Drug discovery scientists have a new shortcut. They do not need to leave Claude to run ADMET predictions anymore. That is the headline.
But the real story is older and slower — a history of broken workflows, fragmented software, and a quiet standardization push that finally made this integration possible. ADMET stands for Absorption, Distribution, Metabolism, Excretion, and Toxicity.
It is the gauntlet every drug candidate must run. Most fail. The industry has known for years that machine learning models can predict these properties faster and cheaper than lab assays.
But the models lived inside specialist software. Bench scientists did not use that software day-to-day.
So the models stayed on the shelf. That gap — between a good model and a usable tool — is what Anthropic and Inductive Bio have been working to close. The connector announced June 30 is live, not a demo.
Any scientist using Claude can call Inductive’s prediction models straight from the chat interface. No new platform. No toggling between tools.
Under the hood sits Anthropic’s Model Context Protocol. MCP is an open standard.
It handles the handshake between a large language model and external tools or data. Anthropic released it quietly. The company has been assembling a life sciences ecosystem around Claude piece by piece.
Inductive is the latest addition to that build. Inductive Bio describes itself as an AI drug discovery partner developing virtual chemistry labs.
The company built global prediction models for all five ADMET properties. But building models is one thing. Getting them used is another.
The drug discovery industry is famously conservative. Scientists trust their own assays. They trust software they already know.
They do not trust black boxes they have to learn. So Inductive bet on integration instead of replacement.
The MCP connector does not ask scientists to change their workflow. It puts the prediction engine inside the tool they already use. The handshake is invisible.
The scientist asks a question. Claude calls Inductive’s model.
The answer comes back in the chat window. This is not a research preview. The connector is live.
That matters because the industry has seen plenty of demos that never made it to production. Inductive is betting that convenience will drive adoption where accuracy alone could not. The best model in the world is useless if nobody calls it.
Anthropic’s life sciences ecosystem remains small. But the logic is clear.
Claude is a general-purpose assistant. Drug discovery is a specialized domain. The ecosystem model lets Anthropic bring in specialized tools without building them in-house.
Inductive gets distribution. Anthropic gets a more useful Claude for scientists.
The scientists get predictions without friction. Whether that is enough to change how the industry adopts machine learning for ADMET is an open question. The models have been ready for years.
The workflow integration has not. Now it is. The connector exists.
The scientist just has to use it.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Inductive Bio integrates ADMET predictions directly into Anthropic’s Claude platform”, “description”: “Drug discovery scientists have a new shortcut. They do not need to leave Claude to run ADMET predictions anymore. That is the headline. But the…”, “datePublished”: “2026-06-30T16:39:00+00:00T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Ava Anderson”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
