Nicholas Francoeur manages infrastructure technology for Compucom, a 5,000-employee company running technology deployments and on-site support across hundreds of client sites nationwide. When off-the-shelf tools couldn't handle complex field operations at the scale he needed, Nicholas did not wait for vendors to catch up.
The disconnect that costs hours every day
Technicians spent entire eight-hour shifts manually logging 50 to 100 hardware assets per site. They photographed equipment, then hand-typed make, model, and serial numbers into inventory systems. The error rate reached 25%—one in four entries had mistakes that surfaced later as mismatched records or scheduling conflicts.
Building weekly schedules for 600 to 700 site projects consumed 20 hours of manual work. Resource Guru offered no integration or bulk upload capabilities, so every site booking required individual entry. When projects paused, and hundreds of sites needed rescheduling, the manual catch-up took 40+ hours.
“Having technicians spend full shifts doing inventories felt, for lack of a better term, dated and old school,” Nicholas says. “You could hear it in their voice when we started the conversations that they dreaded doing it. We needed to find a better way.”
The moment when building became the only option
When Compucom’s scheduling vendor said it didn't support the team's needs, Nicholas looked for another path. Zapier already supported the vendor through its connected apps program, so an API bridge became possible.
Nicholas operates from a simple principle: “When the tools we have don’t do what we need, I build the bridge.” The business needed one source in one place rather than information scattered across systems.
He identified the two highest-drag jobs—asset logging and operational scheduling—and designed bridges that let AI handle messy real-world input while structured tools carried clean data into scheduling systems.
Building bridges between systems
The asset-logging bridge starts with a web form. Technicians submit their name, site ID, and workstation number, then upload photos of equipment. AI by Zapier processes the images and extracted the make, model, and serial numbers. The extracted information is matched against Foresight inventory reference data and stored in Zapier Tables.
An OCR cross-check catches mismatches against existing records, flagging data-quality problems before they propagate. The reference table is specific to each client, so devices that fall outside the known list are flagged for review.
The scheduling bridge reads spreadsheet data and automatically creates bookings through the Resource Guru API. Changes can be verified in Resource Guru in real time when updates are made.
The key insight is separating judgment from transfer. AI reads messy input such as equipment photos and survey data. Deterministic logic handles the structured transfer into scheduling systems.
Steal this pattern: Let AI read the messy real-world input, then let structured tools carry clean data into operational systems.
When accuracy and speed both improve
Site surveys that once consumed a full eight-hour shift now take two to four hours. Asset-entry errors dropped from 25% to roughly 1% after Nicholas added more intelligence and cross-checking to the original proof of concept.
“The first time we saw a successful run was the big lightbulb that this was going to work,” he says. “The proof of concept was becoming a reality, and it felt very rewarding.”
The scheduling wins proved even more dramatic. Weekly schedule builds for programs covering roughly 700 sites dropped from 20 hours to about one hour. When a project pause required rescheduling, work that previously consumed 40 hours now takes about two.
This year, roughly 600 sites have run through the survey workflow. Technicians average about two sites a day, with some reaching a third site in a day.
The pattern that scales beyond one company
The system is live nationwide, used by technicians at hundreds of client sites. Compucom is now working with additional internal teams to package the pattern for more clients and internal support groups.
Nicholas describes the process as building a house: the initial structure exists, and each team helps decide how to furnish it. Teams bring their requirements and use cases, walk through the current tool, and then help shape the next configuration.
“We are in a time and space where we don’t need to wait for a vendor solution,” Nicholas says. “If there is an idea or business requirement that’s not directly available, there’s going to be a way to build it. You just need to figure out how.”
