October 30, 2026
Barry Bunin, PhD
Founder & CEO
Collaborative Drug Discovery
“A.I. Agents as ‘Co-Scientists’? This Lab Says They Could Speed Drug Research.” That’s the headline for an article in The New York Times about how scientists unleashed thousands of AI agents on drug discovery tasks and were impressed with what they found. Actually, this is a fairly common type of headline as everyone tries to filter the very real substance from the very real hype in AI. The work of Stanford University’s Dr. James Zou and colleagues was recently published in Science with the title “The Virtual Biotech: A Multi-Agent AI Framework for Therapeutic Discovery and Development.” The researchers describe a process in which multiple AI agents discussed the same points in parallel conversations which could take place within seconds, in an effort that ran 24 hours a day, seven days a week. “I’m very jealous that it doesn’t get tired,” Dr. Zou said. “It just goes 24/7, and its meetings are so much more efficient than mine.” The researchers say that within a few days the AI agents had found dozens of novel proteins that would theoretically bind to new COVID variants. The article says testing showed two of them actually worked — created at a fraction of the time and cost it would take a lab of humans. The lab also reports that tens of thousands of AI agents predicted which approach to attacking a protein involved in lung cancer would be most effective at treating the disease in clinical trials, a finding they say was later corroborated by work performed independently by researchers at a large pharmaceutical company. Dr. Kyle Swanson, a former graduate student in Dr. Zou’s lab who was involved in the COVID vaccine research, said “It really does feel like a new era where one person plus a bunch of GPT or Claude models can really make significant, fast progress.”
The CDD Vault AI module focuses on delivering practical capabilities without overpromising like reasonable generative bioisosteres to experts for every molecular moiety, zero click models continuously optimized as additional data is generated, and best of breed open models like Boltz-2, Alphafold-2 and most recently Lilly TuneLab. See the latest technical release notes for details around ChemLab and AbLab TuneLab models in CDD Vault.
* * *
“Autonomous AI Agents Discover Reverse Transcriptases with Tandem Repeat Arrays.” That’s the title of a paper recently released by Anthropic reporting that a team of autonomous large language model agents surveyed reverse transcriptase (RT) genes across 1.9 billion metagenomic protein clusters and surfaced a new family of the enzymes, along with the repeat arrays that define it. Drug Discovery News writes about the study: “AI has already changed how scientists predict protein structures and design new molecules. The next leap may be discovery itself, with agents that read raw biological data, spot what looks strange, and chase it down on their own. [This] study offers an early glimpse of that future, and of what it could mean for the hunt for the next generation of therapeutic tools.” Of note, the article was written by people Peter H. Yoon, Januka S. Athukoralage, Emmanuel Ameisen, Eric Kauderer-Abrams, Nicholas T. Perry, and Matthew G. Durrant. The Anthropic team built a harness in which worker agents planned and ran analysis tasks while supervisor agents reviewed the results and opened follow-up tasks. The campaign ran for 21.5 hours without human intervention, spanning 949 agent sessions and 119 tasks, 98 of which the agents opened themselves. Insight into the logic the AI agents used is shown in this passage reported by Drug Discovery News: “‘I can see by eye a tandem repeat array,’ the agent wrote, according to the session transcript published in the paper. After a literature search turned up no match, the agents filed a report for human review.”
For broader societal context, I’ve written a perspective that I tried to keep balanced titled, “CDD’s Response to Anthropic CEO Dario Amodei’s Provocative Blog Post ‘We Must Pace the Future.’”
* * *
“Evaluating Generative Models for Structure-Based Drug Discovery from De Novo Design to Lead Optimization.” That’s the title of a study recently published in Nature, which evaluates current methods of structure-based drug design, describes their limitations, and makes the case for the value of MolGenBench. While MolGenBench was released earlier this year, the just-published study provides data behind why the new benchmarking tool is seen as significant. The researchers write in the abstract to their 23-page study: “We introduce MolGenBench, a comprehensive benchmark for de novo molecular generation and hit-to-lead optimization built on a large, structurally diverse dataset. Evaluation of 17 leading methods reveals persistent limitations, including weak virtual-screening performance, limited coverage of target-specific bioactive space, frequent high-risk structural motifs, poor target awareness, and reduced generalization to unseen proteins. Although some models improve three-dimensional conformation generation, they still struggle to produce active molecules reliably. MolGenBench provides standardized data, metrics, and analyses that connect structural prediction with practical molecular design and guide the development of more effective structure-based generative models.” CDD has been a partner of other model comparison methods as part of the OpenADMET consortium.
* * *
“Here’s How Much a Flu Vaccine May Lower Your Dementia Risk.’” That’s the headline from The Washington Post about research suggesting that older adults who get the vaccine may have a 40 percent lower risk of Alzheimer’s than those who do not. Dr. Trisha Pasricha, Assistant Professor of Medicine at Harvard Medical School and a physician at Beth Israel Deaconess Medical Center, points to a large study in which researchers analyzed health records of 1.8 million adults ages 65 and older, and found that those who received at least one flu vaccine were 40 percent less likely to develop Alzheimer’s disease in the following four years. Dr. Pasricha points to a 2025 meta-analysis of eight studies reported that people already at higher risk for dementia had greater benefits with more flu shots: Two or three doses reduced dementia risk by 16 percent, while more than four reduced the risk by 57 percent. Other vaccines, including for shingles, RSV and Tdap, are also linked to reduced dementia risk.
With all the irrational fears circulating about vaccines, which have a higher bar for safety than other drugs, it is amazing to see all these additional, unexpected benefits. Vaccines have proven to be among the safest and most effective tools in the modern health arsenal.
Barry A. Bunin, PhD, is the Founder & CEO of Collaborative Drug Discovery, which provides a modern approach to drug discovery research informatics trusted globally by thousands of leading researchers. The CDD Vault is a hosted biological and chemical database that securely manages your private and external data.