From a full day’s work to just a few clicks: How PG automates the preparation of its geotechnical reports
Construction & Expertise PG turns a manual, repetitive writing process into an AI-powered report generator.
Construction & Expertise PG, a Quebec-based civil and geotechnical engineering firm founded in 2006, delivers complex projects—from Nunavik to major infrastructure projects—where every deliverable counts toward the company’s reputation. At the heart of every project is a geotechnical report ranging from 40 to 400 pages that compiles laboratory tests, field analyses, and engineering recommendations. This report is PG’s signature. And it was written in a very manual way, requiring an engineer to spend a full day on each project. Mondrian designed and deployed “Mr. Sol,” an automated generator that compiles field and lab data into a structured Word report in just a few minutes. Built iteratively—without waiting for the perfect solution—we delivered value quickly, then automated sections over time, until we reached 45 sections in production today. At a rate of 50 reports per year, that's the equivalent of more than two months of engineering time saved each year.
A Quebec-based leader that relies on AI to scale up without compromising its rigor.
Civil Engineering, Geotechnical Engineering, and Northern Construction
51–200 employees
Canada
Founded in 2006 by Pierre Gouslisty, Construction & Expertise PG has established itself as a Quebec leader in geotechnical engineering, northern construction, and the rehabilitation of concrete structures. Operating on demanding job sites—from Nunavik to major infrastructure projects—PG combines the expertise of its engineers and technologists with an in-house geotechnical laboratory. Faced with sustained growth and a growing project pipeline, management has chosen to invest in AI—not to replace its engineers, but to free up their time for high-value-added tasks.
A whole day's worth of writing to redo, project by project.
Every geotechnical project culminates in a final report ranging from 40 to 400 pages: a comprehensive document that compiles laboratory test tables, stratigraphy, foundation recommendations, environmental conclusions, and appendices. This report used to be written entirely by hand, involving copying and pasting data from Access, Excel, and PDF files into Word—a task that took an engineer or technician a full day per project.
A full day of writing per report means an engineer's time locked into low-value work, in an industry where every billable hour counts. With projects involving anywhere from 2 to 5,000 boreholes and data scattered across multiple systems, copy-paste errors accounted for 90% of post-delivery corrections — driving back-and-forth with clients and chipping away at the reputation for rigor PG had built over 20 years.
The report’s data comes from a variety of sources: an Access database via Sobek, laboratory Excel files, PDFs of drilling logs, and GPS data. The project folder structure varied from one technician to another, and certain sections of the report required an engineer’s judgment rather than simple data extraction. Standardizing, connecting, and orchestrating all of this within a robust pipeline required specific expertise in AI and data engineering that PG did not have in-house.
Mr. Sol: A complete report in just a few minutes, all from a single interface.
Mondrian designed and deployed “Mr. Sol”: a Power Apps application accessible with a single click for PG engineers. The user selects a project, enters a few parameters (project type, presence of an environmental study, foundation type), and the solution automatically retrieves the data from the local PG server, copies it to the Azure cloud, and then generates a structured Word report in just a few minutes—complete with numbered appendices—ready to be reviewed and delivered to the client. The 44 automated sections cover laboratory tests (particle size analysis, Atterberg limits, shear test, consolidation, compression), stratigraphy, foundation recommendations, and conclusions.
The project ran in agile mode, with weekly stand-ups and biweekly sprints that actively involved PG's geotechnical engineers. Their domain expertise — knowing how to interpret a standard penetration index, telling a driven pile apart from a bored pile, or recognizing when a section on bedrock is needed — was the raw material behind every automated section. Mondrian translated those business rules into code; PG validated them project by project against real data.
| Component | Role in the solution |
|---|---|
| Power Apps + Azure Functions | User interface and pipeline orchestration |
| Azure Data Factory | Automated data transfer from local server to cloud |
| Azure Container App + AI Foundry | Report generation using language models (LLMs) |
| MySQL + Azure Cloud Storage | Centralized project data storage |
Fewer errors, less friction, more time for what actually matters
| Indicator | Result | Business impact |
|---|---|---|
| First-draft writing time | 1 day → a few minutes | Engineers freed up for higher-value work |
| Report errors | ~90 % reduction | Fewer client back-and-forths, stronger reputation |
| Automated sections | 44 of 54 in production | A nearly complete report generated with the first click |
| Adoption | Live in production | Zero adoption friction, familiar Power Apps interface |
Facing a similar challenge in your industry?
A 30-minute call is enough to find out where AI can create value for you.
Planifier un appel