Product
·
April 9, 2026
·
Watch · 34 min

Michael Bachelor walks through a new smart-city app on EdgeLake and MCP

TSC session. Michael Bachelor demos his next smart-city app on AnyLog Edge with EdgeLake and MCP-driven natural-language queries against live municipal data, no cloud round-trip.

At Sabetha, Kansas (the same utility deployment we've covered before), Michael Bachelor presented the next iteration of the city's smart-utility stack: a working CMMS-style maintenance app that replaces a 15-year-old Visual Basic tool the operators had been propping up since long after support ended.

The application was built end-to-end on Base44 with Claude as the architect. Work orders, asset registries, and a starter alerting view all run against the same EdgeLake fabric that already collects power, water, and wastewater telemetry from the three Sabetha plants. Bachelor's point in the session: the app itself is the thin layer. What makes it stick is the foundation underneath.

That foundation is three Lenovo SE30 industrial PCs at the plants, each running Linux and containerized AnyLog operator nodes alongside Dynics ICS 360 Fusion. Fusion pulls data out of the Allen-Bradley ControlLogix PLCs and over Modbus / DNP from the protective relays and synchrophaser meters. A master node and an EdgeLake query node sit in the cloud purely as hosts, with no cloud services in the data path. Data stays at each plant. Queries traverse the metadata layer.

The result is a single data-lake view with an asset-framework metadata ontology (not ISA-95) that's intentionally lightweight: five nodes today, with the same architecture extending to five thousand or five hundred thousand. EdgeLake speaks OPC UA natively for the meters that support it. Bachelor closed by demonstrating MCP-driven natural-language queries against the live municipal data; the same pattern lets Claude act as a CMMS co-pilot inside the maintenance app itself.

Watch the full TSC session on YouTube →