San Jose and Ladris deploy AI evacuation planning software

San Jose’s contract with Ladris is a good example of AI quietly moving from pilots into critical public infrastructure. Evacuation planning is a high-stakes optimization problem: you’re juggling dynamic data on population, traffic, weather and physical constraints where minutes translate into lives. Embedding an AI system into that workflow creates a real-world testbed for decision-support agents that operate under uncertainty, time pressure and noisy data.([sanjosespotlight.com](https://sanjosespotlight.com/es/san-jose-adds-ai-tools-to-its-disaster-tactics/)) That’s much closer to the environments AGI proponents ultimately care about than benchmark leaderboards.

The strategic angle is that small, specialist vendors like Ladris are becoming the connective tissue between cutting-edge modeling and conservative public-sector buyers. They wrap AI into narrow, auditable applications that city officials can understand and procure. As more cities standardize on similar tools for floods, wildfires or mass events, these systems will accumulate rich, longitudinal datasets on human movement and behavior in crises. That data is gold for improving future planning models—and potentially for training more general decision-making agents. It also raises hard questions about transparency and accountability when an algorithm’s recommendation shapes who gets out of harm’s way first.