# Anthropic reports early results from using Claude across drug development with six companies

URL: https://ai-pulse.aglaganova.com/blog/anthropic-reports-early-results-from-using-claude-across-drug-development-with-six-companies
Published: 2026-11-04
Author: Roman Aglaganov, Full-stack developer
Category: Research
Region: Global
Tags: science, agents

> A shorter CAR-T design, 35,000 antibody candidates in a month and a manufacturing review that caught issues people missed. The results are early and mostly preclinical, and Anthropic now owns a small drug programme of its own.

![Illustration: a long chain of grey hexagon outlines, like a molecule, with one black ring at its centre and a single red atom](https://ai-pulse.aglaganova.com/api/media/file/drug-development-1.png)

Most stories about AI in medicine focus on one step, such as predicting a protein’s shape. Anthropic’s life sciences team argues that this misses the real problem. “There is no single problem to solve; the bottlenecks are distributed all throughout the drug development process,” it writes in a report published on 27 October. So for several months it worked with six drug developers to try Claude at many different stages, from designing molecules to planning clinical trials.

## What the partners say they got

**Design:** At Generate Biomedicines, six scientists used Claude Code to propose new versions of a genetic component of CAR-T cell therapy, aiming to make it work against solid tumours. One design was about 70% shorter than the standard part, and in lab tests the engineered cells carried about ten times as many receptors, killed tumour cells more efficiently and stayed active longer. At Divergence Labs, which works on food allergies, Claude produced roughly 35,000 antibody designs in about a month; early tests found 20 that dislodge the allergy antibody IgE, three of them more strongly than the standard medicine.

**Optimisation:** Connected to BigHat Biosciences’ lab data, Claude proposed changes to two antibody fragments, and the lab made and tested about 1,010 variants in ten weeks. When the usual stability measures looked misleading, Claude and an Anthropic researcher reanalysed the raw data and proposed new ones. The best variants bound more than ten times more tightly than the originals.

**Paperwork:** At Enveda, Claude reviewed 1,061 pages of manufacturing documents, caught all 11 issues human reviewers found and flagged seven they had missed, cutting review time from more than 25 hours to under three. Bristol Myers Squibb is using Claude to assess plans for an early-stage drug and estimate a trial’s chance of success, and The Column Group is setting up a new company, Pleiades, meant to run drug programmes with a small team and Claude.

## Anthropic’s own drug

Anthropic has also bought Mithril Medicine, a one-person company working on treatments for two rare forms of muscular dystrophy, whose founder used Claude to coordinate contracts and experiments with 11 outside labs. According to the report, a candidate went from idea to promising results in mice in under 18 months, and if it keeps holding up, Anthropic expects it could be ready for human trials as soon as late 2028.

## How much to read into it

These are the company’s and its partners’ own accounts, and almost all of the work is preclinical: none of these designs has yet been shown to help a patient. Divergence’s results are early hits rather than a drug, BigHat’s new measures were refined and run by human scientists, and Pleiades’ first tool was tested on just 45 target and disease pairs. What the report does show is a shift in ambition, from using AI to speed up one task to handing it pieces of the whole pipeline, with people still checking the work.

Disclosure: AI Pulse is written with AI tools, including Claude, which is made by Anthropic.

## Sources

- [Anthropic: Early progress on accelerating drug development with Claude](https://www.anthropic.com/research/early-progress-drug-development)