Anthropic Launches Claude Science and Plans to Develop Its Own Drugs

At an event called "The Briefing: AI for Science" this week, Anthropic unveiled Claude Science, a new AI workbench aimed at researchers. The product pulls together fragmented tools and datasets into a single environment and can generate figures and visuals. Anthropic also announced it intends to develop its own drugs, not just sell AI tools to companies that do. The company already has a list of biotech and pharma customers using Claude, and framed the launch around AI's potential to accelerate scientific discovery and healthcare development.

What happened

Anthropic held an event this week called “The Briefing: AI for Science” and used it to announce Claude Science. The company describes it as an AI workbench built for scientists, one that brings together tools and datasets that are normally scattered across different platforms. It also generates figures and visuals, which is a step beyond what a general-purpose AI assistant typically does.

Anthropic positioned the launch around a specific claim: that AI can “dramatically accelerate the pace of scientific discovery and the development of healthcare interventions,” according to the company’s own framing at the event. A list of biotech and pharma customers already using Claude was cited to back that up, though specific names from the source excerpt were not detailed.

The bigger headline, though, is that Anthropic said it plans to develop drugs itself. That moves the company from selling AI to researchers into becoming a direct participant in drug development.

Why it matters

There is a real difference between building tools for scientists and deciding to run your own drug development program. The first is a software business. The second puts Anthropic in competition, or at least in the same space, as the pharma and biotech companies it is also trying to sell to.

Claude Science as a product makes sense. Researchers genuinely deal with fragmented software environments, and an AI layer that connects those pieces and outputs publication-ready visuals has obvious value. Anthropic already has traction here with existing biotech and pharma customers, so the workbench is building on real usage rather than speculating about a market.

The drug development ambition is a different kind of bet. It signals that Anthropic sees its AI as capable enough to pursue outcomes, not just assist humans pursuing them. Whether that turns into an actual drug pipeline or remains a long-range aspiration is still unclear based on what was announced.

Our take

The Claude Science workbench is the part of this announcement worth watching closely in the short term. A consolidated research environment with AI-generated figures is a practical, sellable product. If it works well enough to cut down the time scientists spend wrangling data and tools, adoption in biotech will come quickly, and that is a meaningful business.

The drug development announcement is harder to evaluate. Anthropic is a well-funded AI lab, but building a drug pipeline requires a very different set of capabilities: clinical expertise, regulatory navigation, and long time horizons that go well beyond model training cycles. It is possible this means they plan to use Claude Science internally to run experiments and license or partner on any resulting compounds. Or it could mean something more ambitious. The source does not spell it out, and that ambiguity is worth noting.

For businesses already using Claude for research or document work, this is a signal that Anthropic is investing seriously in the science vertical. That usually means better models, better tooling, and more specialized features are coming, even if the drug angle never materialises into a product you interact with directly.

What to do about it

If you work in biotech, pharma, or any research-adjacent field, put Claude Science on your evaluation list now. The key questions to answer before committing: Does it connect to the specific data sources your team already uses? How does it handle proprietary datasets? And what does the pricing look like compared to your current research stack? Getting those answers early puts you ahead of a likely wave of interest once case studies from early customers start appearing.

Source: The Verge · AI