2025: AI-Designed Drug Enters Phase II (2025)
By 2025, multiple drug candidates designed with substantial AI contribution had advanced into
Phase II clinical trials, with INS018_055—a small-molecule inhibitor of TRAF2- and NCK-
interacting kinase (TNIK) for idiopathic pulmonary fibrosis (IPF), discovered by Insilico Medicine
using its end-to-end generative AI platform—reporting Phase II interim results in 2024 and
continuing enrollment in 2025, representing the most advanced fully AI-generated clinical asset.
The compound was discovered in under 30 months from target identification to IND-enabling studies,
compared with an industry median of 4–5 years for the same stages.
Insilico's PandaOmics platform applied multi-omics data from lung fibrosis patients to rank novel
targets by disease relevance and druggability; the Chemistry42 generative chemistry platform then
used reinforcement learning and graph neural networks to design de novo molecules satisfying
specified ADMET and binding profile objectives against TNIK. The resulting compound was synthesised,
profiled in vitro and in vivo, and advanced to Phase I without identifying a series by conventional
high-throughput screening or fragment-based approaches—relying entirely on in silico generation.
Phase I results, reported in Nature Biotechnology in 2023, demonstrated a safety and
pharmacokinetic profile consistent with the AI-predicted ADMET characteristics and warranted
Phase II evaluation in IPF patients. The Phase II trial, initiated in 2023 and actively enrolling
in 2025, is measuring change in forced vital capacity (FVC) as the primary endpoint—the accepted
efficacy measure in IPF.
The broader AI drug discovery landscape in 2025 includes dozens of Phase I and Phase II assets
from companies including Recursion Pharmaceuticals, Exscientia, BenevolentAI, Relay Therapeutics,
and Schrodinger, across oncology, immunology, and rare disease indications, collectively marking
the transition of AI-driven drug discovery from a research capability to a clinical development
reality.
Bu Neden Önemliydi
The advance of AI-generated drug candidates into Phase II trials demonstrated that artificial
intelligence can meaningfully compress the drug discovery timeline—from years to months—by
integrating target identification, molecular generation, and ADMET prediction into end-to-end
computational pipelines. The 2025 milestone marked the transition from proof-of-concept to
clinical evidence that AI-designed molecules can reach therapeutic evaluation, establishing a
new benchmark for R&D productivity and attracting unprecedented investment in AI-first
pharmaceutical development.