical giant GSK, French firm Ipsen and Japan’s Kyowa Kirin.
Cohere plans to integrate Reliant’s technology into a version of North, its platform for building AI agents, designed for the pharmaceutical industry. The firm has previously tailored North for the financial services and telecom sectors. It also recently sold the system to the federal innovation department.
Steve Woods, a partner at Inovia Capital who has led financings for both firms, said there was “real alignment” between the companies. Cohere is “really good at model-building,” he said. Reliant, meanwhile, has strong capabilities in technical areas like reinforcement learning and optimizing the way AI agents do tasks, he said.
Both have tried to help clients adopt generative AI faster by selling tools that mimic how their employees work, rather than just offering models, Woods said. “Otherwise it’s hard for anybody to use your product.”
The two firms did not disclose terms of the deal. Reliant had previously raised US$13.5 million in venture capital, according to PitchBook data, including a US$11.3 million seed round in mid-2024 that Inovia co-led. Woods claimed Reliant had been doing well on its own, and that the deal would generate a good return for Inovia in a short period of time.
Cohere, meanwhile, has taken in US$1.6 billion to date, and is currently raising a new round to which Aleph Alpha backer Schwarz Group has already committed US$600 million.
Cohere has long focused on the specific AI needs of big businesses in regulated industries. It now faces increased competition from AI giants OpenAI and Anthropic who are increasingly setting their sights on the enterprise market, launching new tools for sectors like law and finance, and buying up startups already targeting specific verticals.
Cohere typically works with an anchor customer—RBC for financial services, STC Group for telecom—to adapt North for a particular sector. The AI firm must learn which tasks firms in that industry commonly set for agents, what prompts their workers use to control them, what data they rely on, and how to connect to their existing databases and software systems.
“The playbook is the same from one vertical to the other, but the content is different,” said Pineau. She added that Cohere has identified as many as 20 industries to which it could sell a sector-specific version of North.
Hermann, based in Berlin, will become vice-president of AI verticalizations at Cohere, leading work to tailor the firm’s models for target industries. That’s a continuation of Reliant’s strategy, which aimed to expand from pharma to other fields with lots of data to parse and a high need for accuracy. “Any industry where there’s a large amount of knowledge work, there will be mental menial labour,” Bellemare said in an interview with The Logic in 2024.
Cohere already lets clients run its models on their own hardware, which it touts as a way to address concerns about control over the technology and access to sensitive data. The expertise of Reliant researchers in upgrading already-trained models with specialized data will add to North’s appeal, Woods claimed.
The deal also brings together two Canadian companies, amid renewed concerns that the country is losing top AI researchers and breakthroughs to Silicon Valley. Reliant’s 30-plus team will join Cohere’s workforce of over 550, including at a recently opened office in Montreal that Pineau heads.
Amid a fierce war for AI talent, many tech firms are using acqui-hire deals to bring in top researchers and engineers. That’s not Cohere’s strategy, according to Pineau. The firm is only looking for teams that are a good fit in terms of where they are and what they’re working on, she said. “We’re not really interested in just sweeping up anything that’s on the market.”