The Google team is making significant strides in constructing an autonomous wet lab, with the ambitious goal of automating the entire workflow of molecular design, experimental setup, feedback analysis, and solution iteration through AI-driven agents. This initiative aims to establish a recursive self-improvement system fueled by real-time biological feedback. To date, the team has achieved a notable milestone by successfully identifying protein molecules that exhibit strong binding affinity to targets linked to solid tumors. However, the integration of three core technologies—robotic operations, molecular screening, and protein design—continues to pose challenges.

To overcome these hurdles, the team has developed the VERID system, leveraging an open-source framework to create a dynamic digital twin model of the lab workstation. This innovative approach enables AI agents to conduct simulation validations prior to actual experiments and maintain detailed operation logs for efficient troubleshooting.

In parallel, DeepMind is advancing its research on the enzyme design pipeline, AlphaProtein Novo, and the antimicrobial peptide screening platform, Antigravity. Additionally, DeepMind has committed $300 million in collaborative efforts to develop foundational biological data and measurement technologies.

Currently, the pace of AI-generated molecular design far outstrips the capacity for experimental follow-up. The reduced complexity of instrument connectivity, thanks to open-source frameworks, is spurring the industry to expedite the exploration of deep AI applications in scientific research experiments. Notably, companies like OpenAI and Anthropic are also venturing into this domain, each bringing unique technological approaches and commercial strategies to the table.

The establishment of autonomous wet labs will empower AI companies to directly validate model effectiveness, meticulously record experimental processes, and accumulate valuable trial-and-error data that is often elusive in publicly available sources. This has the potential to substantially enhance R&D efficiency during the early stages of drug discovery. Nevertheless, a fully autonomous experimental process has yet to be realized, with critical decision-making steps still necessitating human intervention. The effectiveness of accelerating the entire pharmaceutical R&D chain through these advancements remains to be thoroughly validated.