AI is being used across drug discovery, clinical operations, manufacturing, and regulatory review—but “deployment” can mean anything from a company-described workflow to an early proof of concept. These 11 examples show what organizations say they are doing, how mature each use appears to be, and what the evidence does—and does not—establish.
Eleven documented uses of AI in pharma and biotech
The examples below are grouped by lifecycle stage. Company filings and annual letters describe the organizations’ own work; their reported milestones and performance figures are not independent evaluations. The FDA example concerns the agency’s internal review process, not a pharmaceutical company discovering a drug.
As an Amazon Associate I earn from qualifying purchases.
| Example | Stage and maturity | What is reported—and what it establishes |
|---|---|---|
| 1. Pfizer’s OncoScout | Drug discovery workflow | Pfizer’s 2024 annual review says its AI-powered OncoScout platform supports the search for compounds with potential to become cancer medicines. This documents a discovery workflow; it does not establish that OncoScout produced an approved medicine. |
| 2. Recursion OS | Integrated discovery and development platform | In its 2024 Form 10-K, Recursion describes Recursion OS as combining proprietary experimental data with computational models across target discovery and development. The company presents it as a way to connect wet-lab experiments with computational work; this is the company’s description of its platform. |
| 3. Recursion’s clinical-stage programs | Drug development; candidates in clinical stages | Recursion’s filing reports clinical-stage programs and says REC-617 was optimized using its AI platform, alongside updates on multiple candidates. That supports describing AI as part of the development process, not as the sole cause of clinical progress or evidence of efficacy. |
| 4. Recursion with Roche and Genentech | Partnered discovery research | Recursion reports whole-genome and chemical perturbation maps in gastrointestinal oncology, plus a neuroscience phenomap. The company says the neuroscience work triggered a milestone. Both the platform contribution and milestone are reported by Recursion; no downstream clinical benefit is quantified in the cited filing. |
| 5. Recursion with Sanofi | Partnered discovery; target advancement | Recursion reports milestones for advancing immunology and oncology targets into lead optimization. This is a discovery-stage partnership milestone, not a clinical outcome. |
| 6. Recursion with Bayer | Research data and workflow software | Recursion reports delivering 25 multimodal oncology data packages and LOWE, an LLM-orchestrated workflow tool. Its filing does not quantify a resulting clinical benefit. |
| 7. Recursion with Merck KGaA | Discovery partnership | Recursion describes an alliance to identify potential first-in-class or best-in-class targets in oncology and immunology. The cited filing characterizes this as target discovery, not a clinical-stage result. |
| 8. Takeda’s trial recruitment and protocol work | Clinical operations; company-described capability | In its 2024 CEO annual letter, Takeda says it can use data and AI to identify potential trial participants, facilitate recruitment, and optimize protocols. The letter does not quantify recruitment gains or describe a measured outcome for this capability. |
| 9. Takeda’s real-world evidence for GammaGard Liquid | Evidence supporting a regulatory submission; not described as AI-driven | Takeda says real-world evidence from licensed databases supported, in part, the January 2024 U.S. FDA approval of GammaGard Liquid for chronic inflammatory demyelinating polyneuropathy (CIDP). Takeda also says the approach reduced timelines by several years. The letter does not say AI generated the evidence, so this is an example of data-supported evidence generation, not an AI-driven approval. |
| 10. Takeda’s manufacturing quality work | AI/ML proof of concept; scale-up planned after testing | Takeda describes proof-of-concept work using AI and machine learning for risk classification and deviation summaries, with scaling planned after testing. This is not evidence of established, network-wide deployment. |
| 11. FDA’s generative-AI scientific-review pilot | Regulatory review; completed internal pilot | In a May 8, 2025 announcement, the FDA said it had completed its first generative-AI scientific-review pilot and set a goal of deploying a common secure system across centers by June 30, 2025. The announcement states the target, not that every center met it. FDA reviewer Jinzhong (Jin) Liu said of the pilot: “This is a game-changer technology that has enabled me to perform scientific review tasks in minutes that used to take three days.” That is his account of the pilot, not a validated average across reviewers or tasks. |
What the reported performance figures mean
Pfizer’s annual review also describes AI and machine learning for demand prediction across about half of its products. For its Medical AI Assistant, Pfizer reports a 40% reduction in first-draft manuscript production time and a 15% reduction in overall manuscript submission time. It also estimates $735 million in annual impact from AI use cases. These are Pfizer’s 2024 company-reported figures; the cited passage does not present them as independently audited causal estimates. They illustrate why the source and scope of an impact number matter as much as the number itself.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →What FDA activity says about AI in drug development
The FDA’s Center for Drug Evaluation and Research (CDER) says it drew on experience with more than 500 submissions containing AI components from 2016 through 2023 while developing its approach to AI in drug development. That is a count of submissions with AI components—not a count of approvals, AI-created medicines, or proof that the AI component determined an outcome. CDER’s page also points to 2025 draft guidance on AI information intended to support regulatory decisions about safety, effectiveness, or quality.
#1 Best Overall
The FDA’s Center for Biologics Evaluation and Research (CBER) describes a risk-based review approach that considers the AI method, its specific application, and relevant product and clinical factors. The agency’s AI/ML information for products regulated by CBER is useful context, but neither agency’s material amounts to blanket validation or approval of “AI-developed” medicines as a class.
How to judge a claimed deployment
- Check the maturity level. An operating workflow, a development-stage program, a partnership milestone, a pilot, and a proof of concept are different kinds of evidence. The Takeda manufacturing example, for instance, is explicitly a proof of concept with scale-up planned after testing.
- Separate the tool from the outcome. A platform being used in discovery does not show that it produced a successful candidate. Likewise, a milestone reported by a partner does not establish clinical benefit.
- Read company figures as attributed claims. Annual reports and shareholder letters can document what a company says it is doing, but self-reported time savings, financial impact, and milestones should not be mistaken for independent measurement.
- Ask whether the use is actually AI. Takeda’s GammaGard account concerns real-world evidence and licensed databases; the cited description does not attribute the evidence or approval to AI.
- Keep expert and experimental judgment in the picture. AI can support prioritization, analysis, and workflow, but the examples do not establish that it replaces scientific, clinical, manufacturing-quality, or regulatory judgment. Drug candidates still require appropriate experimental and clinical evaluation.
What these 11 examples show
AI use in pharma and biotech is not one uniform technology or maturity level. The documented cases range from discovery platforms and partnered research to clinical operations, a manufacturing proof of concept, data-supported evidence generation, and an FDA internal review pilot. The strongest conclusion is that organizations are incorporating AI into specific workflows—not that AI has independently been shown to deliver approved medicines or a general, quantified improvement across the industry.
Quick Recap
Rank #2
- Used Book in Good Condition
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
