Agentic AI Turns Transit Data Into More Targeted Decisions

AI summary of the source article

Researchers in Salvador, Brazil, developed an agentic artificial intelligence tool named SUNTInsight to bridge the gap between large transit datasets and actionable decisions. The system allows transit managers to use natural language to generate SQL queries, retrieve operational data, and view visualizations alongside the underlying queries. In tests on a dataset covering roughly 700,000 passengers and 2,000 vehicles, the system revealed localized congestion on specific route segments that broad network averages obscured. This insight pointed toward targeted interventions, such as short turns and segment-specific headway control, while maintaining human oversight and isolating generated code for security.

Why it matters

The system illustrates how agentic AI can translate complex public transit data into targeted operational decisions while maintaining human oversight.

Key facts

• SUNTInsight translates natural language prompts into SQL queries and visualizations to analyze public transit datasets.
• The system was tested on Salvador, Brazil transit data covering roughly 700,000 passengers, 2,000 vehicles, and nearly 400 lines.
• Targeted measures identified by the tool included short turns and segment-specific headway controls instead of adding buses across an entire route.