ReticularAbout UsOur Mission: Use Interpretable AI to Accelerate Biological Discovery
Data scarcity makes steering biological AI systems difficult.
Reticular builds mechanistic understanding to steer these powerful models with precision - even with limited validation data. We've chosen to focus on antibody engineering challenges first, as we see them as a critical bottleneck in therapeutic development pipelines. We help you:
- 1peer inside biological AI to understand its decision-making
- 2steer models reliably even with limited validation data
- 3accelerate development by reducing experimental cost
- 4build safeguards directly into model capabilities
Our approach prioritizes transparency by design, incorporating biophysical principles rather than just statistical patterns. By providing not just what to try next, but why, we help scientists design smarter experiments with greater confidence.
Our Research
Research forms the foundation of our biological AI interpretability efforts. We focus on:
- Translating interpretability techniques from natural language to biological sequence models
- Pioneering semantic biological interpretability techniques using DNA & protein annotation databases
- Developing and empirically validating sample-efficient guidance algorithms for practical protein engineering use cases
Our Team
The Reticular team was formerly publishing cutting-edge research at MIT, with over 10+ years of combined experience in machine learning and biology. Our work has been published in leading venues including ICLR, NeurIPS, and Nature, and is protected by multiple patents.
As scientists ourselves, we're deeply committed to building tools that genuinely enhance the work of other scientists. Reticular was founded by Nithin Parsan and John Yang with the belief that the most powerful biological engineering comes from AI systems that enhance scientific understanding, not replace it.
