MIT News reported on April 17, 2026, that OpenProtein.AI has launched an open-source, no-code artificial intelligence platform. Powered by robust foundation models like PoET, the platform helps biologists worldwide design proteins and predict their structure and function without needing machine learning expertise. This innovative tool aims to lower the barrier to AI adoption and accelerate the protein engineering process in the pharmaceutical and biotech sectors.
Core Features and Technical Highlights
The platform offers an intuitive web interface where researchers can easily upload experimental data and use machine learning for protein design, structure prediction, and functional analysis. It integrates various open-source models, including its flagship protein language model, PoET (Protein Evolutionary Transformer), which is trained on protein family data to generate related sequences and generalize evolutionary constraints. The PoET-2 version, released last year, outperforms models with more parameters while requiring minimal computational resources and experimental data. PoET-2 also supports zero-shot variant effect prediction and sequence optimization.
Founder Background and Open Mission
OpenProtein.AI was co-founded by Tristan Bepler, who holds a Ph.D. in computational and systems biology from MIT, and former MIT associate professor Tim Lu. The company is dedicated to building an open ecosystem for AI and biology, preventing advanced tools from being confined to a few large institutions. Bepler emphasizes that the platform is designed as an open toolbox, allowing researchers to train models on their own experimental data and optimize protein sequences. We’re trying to make the platform an open toolbox, Bepler stated.
Practical Applications and Industry Collaboration
Pharmaceutical and biotech companies have begun to adopt the platform. For example, Boehringer Ingelheim has been using OpenProtein.AI’s tools since early 2025 and has expanded the collaboration to embed the platform into its antibody discovery workflow for engineering protein therapies for cancer and autoimmune diseases. Scientists in academia can access all features for free, promoting broader scientific research and innovation.
Technical Advantages and Future Potential
Using generative AI, the platform can rapidly generate libraries of protein sequences on a computer and perform predictive validation, significantly shortening laboratory testing cycles. It not only supports sequence-to-structure prediction but can also analyze the dynamic features of proteins. The company plans to expand to non-protein biological modalities in the future, developing domain-specific languages to describe biological systems.
