AI & ML · Jul 8
Citrine launched Catalyst and Apex to move materials teams from goals to experiments, citing 550 models and 70,000 monthly suggestions
The useful lesson is not bigger AI theater, it is workflow trust for scientists who actually have to run the next experiment.
- Evaluate materials AI by the decision loop it creates, not by model branding alone.
- Watch adoption metrics like accepted recommendations and lab feedback, not just suggestion volume.
- Build domain AI around goals, explanations, model quality, and experiment feedback from day one.