Home 9 AI 9 Teaching MicroStation to Understand Engineers

Teaching MicroStation to Understand Engineers

by | Sep 28, 2026

An archaeologist’s path to AI reveals new ways to automate design tasks and simplify engineering software.
A young Louis-Martin, inspired by his visit to a replica of King Tutankhamun’s tomb in 1986 (source: Bentley).

 

Louis-Martin Losier’s journey from archaeology to engineering software reflects an ambition: making complex information easier to understand and use. Now a senior product manager at Bentley Systems, he is helping engineers control MicroStation through natural language, reducing the need to memorize commands or write automation code, tells Bentley.

His interest began with discovering an arrowhead as a child in Canada. At Laval University, he combined archaeology with geomatics, reconstructing a Syrian excavation site from records. Unlike conventional drawings, his 3D model let researchers examine layers and search by period or artifact.

Later, in Brazil, Losier applied digital modeling to civil engineering. He developed a monitoring platform for a landfill, combining ground information and sensor readings to flag safety risks. That approach helped inspire Geovoxel, the startup he cofounded in 2011. Its systems supported infrastructure monitoring and landslide warnings, giving authorities time to respond when rainfall crossed dangerous thresholds.

After joining Bentley in 2018, Losier managed geospatial products before moving to MicroStation. Workshops with users in 2024 identified repetitive work and difficult onboarding as major productivity barriers.

MicroStation 2025’s Python Assistant addressed automation by generating scripts from ordinary language. Bentley Copilot helps newcomers translate familiar workflows into Bentley software and consult company project manuals. Losier says onboarding can shrink from days to hours.

His latest work involves an experimental Model Context Protocol server that connects assistants such as Claude or GitHub Copilot to MicroStation. Agents can execute requested operations inside the software.

Demonstrations included reconstructing the ancient Antikythera mechanism with less than a day of hands-on work and recreating the Quebec Bridge’s geometry in days. A separate connection to STAAD enabled structural analysis.

The central challenge remains aligning probabilistic AI with engineering’s exacting requirements. Bentley’s profile presents these tools as a way to make sophisticated software more accessible while extending what engineers can accomplish.