Privately-held Mindbeam AI, a member of the NVIDIA Inception program, said its latest research suggests generative AI could accelerate the search for safer pain medications by identifying novel compounds with improved predicted liver safety compared with acetaminophen. The company develops infrastructure for optimizing large language model training and inference using NVIDIA GPU-based computing. Mindbeam designed and evaluated 24 novel drug candidates targeting TRPV1, a receptor involved in pain signaling, and efficacy and toxicity assessments narrowed the group to three lead compounds that showed strong potential as future pain therapies. One of the three candidates delivered the most favorable overall profile, balancing predicted efficacy, bioavailability, and tolerability characteristics. A central objective of the research was to address liver toxicity risks associated with acetaminophen, as chronic high-dose use can cause liver damage. The company said its AI-driven approach identified compounds with improved predicted liver safety. Mindbeam noted the research also illustrates how AI can compress early-stage drug development timelines while improving efficiency, building on its Litespark framework designed to accelerate generative AI model training while reducing computing costs and energy consumption. The company emphasized these findings remain preliminary.