Stanford researchers will discuss their agentic 'scientists' that are on course to reshape drug discovery at VB Transform 2026
Our take
The inefficiencies plaguing drug discovery are well-documented, a frustrating reality for researchers and a significant barrier to medical advancement. The staggering failure rate – reportedly 90% to 95% – and the exorbitant costs associated with bringing a single drug to market underscore a systemic problem. This isn't simply about bad luck; it's about fractured workflows and the inevitable loss of context as projects are handed between specialized teams. Recent advancements in generative AI have offered glimpses of potential solutions, but Stanford’s work with agentic AI represents a truly transformative shift, building upon the foundational work seen in articles like Xiaomi's HarnessX rewrites its own AI scaffolding mid-task — and smaller models gain the most and mirroring the strategies explored in Alibaba's model never trained as an agent — and improved agent performance across seven benchmarks, where the focus is on enabling AI to handle increasingly complex, long-horizon tasks. Professor Zou’s team’s approach, deploying thousands of autonomous AI "scientist" agents within a virtual biotech, promises to address these shortcomings head-on by maintaining continuity and context throughout the entire drug development lifecycle.
The hierarchical orchestration framework is particularly compelling. Rather than a monolithic AI attempting to manage every aspect of drug discovery, the system leverages a team of specialized agents, each focused on a specific task—discovery, safety, analysis—all guided by a central “chief scientist officer.” This structure mirrors the way human research teams operate, but with the added benefit of constant communication and data sharing within the AI ecosystem. The emphasis on "agent-native" data, ensuring the AI has access to and can effectively synthesize vast datasets, further strengthens the approach. The use of a mixed-model architecture, drawing on tools like Claude for coding and data analysis while incorporating fine-tuned models for specialized use cases, demonstrates a pragmatic and adaptable approach to leveraging the power of AI. It’s a move away from the hype around singular, all-powerful models towards a more modular and robust system – a strategy that aligns with the growing understanding of how to best deploy AI in complex environments.
The potential impact of this technology extends far beyond simply accelerating drug discovery. By streamlining workflows and reducing failure rates, it could dramatically lower the cost of bringing new treatments to patients, expanding access to life-saving medications. The Human Intelligence startup, currently valued at roughly $1 billion, is clearly banking on this transformative potential, and the insights Zou will share at VB Transform promise to offer a glimpse into the practical considerations of building and managing such a system. The challenges of context management, data transformation, and ensuring agent trustworthiness, as highlighted in Zou’s session and echoed by the work of companies like Zillow, discussed in Mistral launches OCR 4, turning document extraction into a full enterprise AI play, are critical to successful implementation.
The emergence of agentic AI in drug discovery signals a fundamental shift in how we approach scientific research. While the technology is still in its early stages, the potential to automate and accelerate the process, while simultaneously improving success rates, is undeniable. The question now is not *if* AI will reshape drug discovery, but *how quickly* and what new ethical and regulatory considerations will arise as these increasingly autonomous systems take on greater responsibility for human health. The ability to effectively manage context and ensure the integrity of agent actions will be paramount to realizing the full potential of this transformative technology, and the lessons learned from Stanford's virtual biotech will be invaluable to anyone seeking to build the future of medical research.
Drug discovery is notoriously inefficient. Pharmaceutical projects span years, moving from one specialized human team to the next through disconnected workflows that result in knowledge loss during each handoff.
A shocking 90% to 95% of drug discovery projects reportedly fail — one of the highest failure rates of any industry. A single successful drug can take over a dozen years and up to $1 billion from initial discovery to patient distribution, according to published reports.
Generative AI is being used to solve some of the challenges, but Stanford researchers have moved the ball forward with agentic AI.
A team led by James Zou, associate professor of Biomedical Data Science at Stanford University, has deployed thousands autonomous AI "scientist" agents in a virtual biotech that simulates the full lifecycle of drug development. The agents handle everything from initial discovery through safety testing and clinical trial design, while maintaining the continuity that’s lacking in today’s drug discovery processes, according to Zou.
The project uses a hierarchical orchestration framework. At the top sits a chief scientist officer agent that acts as a planner, delegating tasks to teams of specialized agents, Zou told VentureBeat during a call ahead of his upcoming session at VB Transform 2026.
While one team of agents focuses on discovery, another manages safety, and others handle specialized analytical tasks. Because these agents operate within a unified, hierarchical ecosystem, they retain the full context of a project, maintaining continuity from the first molecule identified to the final clinical outcome.
The "brain" of the system relies on a vast amount of primary data. The agents are granted access to data sources ranging from genomics and FDA chemistry data to clinical trial databases using a model context protocol.
The team has invested heavily in agent-native and agent-friendly data, allowing the AI to synthesize complex information more effectively. The system relies on a combination of models, with Zou noting that while Claude often serves as the backbone for coding and data analysis, the architecture employs a mixture of models, including those fine-tuned specialized use cases.
Zou is raising money at a roughly $1 billion valuation for his startup, Human Intelligence, based on the research.
During Zou’s session at VB Transform on July 15, titled How 10,000 agentic scientists in Stanford’s lab are set to revolutionize medical research and discovery, he will share valuable insights including strategies for managing context and long-running, multi-step workflows in a multi-agent system, the process of transforming and indexing raw enterprise data to make it agent native, and how to use human auditing and experimental reward signals to verify agent actions.
Another session at VB Transform focused on the value of agentic context includes Building a trustworthy agentic AI foundation: How Zillow accelerated engineering by 40%, with Zillow's SVP of engineering and technology, Toby Roberts and Glean’s CEO Arvind Jain.
Interested in attending VB Transform 2026? Register here. A select number of complimentary passes are also available to senior technology leaders. Contact us to get yours.
Read on the original site
Open the publisher's page for the full experience