Cohere, a Toronto-based enterprise AI startup, acquired Montreal-based Reliant AI on May 19, 2026, to build specialized AI tools for pharmaceutical research and drug discovery. The deal marks a significant shift in how AI companies are approaching regulated industries, moving beyond general-purpose chatbots toward domain-specific systems that can handle sensitive biomedical data securely.

What Is Cohere Doing in Pharma?

Cohere is integrating Reliant's technology and expertise into its flagship North platform, creating a new product called "North for Pharma." This isn't just a rebranded version of existing software. North for Pharma is described as "an agentic AI system purpose-built to enhance productivity and efficiency" across drug research and development, clinical trials, and scientific analysis. The system can run entirely on-premises or in private cloud environments, meaning pharmaceutical companies never have to send their proprietary research data to external servers.

Reliant AI, founded in 2023 by former Google and DeepMind researchers, had already built exactly what Cohere needed. Reliant's core product, called Reliant Tabular, was a research workbench designed to automate tedious tasks in drug discovery. According to Reliant's own claims, the system could scan scientific literature nearly five times faster than manual methods while making far fewer errors. For an industry drowning in research papers and competing datasets, that speed advantage matters enormously.

Why Does This Deal Matter for Enterprise AI?

The acquisition reflects a broader trend in enterprise AI: companies are moving away from one-size-fits-all models toward specialized systems built for specific industries. Cohere's strategy centers on what it calls "sovereign AI," meaning AI systems that can be deployed securely without exposing sensitive data to third parties. For pharmaceutical companies handling proprietary drug formulas and clinical trial data, that promise of data sovereignty is a major selling point.

The deal also brings real customers into Cohere's fold. Reliant AI was already working with major pharmaceutical firms including GSK (GlaxoSmithKline) and Kyowa Kirin, a Japanese pharmaceutical company. These aren't small pilot projects; they're established relationships with some of the world's largest drug makers. Cohere now inherits those contracts and the credibility that comes with them.

How Does This Fit Into Cohere's Bigger Picture?

The Reliant acquisition isn't Cohere's first move into specialized enterprise AI. In April 2026, just weeks before the Reliant deal, Cohere announced a tie-up with Aleph Alpha, a German AI firm, to deepen its "sovereign AI" offerings in Europe. These moves suggest Cohere is deliberately building a portfolio of industry-specific AI tools rather than competing head-to-head with OpenAI or Anthropic on general-purpose models.

This strategy makes sense given the competitive landscape. While OpenAI's ChatGPT and similar models dominate consumer attention, the real money in enterprise AI lies in solving specific problems for regulated industries. Banks, healthcare systems, and pharmaceutical companies have compliance requirements, data privacy concerns, and domain-specific workflows that general-purpose AI struggles to handle. Cohere is positioning itself as the vendor that understands those constraints.

What's Happening With Cohere's Open-Source Models?

Around the same time as the Reliant acquisition, Cohere made a significant move in the open-source AI space. On May 23, 2026, Cohere released Command A+, a 218-billion-parameter mixture-of-experts model, under an Apache 2.0 license. This licensing change is more important than it might sound. Cohere's previous open-source models used a CC-BY-NC 4.0 license, which prohibited commercial use. That restriction meant enterprises couldn't legally build revenue-generating products on top of Cohere's open models without paying for proprietary API access.

Apache 2.0 removes that legal friction entirely. It grants commercial deployment rights without royalty obligations, meaning companies can now use Command A+ in production systems without vendor lock-in. The model itself is substantial: 218 billion total parameters with approximately 25 billion active per token, and it can process a context window of 128,000 tokens, roughly equivalent to 100,000 words at once. Cohere claims the model runs approximately twice as fast as prior Command versions, though independent benchmarks don't yet exist to verify that claim.

Steps to Evaluate Cohere's Enterprise AI for Your Organization

• Assess Data Sovereignty Needs: Determine whether your organization requires on-premises or private cloud deployment due to regulatory or competitive concerns. If data residency is critical, Cohere's North platform's ability to run in fully isolated environments is a key advantage over cloud-only competitors.
• Evaluate Domain Fit: Consider whether your industry has specialized AI requirements similar to pharmaceuticals, banking, or healthcare. Cohere's strategy focuses on regulated sectors where general-purpose models often fall short. If your use case is highly specialized, domain-specific systems like North for Pharma may outperform generic alternatives.
• Test Open-Source vs. Proprietary: If you're evaluating Command A+ or other open-weight models, verify the Apache 2.0 license claim on the Hugging Face repository before committing to integration. Wait for independent benchmarks from third-party evaluators before migrating production workloads, as vendor performance claims lack external verification.
• Review Customer References: Ask Cohere for references from existing North deployments, particularly in your industry. The fact that Cohere now has GSK and Kyowa Kirin as customers through the Reliant acquisition provides concrete proof points for pharmaceutical companies considering adoption.

What Does This Mean for the Broader AI Market?

The Cohere-Reliant deal signals that the enterprise AI market is fragmenting. Rather than a single dominant model serving all industries, we're seeing specialized platforms emerge for healthcare, finance, legal, and other regulated sectors. This fragmentation creates opportunities for startups like Cohere that can build deep expertise in specific domains, but it also means enterprises will need to evaluate multiple vendors rather than relying on a single AI provider.

The pharmaceutical industry is particularly attractive because drug discovery is expensive, time-consuming, and data-intensive. AI that can accelerate literature reviews, identify drug targets, or predict clinical trial outcomes could save billions in development costs. That's why Amazon, Nvidia, Google, Microsoft, and numerous biotech startups are all racing to build AI workbenches for drug R&D. Cohere's acquisition of Reliant positions it as a credible player in that race, with both the technology and the customer relationships to compete.

Cohere's dual strategy of building proprietary domain-specific products like North for Pharma while simultaneously releasing open-source models under permissive licenses like Apache 2.0 also reflects a maturing AI market. Companies can now choose between paying for specialized, managed solutions or deploying open models themselves. That choice benefits enterprises by preventing vendor lock-in and encouraging competition on quality rather than just market dominance.