Written by Navid Faghih, operations manager and co-founder at Canzone Construction
It was the final day of tender on a major laboratory tenant improvement project at an institutional site, the kind of bid where the architectural and engineering packages run so dense that every page assumes you’ve already absorbed the fifty before it. Hours before the tender closed, our team caught something buried in the gray area between two specialized trades. A significant piece of the project. Neither trade had priced it. It had simply fallen through.
Veteran eyes on the documents caught it, nothing more sophisticated than that. The relief lasted about as long as it took me to realize how close we’d come to missing it entirely.
A missed scope on a hard institutional bid carries real weight. Had that gap gone unnoticed, we would have been legally bound to deliver the project at our bid price, forcing us to absorb a margin-erasing loss or go back to a major client asking for more money before a single wall went up. Either path damages something hard to rebuild. Our entire margin and that client relationship came down to whether one person happened to notice a single line item at the eleventh hour.
Once the bid was submitted, the adrenaline wore off and the dread set in. I kept circling the same thought: if we’d almost missed this one, what else had slipped past us and gone uncaught? As we took on more complex institutional work, I had to face the fact that manual review under tight deadlines was a structural weakness, one that would only get more exposed as we grew.
Praising the team for the catch and moving on would have been simple. We stopped instead, and audited our past projects and bids to see whether this was an isolated scare or a pattern. It was a pattern. That finding pushed us toward building a different kind of solution, one that didn’t depend on nobody having a bad day.
That near-miss led us to build our own in-house AI document extraction tool. It ingests complex architectural blueprints and cross-references drawing notes against trade requirements, flagging potential scope gaps before a human review even begins. Final decisions still rest with our estimators; the tool is an assisting layer, a safety net we didn’t have on that tender day. It’s freed our team to spend their time strategizing on projects rather than hunting for what might be hiding in the gray areas.
That day taught me to protect expertise rather than spend it on manual document processing. The real shift was building systems smart enough to let our team’s judgment go toward decisions that actually need it. A single scope gap left us with that lesson, and our process today looks nothing like it did before that day.
