What does it take to make AI reliable enough to design a bridge? Peggy Smedley sits down with Francois Valois, senior vice president of Bentley Open Applications at Bentley Systems, to unpack the MCP (model context protocol — an open, agnostic standard that connects AI agents to deterministic engineering tools like STAAD and MicroStation ensuring trust, validation, and accountability in engineering workflows.
They explore why generative AI alone needs to be more trustworthy for infrastructure design, how MCP bridges the gap between non-deterministic LLMs (large language models) and precise calculation engines, and what Bentley's open-source strategy means for the future of connected-infrastructure data.
Topics covered:
What MCP is and why Anthropic invented it
Why "90% accurate" isn't acceptable in engineering
How STAAD and MicroStation are now AI-accessible via MCP
The role of structured data (and dark data) in AI workflows
How to start experimenting with AI agents in your firm
The future of the engineer in an AI-augmented world
infrastructure.ai | connectedworld.com | thepeggysmedleyshow.com
https://www.bentley.com/
Freight rail moves 40% of long-distance goods in the U.S., but the industry's biggest challenge isn't infrastructure, it's intelligence. In this episode of The Peggy Smedley Show, host Peggy Smedley sits down with V. Krishnan, industry advisor for Microsoft manufacturing & mobility, to explore how AI agents and realtime decision tools are reshaping the future of rail.
They discuss:
Why railroads are drowning in data but struggling to act on it
How CSX used AI to transform customer experience
The Rumo case study: cutting lookup times from 4 minutes to 3 seconds
Fuel optimization and how AI can cut costs 3–7% without new infrastructure
Weather analytics, workforce challenges, and the retiring knowledge crisis
Why the next 10% of performance won't come from more track — it'll come from smarter decisions
Whether you're in logistics, supply chain, or enterprise tech, this conversation is a must-listen.
Join the Fin and AWS team on July 9 to see how you can scale customer support with AI. Save your spot.
AI is transforming society as profoundly as electricity did — but who's footing the bill? In this episode, Peggy Smedley sits down with Bryan Reimer, Research Scientist at MIT's Center for Transportation and Logistics, to tackle the hard questions the tech industry isn't asking loudly enough.
They discuss:
Who pays for AI's physical, social, and institutional footprint?
Should AI be taxed like labor, capital — or something entirely new?
Who's liable when AI systems fail in critical infrastructure?
How do we protect workers as automation reshapes the labor market?
What does "safe AI" actually mean in practice?
Why automated vehicles are the real-world proving ground for AI governance
Bryan also shares insights from his book How to Make AI Useful and why the goal shouldn't be replacing people — but putting people into better loops.
Follow Peggy on X @ConnectedWorld and LinkedIn
Looking to sponsor an upcoming segment of The Peggy Smedley Show? Check out the exciting topics planned for the coming months and send your ideas to [email protected].