Anthropic is rolling out Claude Science, an AI workbench for pharma researchers. The buzzy AI startup says this new product goes beyond assisting scientists to actually doing their work.
“Claude can run the work — not help with it, not accelerate it — even run it,” said Zubair Jandali, who leads Anthropic’s healthcare and life sciences commercial team, during a launch event the startup held Tuesday in San Francisco.
Claude Science is designed to pull together the databases, lab tools and computing power researchers normally juggle across dozens of programs. Researchers can give the platform a single instruction, like asking it to screen for a promising drug compound, and it will carry out the analysis, run the necessary computing tasks and return results on its own.
Jandali drew a parallel to software development, where AI progressed from autocomplete to increasingly autonomous coding. Anthropic is betting that the same arc is now underway in life sciences labs.
The goal is to speed up the drug discovery cycle, which could mean faster access to treatments and lower costs across an industry that spends hundreds of billions of dollars per year to bring new treatments to market.
Novartis CEO Vas Narasimhan, who also sits on Anthropic’s board, broke that timeline into three types of delay — information, operational and biological. He believes AI is already collapsing the first two.
“I think with the tools you saw today, we can bring information latency down almost to zero — I mean, the information will be at scientists’ fingertips. Operational latency, we can probably reduce significantly — this is organizing trials, organizing experiments, getting all of the work to happen in a large organization. But I think the biological latency we’re stuck with. To actually run the experiment in an animal model and cellular model, or in humans — that’s about 60% of this timeline,” Narasimhan explained during the event.
He estimated that AI platforms like Claude Science could help cut a drug candidate’s approval timeline from roughly 12 years to seven or eight — as well as potentially double the pharma industry’s success rate, from around 8% to 16%.
While Narasimhan thinks the impact on public health could be “massive,” he cautioned that one of pharma’s most persistent challenges — figuring out whether a given biological target is actually the right one to go after with a drug — remains largely unsolved by AI.
Another pharma CEO — Chris Boerner, who sits at the helm of Bristol Myers Squibb — also cautioned against overselling AI’s near-term potential.
“We don’t want to get over our skis,” he said, pointing to talk of AI “curing cancer in our lifetime” as the kind of promise the industry should be wary of making.
Still, he said the technology is already paying off in tangible ways — and he offered a window into what adoption looks like inside a major drugmaker already using Claude.
Boerner said BMS now runs all of its small molecule candidates, as well as a large share of its large molecule candidates, through AI screening before they ever reach a wet lab. The company has set an internal target of cutting drug development cycle times by 30% using AI tools — a goal Boerner said BMS is “well on our way” to surpassing.
The launch of Claude Science arrives at a pivotal moment for Anthropic. The announcement is one of the company’s most ambitious moves in the healthcare world — a sector it has increasingly courted as it looks to diversify revenue beyond coding tools like Claude Code.
The stakes are especially high as Anthropic prepares for its much-anticipated initial public offering. It’s a time during which a huge new customer base like pharma could be a major plus for the investor pitch.
The product launch also sharpens Anthropic’s rivalry with OpenAI, which has been pushing its own tools for pharma and medical research. In January, both companies rolled out platforms for healthcare providers within a week of each other.
In May, Anthropic closed a $65 billion Series H funding round, bringing its valuation to $965 billion.
Photo: BlackJack3D, Getty Images
