Every stage keeps its own copy
From site scan to project closeout, each stage of a project produced new files without replacing the old ones, so every build left several copies of the same drawings behind.
FormTek Construction Inc. is a general contracting and construction management firm based in Canada, with a U.S. office and roughly 45 construction team members. The firm carries a project from site scanning and modelling all the way through issued drawings, as-built documentation, and project closeout.
This story focuses on experiences shared by Arsalan Khan, Director of Construction. His team’s files move between project coordinators, site teams, engineers, clients and subcontractors, and the more often they move, the harder it becomes to know which copy is the one the project actually needs.
One project, many versions of the same file, less certainty
On a construction project, drawings are never really finished - they are superseded. A site scan becomes a point cloud, the point cloud becomes a Revit model, the model is issued as CAD sheets, the sheets are revised, and the final set is archived as as-builts. Each stage produces its own files, and each stage tends to keep the one before it.
FormTek’s workflows produce an unusually wide set of file types: point clouds, Revit models, AutoCAD DWG and DXF, PDF, image sets, orthophotos, survey files and as-built documentation. These files are shared across teams and stored in project folders, backup drives, and shared locations.
Every project moves through site scans, point clouds, Revit models, CAD revisions, etc., and files get passed between our coordinators, site teams, engineers and subcontractors.
The result is a familiar problem in construction: several copies of a drawing set, no reliable way to tell which one is current, and a manual clean-up that never quite happens. At roughly 5 TB of active and archived project files across four drives, checking every file manually could take several days. Yet removing the wrong version of a deliverable was catastrophic.
Over the years, the same drawings ended up saved in multiple project folders, backup drives and shared locations, and it was hard to tell which version was the one we actually needed.
The team faced four connected problems:
From site scan to project closeout, each stage of a project produced new files without replacing the old ones, so every build left several copies of the same drawings behind.
Project folders and drawing sets were copied between shared directories, archive locations, and backup drives.
Comparing drawings and folders one at a time took several days - and at multi-terabyte scale, reviewing files individually was not a realistic option.
Names alone did not reliably show whether files were identical or which project copy the team should retain.
After testing generic cleaner tools that lacked folder-level granularity or failed when scanning complex multi-terabyte directory structures, FormTek deployed Cisdem Duplicate Finder across its engineering and project coordination workstations. Cisdem offers a unified, multi-algorithm scanning engine engineered specifically for rapid cryptographic hashing, large-scale storage handling, and cross-drive directory comparison.
We needed a tool built specifically for duplicates, not a feature buried inside a larger app. The customizable selection rules mean we don’t have to click through each file one by one to decide what to remove.
Arsalan’s team pointed Cisdem Duplicate Finder at four drives holding 5 TB of project data. The full scan finished in about 1 hour and 27 minutes. Cisdem’s cryptographic MD5/SHA256 checksum engine detected identical content even when files had been renamed across departments.
The scan completed in 1 hour 27 minutes across our 5TB dataset without a hitch. It gave us an instant, categorized breakdown of exact duplicates and let us purge over half a terabyte of waste with complete certainty.
The biggest duplicates on FormTek’s drives weren’t single files - they were entire project directories that had been copied during archiving or synced between drives. Hunting those down file by file misses the bigger picture.
The duplicate folder and similar image views were the biggest time saver. When an entire project directory had been copied during archiving or synced between drives, we could remove it in one step instead of file by file. Side-by-side similar image comparison also made it clear which drawing set we were keeping.
Cisdem’s Duplicate Folder Finder shows a copied project folder as a single result, so it can be reviewed and removed in one step. For near-identical drawing exports and site photo series, similar image comparison puts each group side by side, so the team can see exactly which set it is keeping.
For projects, speed only matters if the right files survive. Customizable selection rules decide which copy is kept by default, and AI Photo Quality Scoring points to the highest-quality image in each group of similar photos - so nobody has to check them one by one.
Keeping the final confirmation with the user was especially useful for project records that needed to remain available.
With project deliverables, deleting the wrong file isn’t an option. Cisdem Duplicate Finder offers an AI Photo Quality Scoring feature, which helps us choose the photos with the highest quality, so we don’t have to check them one by one. Plus, the app asks for confirmation before the real delete, so our team can use it with confidence.
Less time managing files. More confidence in what’s kept.
The impact shows up in storage reclaimed and time saved - but also in what the team no longer worries about: deleting a current drawing by accident, or spending days comparing project folders by hand.
FormTek has integrated Cisdem Duplicate Finder directly into its standard monthly closeout protocol. Before archiving any completed project milestone or syncing large datasets across field offices, coordinators run a routine scan to ensure primary drives stay clean and free of redundant versions.
This is by far the most reliable duplicate file finder for engineering and technical data. It’s fast, mathematically precise, and powerful enough to clean anything from a few gigabytes to multiple terabytes.