- Case Studies
- Deep Industries
Deep Industries uses Source Code Lab’s AI agent to track 35,000+ parts across its oil and gas sites
How an oil and gas services company gave every gas extraction site a live, accurate inventory ledger, without asking a single site manager to learn new software.
Book a strategy call| Spiral wound gasket, 4"Site 7 · +40 | Awaiting approval |
| Ball valve, 2", Class 600Site 7 · +6 | Awaiting approval |
| Drill pipe, 5"Site 3 · -12, yard +3 | Approved |
| Unknown: “HP flange kit”Site 11 · learning from supervisor | New term saved |
Inventory parts tracked across every gas extraction site, in one live ledger.
Site updates logged accurately by the AI agent from the start, and it learns from every one it misses.
35,000+ parts, many sites and no clear picture of what was where.
Deep Industries delivers gas extraction projects across multiple sites, each running on a constant flow of equipment and spare parts. Its inventory spans more than 35,000 different items, from gaskets and valves to drill pipe and compressor components.
Keeping track of all of it had become a serious problem:
- No reliable view by site. The team struggled to know which parts were at which site, and in what quantity.
- Every site speaks its own language. The same part could be called by a nickname, an abbreviation or a size code, depending on who was writing.
- Updates arrived in every format. Site managers and supervisors reported stock in notes, emails, WhatsApp messages and photos of handwritten sheets.
The usual fix is a new system that forces every site to enter data the same way. But asking busy crews on remote sites to change how they work is slow, expensive and rarely sticks.
“We had parts spread across sites and no single answer to a simple question: what do we have, and where is it? Getting every site to fill in the same forms just wasn’t realistic.”
An AI inventory agent that speaks the language of the oilfield.
“Our site managers didn’t have to change a thing. They send updates the way they always have, and the ledger just keeps itself current.”
Instead of building a new data entry system, Source Code Lab built an AI agent that understands the inputs Deep Industries already had. We taught it a detailed ontology of the business: how every part is referred to in the industry’s everyday vernacular, along with the dimensions and specifications of each one.
Takes updates in any format
- Written notes, emails and WhatsApp messages
- Photos of handwritten stock sheets
- Industry nicknames, abbreviations and size codes
- Matches each mention to the exact part and its dimensions
- Records quantities against the right site
Learns from every miss
- Flags any part it can’t confidently identify
- Learns the correct match once it’s confirmed
- Updates its knowledge so the same mistake isn’t repeated
- Gets more accurate with every site update
A person signs off before the ERP changes
- The agent is capable of updating the ERP and inventory on its own
- By design, that final step stays with the overall project supervisor
- Every change is reviewed before it reaches the system of record
The result is an inventory process that fits the way field teams already work, with AI doing the reading and matching, and an experienced supervisor keeping final control.
A live view of every part, at every site.
Deep Industries now knows which parts are at each site and in what quantity, kept current from the same messages site teams were already sending.
| Inventory tracking | Before | With the AI agent |
|---|---|---|
| Parts tracked | Scattered records | 35,000+ in one ledger |
| How sites report | Any format, hard to compile | Any format, read automatically |
| Updates logged accurately | Manual effort | 8 in 10 from the start |
| New software for site teams | Would have been required | None |
One ledger for 35,000+ parts
Stock levels for every site are tracked in one place, so the team can see what’s available before ordering or moving parts.
No new system for site teams
There was no need to build and enforce a data entry system. Managers keep reporting by message, email or photo.
Accuracy that keeps improving
The agent logs 8 in 10 updates accurately from the start and learns from every part it couldn’t identify.
Human control where it matters
The project supervisor approves every ERP and inventory change, keeping a safety check on the system of record.
“Source Code Lab solved our inventory problem without disrupting a single site. The agent understands our parts the way our people talk about them, and our supervisors stay in control of every change.”
See what yours could look like.
Book a 30-minute call and we’ll map how an AI agent could track your inventory across sites, using the messages, emails and photos your teams already send.
Book a strategy call