AI Model Speeds Antibody and Cell-Therapy Design
Biotech partners report faster design and review workflows, but these are early company-reported results rather than validated clinical outcomes.
Updated on Nov. 2, 2026 in Biotech
Live Poll
Do you think AI-assisted drug development will lower what you pay for medicine?
Anthropic is expanding Claude's role in pharmaceutical research through partnerships with biotechnology developers. Several early applications were reported: Generate Biomedicines used Claude Code to shorten a CAR T-cell therapy component by roughly 70%, and Enveda used it to review manufacturing documentation. The reported results are company-described applications, not peer-reviewed clinical outcomes.
Why it matters
Drug development has long been bottlenecked by lab iteration and regulatory paperwork, and these examples show a language model being applied to both design search and document review. The shift matters now because the companies involved are moving from single experiments toward extended collaboration, including early clinical decision support.
Reported figures include a roughly 70% shorter CAR T-cell component and about ten-fold more receptor production in lab cells, both measured against the unmodified baseline. Enveda's documentation review cut manual checking time by 89% compared with human reviewers, who had previously identified 11 issues, while the system caught those 11 plus seven more. These are internal company measurements rather than independent benchmarks.
The players
Anthropic
The AI developer behind the Claude model family, which is now being applied to life-science research workflows.
Generate Biomedicines
A biotechnology company that used Claude Code to modify a CAR T-cell therapy component.
Divergence Labs
A drug developer that used Claude to generate antibody candidates for a potential food-allergy treatment.
BigHat Biosciences
An antibody developer that combined Claude with automated laboratory equipment to improve two antibody fragments.
Enveda
A biotechnology company that applied Claude to review manufacturing documentation.
The details
Claude Code, a version of Anthropic's model built for writing and editing code, was used by six scientists at Generate Biomedicines to redesign part of a CAR T-cell therapy, a treatment that engineers a patient's immune cells to recognize cancer. The team aimed to free up space in the construct for genetic instructions targeting solid tumours. At Divergence Labs, Claude generated roughly 35,000 candidate molecular designs for antibodies aimed at IgE, an immune protein involved in food allergy, and BigHat Biosciences paired the model with automated lab equipment so that proposed antibody fragment changes could be built and tested in bulk.
Timeline
One month: Divergence Labs' system generated approximately 35,000 molecular designs.
Ten weeks: BigHat researchers manufactured and tested roughly 1,010 antibody fragment variations.
The Tech Race
The effort places a general-purpose AI model directly into antibody and cell-therapy design, a field where companies such as Generate Biomedicines and BigHat Biosciences are building proprietary platforms. The collaboration pattern here is one frontier AI developer supplying a model to many smaller specialist firms rather than building drug candidates itself.
The tools described are being used inside partner companies' research pipelines, not offered as products to patients or the general public. Any downstream effect on treatments depends on whether these candidates advance through clinical trials, a timeline the companies have not specified.
The takeaway
The trajectory to watch is whether AI-generated antibody designs and AI-reviewed manufacturing documents move from lab results into human trials. Track any announced early clinical development findings from Bristol Myers Squibb's assessment of clinical study plans, which the companies say will be the next step in their collaboration.
Further reading
For broader background on how software and AI tools are reshaping lab work, see Biotech coverage.
Source note: This article includes information reported by TechRadar.
Live Poll
Do you think AI-assisted drug development will lower what you pay for medicine?
