An environmental consultancy where assembling a site dossier took longer than reviewing it
Built an intake tool that reads every document a client uploads for a site file, recognizes what each one is, orders it by category and date, and assembles a single clean dossier in minutes. A person reviews the result instead of sorting and merging it by hand.
During the busy assessment season, clients drop everything for a site onto a secure portal at once: 20, 30, sometimes 40 or 50 separate documents per file. Lab results, well logs, prior reports, permits, photos, regulatory letters. Before anyone can do the actual review, someone has to pull all of it off the portal, figure out what each document is, classify it into the right category, put it in chronological order, and merge it into one assembled dossier.
On a heavy file, that sorting and merging ran about an hour to an hour and a half. Not the analysis, just the assembly. A simple file with two documents was nothing, but a file with a hundred-plus pages across dozens of attachments was an hour of a senior person reordering and stitching PDFs by hand. Done across a thousand-plus files in a season, it was a large block of skilled time spent on filing, and the firm called it exactly that: heavy, slow, and not very useful work.
We built an intake tool that does the assembly the staff were doing by hand. When a client's documents come in, it reads each one, recognizes the type, sorts them into the firm's defined category order, sequences them by date, and produces a single assembled dossier ready for review. The reviewer opens a finished file and checks it instead of building it.
The recognition is AI-driven with a deterministic fallback behind it, so it still produces a usable, ordered dossier when a scan is messy or a document type is unusual, and never simply hands back an empty result. It works on whatever the client uploaded, regardless of how the file was produced, because most of what arrives is just scanned paper with no clean structure to lean on.
The assembly step that used to take about an hour and a half on a complex file now takes a few minutes, roughly 90 percent off the part of every file that always ate the most time. What is left for a person is the judgment: confirming the tool read things correctly and adjusting the order where it matters, not reordering attachments one at a time.
Across a season of more than a thousand files, that turned a recurring block of senior time spent on filing back into capacity for the assessment work itself. The same team moves through more files without adding people, and the slowest, least rewarding step in the intake is no longer done by hand.
“The bottleneck was never the analysis. It was taking forty documents a client dumped on us, figuring out what each one was, and stitching them into one ordered file by hand. Doing that a thousand times a season was heavy, slow work that nobody should be doing.”
Have a mess like this?
Tell us where time and money are leaking. We’ll map the fastest high-ROI win, usually a Pilot that pays for itself.