en.Wedoany.com Reported - Chinese AI pharmaceutical company Insilico Medicine officially released its next-generation intelligent laboratory operating system, LabClaw, on May 7, 2026. The core architecture of LabClaw consists of five AI agents with clearly defined roles. The entire drug R&D process is encapsulated into 20 combinable experimental step modules, covering areas such as AI computing services, sample and material preparation, automated experiment execution, data analysis and report generation, and notification and collaboration. At critical operational nodes, the system retains a "human-in-the-loop" manual confirmation mechanism.

The "Claw" in LabClaw's name originates from the open-source AI agent tool OpenClaw. The system uses Pharma.AI as its core decision-making engine, synergistically completing the entire drug R&D chain through five agents: the Experiment Coordinator, Experiment Orchestrator, Science Analyst, QC Inspector, and Data Specialist. According to a press release published by EurekAlert on May 6, LabClaw is equipped with 28 specialized agent skill modules, constructing an end-to-end intelligent closed loop covering the entire process of "target discovery—compound screening—automated experiment execution—data analysis—report generation," seamlessly integrating both dry lab and wet lab phases.

Dr. Alex Zhavoronkov, Founder and CEO of Insilico Medicine, stated in the announcement that LabClaw embodies the company's over a decade of accumulation in the field of end-to-end AI drug discovery. He pointed out that this system is not intended to replace scientists, but to change the way scientists collaborate and their work rhythm.

A white paper released concurrently with LabClaw disclosed the system's operational data in real drug R&D projects. After adopting the LabClaw system, the efficiency of acquiring data from acute toxicity animal models increased by 5.7 times, shortening the traditional acute toxicity experiment process, which requires 7 to 10 days of manual operation, to approximately 1.5 days in concurrent mode. In stable cell line generation scenarios, LabClaw compressed the delivery time for a single process from 14 to 21 days down to 7 days, achieving a 100% efficiency increase. The efficiency of excipient compatibility experiments improved by about 10 times, meaning approximately a 90% reduction in experimental cycle time under the same project schedule.

The collaboration among the five agents follows a clear division of labor logic. The Experiment Coordinator is responsible for synchronizing the parallel actions and decisions of all agents; the Experiment Orchestrator decomposes R&D goals into executable automated experimental workflows; the Science Analyst interprets experimental data in real-time and dynamically adjusts subsequent plans based on results; the QC Inspector continuously monitors data quality and compliance; and the Data Specialist ensures data accuracy and consistency among different agents. Manual confirmation is retained at key nodes, ensuring that the final decision-making power rests with scientists—the entire system is a closed-loop collaboration engine architected between AI-driven automation and human professional judgment.

Insilico Medicine has been continuously investing in automated laboratory construction since 2021. Its fully automated robotic laboratory located in Shanghai's Pudong area commenced operations in September 2024. The laboratory consists of six functional islands, including cell culture, gene sequencing, and high-throughput screening. Material transfer between islands is achieved via automated guided vehicles equipped with robotic arms, featuring nearly 150 automated devices from over 100 suppliers, capable of uninterrupted 24/7 operation. Insilico Medicine's technical team developed skill packages for each functional island, transforming specific experimental operations into standardized, reusable modules, providing the physical execution foundation for LabClaw's multi-agent synergy. In March 2025, the company also took the lead in deploying a humanoid robot "Supervisor" in the laboratory for data collection and generation, conducting systematic training for embodied intelligence.

Observing LabClaw within the broader context of Insilico Medicine's technological evolution, this release serves as a strategic anchor for the company. In February 2026, Insilico Medicine and the U.S.-based Eli Lilly and Company jointly published a paper in ACS Central Science, proposing for the first time the "Prompt-to-Drug" full-stack autonomous AI pharmaceutical vision, covering the entire process from target discovery, molecular generation, and animal validation to CMC (Chemistry, Manufacturing, and Controls) filing. LabClaw is precisely the core infrastructure extending this blueprint from the computer end to the physical laboratory bench—it completes the system-level deployment of the "Prompt-to-Drug" concept's closed loop, where "only a prompt is needed to go from target to clinical candidate compound," within a real laboratory environment.

Founded in 2014, Insilico Medicine was listed on the Main Board of the Hong Kong Stock Exchange on December 30, 2025 (3696.HK). The company's self-developed generative AI platform, Pharma.AI, has supported the generation of over 20 drug assets at the clinical or IND filing stage, with 10 molecules having obtained clinical trial approvals. Its core pipeline asset, ISM001-055, is the world's first drug candidate discovered by AI targeting a novel target to enter Phase II clinical trials. Since 2026, the company has continued to accelerate its BD collaboration layout: in March, it reached a pipeline licensing and AI pharmaceutical collaboration agreement with Eli Lilly, with a total transaction value of up to approximately $2.75 billion, including an upfront payment of $115 million; during the same period, it secured a major BD deal with Daewoong Pharmaceutical valued at $880 million.

The LabClaw white paper also disclosed that the system was built based on over 1,000 benchmark drug R&D tasks and approximately 120 billion tokens of proprietary pharmaceutical industry data. Through multi-task fine-tuning and reinforcement learning, the performance of the Pharma.AI foundation model in pharmaceutical-specific scenarios has been significantly enhanced. Insilico Medicine simultaneously released the MedOS operating room system, covering operating room task scheduling and real-time clinical decision support, forming a multi-scenario AI operating system matrix of "laboratory research—operating room clinical" together with LabClaw and LabOS.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com