A standing-room-only session at the APTA 2026 Rail Conference examined how transit agencies can deploy AI to improve efficiency without compromising safety, data integrity, or human judgment. Four organizations laid out practical approaches, from workforce readiness to real-time vehicle monitoring, offering operations leaders a roadmap for adoption that centers on trust and oversight.
AI adoption is a workforce challenge, not just a technology one
David Jackson, managing partner and transportation industry lead at Gartner Consulting, framed AI adoption as both a technology and a people problem. "AI adoption is a human problem," Jackson said, pointing to the need for an AI-ready culture and stronger employee confidence in using the tools. He stressed that agencies must build AI literacy, establish clear policy, and rethink existing processes before the technology can deliver value in daily work.
Kevin Pellegrini, owner of TransitShine Solutions LLC, outlined a spectrum of AI integration. At one end, stand-alone uses like research and document comparison are already accessible. More advanced applications can query large datasets, power chatbots, and help build or test operational systems. Pellegrini cautioned that AI still struggles with reliability and open-ended tasks, especially when given large amounts of context. He said data privacy, cybersecurity, and human-led data quality and governance must anchor any implementation.
Program optimization and safety engineering get a data-driven reset
Jane Huang, industrial engineer, and Kristen McDonald, principal of digital advisory and transformation at Jacobs, contrasted traditional project delivery-often reactive, siloed, and overloaded-with AI-enabled approaches that are proactive, connected, and optimized. Their presentation showed how AI can help agencies reuse knowledge across projects, identify risks earlier, and cut costs by making siloed data visible to the teams that need it.
Amin Kalbasi, principal systems engineer at Parsons Corporation, tackled safety engineering in the era of big data and AI. He explained that AI models can vary, evolve over time, and resist easy explanation. Yet they can also identify safety risks with more than 99 percent accuracy. When asked if the rail industry can learn from safety standards applied to autonomous vehicles, Kalbasi said, "absolutely, yes."
Edge AI turns revenue fleets into smart safety monitors
Tejas Agarwal, CEO of Scout Robotics, described how Edge AI can process data directly on transit vehicles, reducing the need for constant connectivity. Small devices mounted on trains collect information and generate real-time insights. Potential applications include monitoring tie conditions, identifying missing components, thermal monitoring, detecting and counting assets, spotting obstacles, and assessing clearance.
During the panel discussion, speakers returned to common themes of data standards, storage, transfer, risk, and governance. They encouraged agencies to ask whether new AI-enabled systems are making operations measurably safer and how to sustain momentum beyond pilot projects.
Why this matters for operations
The session made one thing clear: AI in rail operations is not a future concept-it's arriving through incremental, practical steps. For operations managers, the immediate priority is building the data governance and workforce skills that make these tools trustworthy. Without that foundation, even accurate models will stall. Structured training, like an AI Learning Path for Operations Managers, can help teams move from isolated pilots to organization-wide capability. The broader field of AI for Operations is evolving quickly, and the speakers' emphasis on human oversight, data integrity, and safety validation offers a sober, actionable benchmark for any operations group considering its next move.
Your membership also unlocks:
