Cases

Examples of cases that AI is capable of solving

Case 1: First Human Trials of AI-Designed Drugs

Goal: Discover a new antibiotic capable of killing drug-resistant bacteria.

Method: Use a deep learning model to analyze a vast library of chemical compounds and predict antibacterial activity.

Process: The AI algorithm screened over 100 million chemical structures and identified a promising candidate. Researchers then tested this compound in the laboratory and in mouse models.

Result: A new antibiotic, named Halicin, was discovered. It successfully kills multiple strains of multi-drug-resistant bacteria and represents a new class of antibacterial compounds.

Case 2: AI Discovers a New Antibiotic

Goal: Discover a new antibiotic capable of killing drug-resistant bacteria.

Method: A deep learning model was used to analyze a vast library of chemical compounds and predict antibacterial activity.

Process: The AI algorithm screened over 100 million chemical structures and identified a promising candidate. Researchers then tested the compound in vitro and in mouse models.

Result: A novel antibiotic, named Halicin, was discovered. It effectively kills multiple drug-resistant bacterial strains and represents a new class of antibacterial agents.

Case 3: AI-Designed Drug for Fibrosis

Goal: Develop a new drug for idiopathic pulmonary fibrosis (IPF) using AI.

Method: An integrated AI platform was used to identify a novel disease target and generate an active compound against it.

Process: AI algorithms analyzed large omics datasets and identified the kinase TNIK as a promising target. A generative model then designed multiple novel molecules, from which a lead candidate was optimized. Target discovery and compound advancement to the preclinical stage were completed in just 18 months.

Result: An experimental drug (later named Rentosertib) was developed, successfully passing preclinical studies and entering clinical trials—making it one of the first AI-discovered medications of the modern era.

Case 4: AI-Driven Drug Repurposing
(COVID-19)

Goal: Rapidly identify an existing drug capable of treating COVID-19.

Method: A knowledge graph-based AI platform was deployed to map SARS-CoV-2 infection mechanisms against a database of approved drugs (drug repurposing).

Process: BenevolentAI’s system analyzed biomedical data on COVID-19 and profiles of thousands of compounds. Within days, the platform pinpointed baricitinib (an anti-inflammatory drug approved for rheumatoid arthritis) as a dual-action candidate: it could both inhibit viral entry into cells and mitigate cytokine storm. This hypothesis was published in The Lancet in February 2020, prompting expedited global clinical trials.

Result: The AI’s prediction succeeded—baricitinib reduced COVID-19 mortality in trials and was subsequently adopted into WHO treatment guidelines, marking one of the fastest examples of AI-driven drug repurposing in a pandemic.

Case 5: Integrated AI-Lab Platform Accelerates R&D

Goal: Automate and accelerate the design–make–test–learn (DMTL) cycle in drug discovery by integrating AI with robotic laboratories.

Method: Develop an AI-driven platform where generative AI designs candidate molecules and automated labs synthesize and biologically test them with minimal human intervention.

Process: Exscientia implemented an iterative DMTL workflow on a cloud infrastructure. AI continuously proposes novel molecules, while robotic systems synthesize and assay them 24/7. This approach:
- Reduced the number of compounds needed to identify a viable candidate by 10× vs. traditional methods.
- Accelerated drug design by ~70%, compressing years of work into months.

Result: The AI-automation synergy boosted R&D efficiency:
- Design cycles sped up by up to 70%.
- Capital costs cut by 80%.
- Multiple AI-designed candidates advanced to clinical trials.

This breakthrough demonstrates how "self-driving labs" powered by AI can transform pharmaceutical innovation.