Related: AI screens 6 million molecules to find new antibiotics for drug resistant gonorrhea.
Anthropic is moving from making tools for scientists to attempting science of its own. On June 30 the company launched a drug discovery program focused on neglected diseases, the conditions that carry a heavy human burden but that traditional drugmakers tend to skip because they are not commercially attractive. It arrived alongside Anthropic‘s new research product, Claude Science, and together the two mark the company’s deepest push yet into healthcare, a field where several of the biggest AI labs are now placing large bets.
The neglected disease focus is a deliberate choice, and the reasoning is as much strategic as charitable. Eric Kauderer-Abrams, Anthropic’s life sciences head, framed the effort as a way to give the company firsthand experience of drug development while it builds AI tools for biopharma customers, describing the targets as areas outside the scope of what the traditional pharma and biotech landscape might consider attractive but that nonetheless carry real burden. In other words, Anthropic wants to learn how hard drug discovery actually is by doing it, and it chose problems where a success would matter and where it would not be competing head on with the industry it hopes to sell to.
The tool underneath the ambition is Claude Science, an AI workbench built for every stage of research. Anthropic says it brings the scattered tools and data researchers rely on into a single environment, connecting more than sixty scientific databases and a similar number of specialized skills across genomics, proteomics, structural biology, and cheminformatics. A researcher works with a coordinating agent that can run multi step analyses, while a separate reviewer agent checks citations and calculations and flags errors, and every output carries an auditable history so a result can be traced and reproduced. It runs on a Mac, a Linux machine, or a remote server, and can scale compute from a single graphics chip up to a lab’s high performance cluster, and it is available in beta to Claude Pro, Max, Team, and Enterprise subscribers.
The program did not come out of nowhere. In April, Anthropic paid roughly four hundred million dollars in stock for Coefficient Bio, a small startup of former Genentech computational biologists whose tooling covered drug pipeline planning, target identification, molecular design, and clinical trial strategy. Buying an eight person team for that sum signals that Anthropic was acquiring workflow expertise as much as software, the hard won knowledge of how a drug actually moves from idea to candidate. The drug discovery program and Claude Science are where that acquisition starts to show up as product.
Anthropic is not alone in this race, which is part of why the move matters. The launch lands directly opposite OpenAI, which released a rival research model called GPT-Rosalind in April, and it fits a broader pattern of frontier AI companies betting that science, and drug discovery in particular, is where the next serious value lies. The pitch across the field is similar, that an AI system able to read the literature, run analyses, and keep its own work honest could compress a discovery process that routinely takes many years. Anthropic is also funding up to fifty outside AI for Science projects with credits worth as much as thirty thousand dollars each, a way of seeding the ecosystem it wants to serve. It is the same company now putting Claude on NVIDIA chips inside Microsoft Azure, a sign of how broadly it is trying to plant Claude across industries.
The caution worth keeping is that this is an announcement of intent and a beta tool, not a shelf of new medicines. Researchers who have looked at Claude Science are optimistic but measured, noting that a general purpose AI can still hallucinate or miss nuance in regulatory guidance and assay design, errors that carry real consequences in drug development, and that the technology works best as a co pilot inside validated workflows rather than a shortcut to discovery. Neglected diseases have resisted attention for decades precisely because the science is hard and the money is thin, and no workbench erases those obstacles overnight.
Still, the direction is notable. An AI company choosing to take on the diseases the market ignores, and building the tooling to attempt them in the open, is a different kind of statement than shipping another chatbot feature. If Claude Science delivers even part of what Anthropic claims, it could speed the unglamorous middle of research, the searching, checking, and analyzing that eats a scientist’s day. Whether that translates into an actual treatment for a neglected disease is a question only years of work can answer, but Anthropic has now put itself on the hook to try.
Related on Entrelligence: AI screening 6 million molecules for new antibiotics, Claude running on NVIDIA GPUs in Azure, and the US lifting export controls on Claude Fable 5.
Your First 10 AI Skills
10 practical AI skills, copy-paste prompts and a 7-day plan to start using AI with confidence.
Download the guide →