The cocktail containing Omnii-1-designed phage variants eliminated the patient’s antibiotic-resistant isolate in vitro,
Radical Numerics explained regarding a historic treatment announcement.
An artificial intelligence-designed bacteriophage cocktail has been administered to a human patient in the United States for the first time by San Francisco-based AI company Radical Numerics, marking a transition for biological AI from laboratory simulation to actual human administration under a compassionate-use treatment protocol.
Omnii-1 AI Model Designs Bacteriophages for Patient Treatment
The treatment utilized a proprietary AI model named Omnii-1 to design viable bacteriophages, which are viruses that specifically infect bacteria without harming human cells. Radical Numerics announced the milestone on Oct. 22, stating that the investigational cocktail was administered to a patient under compassionate use after successfully eliminating the patient’s antibiotic-resistant isolate in laboratory tests. In 2025, Stanford University and the Arc Institute also conducted joint research on accelerating phage design using artificial intelligence, creating E. coli-targeting phages that outperformed their biological predecessors.
Despite the milestone, the company has not disclosed the treatment outcome, the specific type of infection, or patient details, citing patient privacy and the limitations of drawing conclusions from a single case. Radical Numerics noted that observations from a single patient do not establish general safety or effectiveness, and the company did not respond to questions for comment.
AI Phages Target Antimicrobial-Resistant Bacteria and Superbugs
The development highlights the potential for AI to accelerate personalized treatments targeting individual patients’ pathogens, offering a potential alternative to combat antimicrobial-resistant bacteria that render traditional antibiotics ineffective. Infections caused by multidrug-resistant superbugs pose serious risks during surgeries, organ transplantations, and cancer treatments, as well as for conditions like pneumonia and sepsis. Phages also offer the advantage of sparing beneficial gut bacteria that broad-spectrum antibiotics can affect.
While conventional methods for identifying and preparing suitable bacteriophages can take weeks or months, AI could shorten the design process to just a few days, though actual administration timelines still depend on efficacy validation, production, and safety testing. Radical Numerics hopes the work will help establish a standardized process for designing personalized phage therapies. Drug development typically takes around 10 years with costs exceeding $2 billion including failed efforts, while only about 10% to 12% of drugs entering Phase 1 trials win approval, fostering close industry ties such as Eli Lilly partnering with Insilico Medicine and AstraZeneca collaborating with insitro.
