Genesis achieved its sub-one-angstrom accuracy not through one algorithmic trick, but by making it a core objective from the start. This obsessive focus on the right metric guided countless small, compounding decisions across data, infrastructure, and modeling.

While AI models are effective for developability properties like stability, they fall short on predicting function. Sanofi's Norbert Furtman notes that generalized affinity prediction is a 'holy grail' problem, and predicting interference with a biological pathway is even harder, as function is not solely explained by structure.

To evolve AI from pattern matching to understanding physics for protein engineering, structural data is insufficient. Models need physical parameters like Gibbs free energy (delta-G), obtainable from affinity measurements, to become truly predictive and transformative for therapeutic development.

Similar to how an LLM uses a 'chain of thought' to reason, Genesis's model 'thinks' by iteratively refining an in-memory representation of a crystal structure. This process is guided by physics-based principles, significantly improving the final prediction's accuracy.

A key strategy for improving results from generative protein models is "inference-time scaling." This involves generating a vast number of potential structures and then using a separate, fine-tuned scoring model to rank them. This search-and-rank process uncovers high-quality solutions the model might otherwise miss.

Despite testing millions of compounds, high-throughput screening methods suffer from enormous false-positive rates. The actual predictive value for a re-synthesized molecule is extremely low, creating an opening for high-fidelity AI models to provide cleaner, more reliable predictions.

Genesis's focus on sub-one-angstrom accuracy came from direct experience. When applying models to active drug discovery programs, it became 'pretty obvious' that the standard two-angstrom benchmark was inadequate. This highlights the gap between academic benchmarks and real-world utility.

Beyond accelerating timelines, AI's real value lies in its ability to design molecules for targets previously considered 'hard-to-drug.' These models operate on different principles than traditional lab methods and are indifferent to historical challenges, opening up entirely new therapeutic possibilities.

The immediate goal for AI in drug design is finding initial "hits" for difficult targets. The true endgame, however, is to train models on manufacturability data—like solubility and stability—so they can generate molecules that are already optimized, drastically compressing the development timeline.