Abstract Manager

The modern traffic agency is data-rich but insight-poor. While the industry has successfully deployed ITS technology at signalized intersections including high-resolution data logging controllers and non-intrusive multi-modal detection, operations teams remain overwhelmed by the manual labor required to interpret this data. Engineers are often forced into a reactive cycle, chasing the loudest citizen complaints rather than proactively optimizing the traffic network. 

This session explores the emergence of Agentic AI—a domain-specialized technology that moves beyond static dashboards to function as a "Virtual Traffic Operations Assistant."

Unlike generic large language models or autonomous intersection controllers, these virtual agents are purpose-built for the specific workflows of traffic engineering. We will discuss how this technology acts as a force multiplier by automating the three critical stages of traffic management: Analysis, Diagnosis, and Recommendation.

Attendees will learn how virtual agents:

• Understand Natural Language: Allowing engineers to ask complex questions in plain English, such as "Analyze split failures at King and University" or "What are the top five pedestrian volume intersections in my network?".

• Triage Citizen Complaints: Instantly validate and find data to support citizen complaints, even when unfiltered text is copied and pasted directly into the system.

• Provide Root Cause Analysis: Move beyond simple charts to provide deep analysis, root cause assessments, and actionable recommendations for complex issues like signalized intersection performance.

• Accelerate Onboarding: Act as a "mentor" for new staff by providing guided decision support, engineering rationale, and immediate access to contextual data.

Finally, we will demonstrate how Agentic AI drafts precise, explainable configuration changes—from timing adjustments to phase rebalancing—that are presented to human engineers for final validation. By keeping the "human in the loop," this technology bridges the skills gap, standardizing decision-making and allowing agencies to scale their expertise without increasing headcount.

By accelerating analysis and providing guided decision support, a Virtual Traffic Operations Assistant helps agencies reduce operational overhead, ultimately enabling them to build safer, more efficient, and more responsive transportation networks.

Examples of how traffic agencies across North America are using this AI tool will be included to support attendees understanding of the potential of this type of technology.