Insilico Medicine ("Insilico"), a clinical-stage generative artificial intelligence (AI)-driven drug discovery company , today announced the launch of inClinico Agent, an advanced agentic AI workspace engineered to automate, optimize, and elevate clinical trial intelligence. Powered by a curated proprietary database of over 100,000 trials, endpoint-specific machine learning models, and specialized clinical analysis capabilities, inClinico Agent compresses months of complex clinical landscape analysis into minutes.
Clinical development teams routinely spend months navigating fragmented registries, heterogeneous literature, and disparate databases to inform trial design and assess competitive landscapes. Determining whether a proposed clinical study will achieve its primary endpoint has historically depended on subjective expert opinion that is difficult to replicate or audit.
inClinico Agent is an agentic orchestration framework that converts natural-language queries into structured, end-to-end analytical workflows and pairs frontier LLMs for high-level reasoning with deterministic tools for execution. At its foundation, the system leverages a curated proprietary database of over 100,000 clinical trial records covering drugs, targets, indications, and study phases. The database integrates structured metadata, standardized outcome metrics, and granular free-text results. To overcome data cutoff limitations, the agent dynamically queries live public registries such as ClinicalTrials.gov for the latest records. Every citation in a report is drawn from the agent's own retrievals and checked against the claim it supports.
To evaluate clinical programs, inClinico Agent employs two gradient-boosted machine learning models trained on features available at trial registration, including biological evidence, trial design rigor, and operational metrics. Each trial is routed to the model that matches the type of primary endpoint its sponsor registered. On an independent, held-out test cohort comprising trials initiated in 2021 or later, the Efficacy Endpoint Model achieved an area under the receiver operating characteristic curve (AUROC) of 0.828 (95% CI 0.812–0.845; n=2,347), while the Safety/Toxicity Endpoint Model achieved an AUROC of 0.809 (95% CI 0.787–0.831; n=1,485). These results suggest that both models can meaningfully distinguish trials likely to meet their respective endpoints from those unlikely to. Each prediction is accompanied by feature attributions showing which factors raised or lowered it.
The platform generates a Predictive Trial Index (PTI) ranging from 0 to 100, scaling with a trial's estimated likelihood of meeting its primary endpoint. On a real-world validation set of 939 trials completed in 2025, the platform achieved an overall AUROC of 0.805 and an accuracy of 0.734, with Phase 3 trial accuracy reaching 0.780 (AUROC 0.828).
To rigorously evaluate model performance without retrospective bias, Insilico introduced ClinBench, a prospective benchmark consisting of 3,311 ongoing Phase I–III clinical trials slated for completion by December 31, 2027.
In a head-to-head prospective benchmark, inClinico Agent was evaluated alongside five leading frontier LLMs (GPT-6-Astra, Claude-Fable-5-1, Grok-4.6, Gemini-3.1-Pro, and DeepSeek-V4-Pro) with live search capabilities enabled. Predictions were locked prior to outcome disclosure. While LLMs optimistically predicted success for ~75% or more of trials (75%–82% across Phases I–III), inClinico Agent predicted success for approximately half of the cohort (50%)—aligning closely with historical industry success rates (53% in the test set). Benchmark methodology and predictions have been published on medRxiv (https://doi.org/10.64898/2026.10.13.26365373).
The current version of inClinico Agent integrates eight domain-specific capabilities to support strategic decision-making and regulatory readiness across the drug development lifecycle. These capabilities include:
"In clinical development, data fragmentation and subjective predictions often lead to costly delays," said Frank Pun, PhD, Head of Insilico Medicine Hong Kong, whose team led the development of inClinico Agent. "inClinico Agent unifies rigorous machine learning scoring, transparent reasoning, and regulatory-grade document workflows into a single workspace, empowering biopharma teams to make data-backed clinical decisions with speed and confidence."
inClinico Agent is now available to industry partners and biopharmaceutical enterprise clients. To request a demo or explore partnership opportunities, please contact bd@insilico.com.
inClinico Agent represents another pivotal component of Insilico’s frontier AI suite. Since beginning its strategic expansion into frontier AI in early 2026, Insilico has introduced the MMAI Gym model-training and evaluation ecosystem, PandaOmics Agent, LabClaw, and a range of specialized benchmarking ecosystems, extending its capabilities beyond AI applications built specifically for drug R&D. Building on these cutting-edge technologies, Insilico has established AI co-development partnerships with industry pioneers, including Liquid AI, focused on liquid foundation models, and Human Longevity, focused on precision medicine and longevity research.
About Insilico Medicine
Insilico Medicine is a pioneering global biotechnology company dedicated to integrating artificial intelligence and automation technologies to accelerate drug discovery, drive innovation in the life sciences, and extend healthy longevity to people on the planet. The company was listed on the Main Board of the Hong Kong Stock Exchange on December 30, 2025, under the stock code 03696.HK .
By integrating AI and automation technologies and deep in-house drug discovery capabilities, Insilico is delivering innovative drug solutions for unmet needs including fibrosis, oncology, immunology, pain, and obesity and metabolic disorders. Additionally, Insilico extends the reach of Pharma.AI across diverse industries, such as advanced materials, agriculture, nutritional products and veterinary medicine.
For more information, please visit www.insilico.com