project
Anthropic effort applying agent orchestration to research environments with specialist software, cloud compute and auditable artifacts.
Alec Tarashansky, lead engineer on Anthropic's Claude Science, has joined OpenEvidence, the physician search company valued at $12 billion in January. The part of Claude Science's design that fits most directly into a doctor's search tool is its citation-checking agent.
Reality
Anthropic says roughly 950 Claude agents spent 21 hours and 210 million tokens narrowing 200,000 DNA sequences to one new enzyme system named ART. The filtering ran almost entirely in software, the part of the setup builders can copy.
Reality
Anthropic's new in-house molecular biology lab says a 21.5 hour agent run turned a 1.9 billion cluster database into 19 reports for humans to read, and one of them described a repeat array nobody had recorded.
Reality
The company says the money will train the next generation of its EDEN models. The published evidence so far is one January bioRxiv preprint, co-authored with Nvidia, whose venture arm also put money into the round.
Reality
Reuters says the deliberation includes a safety review of the model itself and a judgment about profitability, and it comes days after Dario Amodei asked the industry to slow down building more capable systems.
Reality
The company's head of life sciences told Reuters that real lab work is still the final test in biology. Anthropic has not said which experiments the lab is running, and it has yet to run a clinical trial.
Perspective Coverage
5 publishersReality
The Danish drugmaker will test Anthropic's science workbench on specific R&D workflows while an earlier company-wide OpenAI agreement keeps running alongside it. Bloomberg reported that financial terms were not disclosed.
Reality
NVIDIA's BioNeMo Agent Toolkit answers a real question for agents that know a task needs folding but not which model to call. The price of that answer is a database download and a Claude Science sandbox you have to open.
Reality
Claude Science adds no new intelligence, only 60-plus database connectors and a reviewer agent. That was enough to move three research-tools vendors, and it reorders how you score AI suppliers.
Reality
Two contract labs made and tested Claude's designs, which is genuinely new. The field baseline those results are scored against comes from the company that ran the experiment.
Reality