Back to case studies

Retail Tech

Gas Station Inventory Management SaaS with AI OCR

Centralized inventory management for 40+ gas stations - AI OCR invoice processing, POS auto-sync, and 20K+ transactions processed.

We built a centralized inventory management SaaS for a multi-location gas station operator managing 40+ stations across US cities. Each location ran Gilbarco Passport POS independently with no unified tracking. The platform centralizes fuel, convenience store, and supply inventory with auto-sync from POS systems, AI-powered OCR for invoice scanning, Bluetooth scanner support, centralized price management, and a single dashboard across all locations.

Start a similar projectUpdated Mar 2026

Client

Regional US Gas Station Operator

Industry

Retail Tech

Timeline

16 weeks

Team Size

5 engineers + 1 PM

Impact

Measurable results

20K+

Transactions processed

40+

Stations managed

AI-automated

Invoice processing

Centralized

Price management

Transactions processed: Transactions processed across all station locations through the centralized platform.

Stations managed: Gas stations unified under a single inventory management dashboard with real-time sync.

Invoice processing: OCR-powered invoice scanning replaces manual data entry, extracting product and vendor details automatically.

Price management: Single price update propagates across all POS terminals with full audit trail.

The Challenge

What we were up against

Each of the 40+ gas stations operated independently - inventory tracked in spreadsheets and email chains, sales and purchase data locked inside individual Gilbarco Passport POS terminals with no way to aggregate across locations.

Invoice processing was entirely manual. Staff at each location typed vendor invoice details by hand into spreadsheets, introducing errors in product names, quantities, and pricing that compounded across dozens of stations.

Price changes had to be updated one-by-one at each POS terminal. A single fuel price adjustment meant contacting every station individually, with no centralized control and no audit trail of when changes took effect.

Non-technical station staff struggled with existing tools. Any solution needed to be simpler than the spreadsheets it replaced, with minimal training and support for Bluetooth barcode scanners already in use at the stations.

Frequent stockouts and overstocking across locations because there was no single source of truth for inventory levels, no automated reorder alerts, and no visibility into which stations were over-supplied while others ran short.

What We Built

Our approach

1
Step 1

Built centralized inventory tracking across fuel,...

Built centralized inventory tracking across fuel, convenience store, and supply items with automatic sync from Gilbarco Passport POS systems - eliminating manual data entry and giving operators a single dashboard view across all 40+ stations.

2
Step 2

Developed AI-powered OCR invoice scanning that extracts...

Developed AI-powered OCR invoice scanning that extracts product names, vendor details, quantities, and pricing from scanned or photographed invoices, auto-filling forms and categorizing products without manual data entry.

3
Step 3

Implemented centralized price management where a single...

Implemented centralized price management where a single update propagates across all POS terminals, with audit trails showing exactly when and where each price change took effect.

4
Step 4

Designed the interface for non-technical staff with...

Designed the interface for non-technical staff with Bluetooth barcode scanner support, large touch targets, and a workflow that is faster than the spreadsheets it replaced - requiring minimal training for station employees.

Tech Stack

ReactNode.jsPostgreSQLAI OCR EngineAWS

Related Work

Frequently asked questions about this project

AI OCR (optical character recognition) uses computer vision models trained on invoice layouts to extract structured data - vendor names, product descriptions, quantities, unit prices, and totals - from scanned or photographed documents. The system learns common invoice formats over time, improving accuracy and reducing the need for manual correction as more invoices are processed.

Next Step

Ready to build something similar?

One call with a founder. No sales team, no follow-up sequence. If we can't help, we'll say so.