01The problem
Equipment records scattered across years of files
In industrial operations, information about equipment piles up over years in many forms: spreadsheets from different sites, PDF service reports, exports from older systems.
The same piece of equipment can show up in many of those files, described a little differently each time. Before anyone can answer simple questions like what do we have, where is it, and when was it last serviced, someone has to reconcile all of it by hand. That work is slow, and mistakes are easy to miss.
02What I built
A platform that does the reconciling
An Azure-based platform that accepts uploaded documents, pulls out the equipment information, and decides for each item whether it is already on file or new.
Matched items are linked to the existing equipment record. New equipment gets a new record. Throughout, service history is preserved rather than overwritten, so each record shows the full life of the asset, not just the most recent file.
- 01
Reads mixed formats
Ingests PDF, Excel, and CSV files, the formats equipment information actually arrives in.
- 02
Creates and matches records
Recognizes equipment that is already on file and creates records for equipment that isn’t.
- 03
Keeps the history
Service history is carried forward with each asset instead of being replaced by the latest upload.
03How it works
From upload to asset register
A simplified view of the path each file takes through the platform.
- Step 1
Upload
PDF, Excel, and CSV files are uploaded to the platform, which runs on Microsoft Azure.
- Step 2
Extract
Equipment details and service information are read out of each document, whatever its format.
- Step 3
Match
For every item, the platform decides: is this equipment already on file, or is it new?
Already on fileLinked to the existing equipment record.New equipmentA new equipment record is created. - Step 4
Asset register
One record per asset, with its service history preserved across every file it appeared in.
04Proving it
A controlled end-to-end test
To validate the whole pipeline, I ran a controlled proof-of-concept test that started from an empty database.
- File uploads processed
- 35
- Assets in the finished database
- 1,966
- Independent audit checks passed
- 27
The 35 uploads populated the empty database with 1,966 assets, and 27 independent audit checks were run against the result and passed. Testing against independent checks, rather than spot-checking a few records by eye, is how you build confidence that a system handling thousands of records is doing the right thing.
What this is, and isn’t
This was a controlled proof-of-concept test, designed to show that the end-to-end process works. It is not a claim about production volumes, speed, or results at any particular company.
05For your business
Where the same approach applies
If your equipment, inventory, or customer records are spread across files and systems, the same idea applies: bring the documents in, let software do the reconciling, and keep a history you can trust.
The same pattern fits maintenance records, parts catalogs, site inventories, or any situation where the same things are described in many places and you need one dependable version.