Records with no signal
Readings and their photos are stored on the phone instantly. When the connection returns, ScanGas syncs by itself in the background.
Every meter, read instantly.
ScanGas turns meter reading into a sequence: snap, automatic reading, next meter. An AI model trained on real gas meters reads the index and the serial number in under a second.
- < 1 sper AI reading
- Offlinerecords with no signal
- GPSto locate every meter
Animated app demo: the reader photographs the meter, the AI reads the index and the serial number, and the app moves on to the next meter by itself.
Built for gas sub-distributors, utility cooperatives and distribution networks
01The problem
Manual reading costs you twice: in the field and at billing.
Paper sheets, stray photos on a phone and data retyped back at the office. Every step adds time and another chance for error.
- 01
Transcription errors
One misread digit means a wrong bill, a complaint and another visit to the address.
- 02
Readings without proof
When a customer disputes a bill, there is no photo or record of what the meter actually showed.
- 03
Anomalies caught too late
Leaks, stuck meters and unusual consumption only surface once the bill has gone out.
02The register sequence
A whole route, one meter at a time.
Load the meter roster once. In the field, the reader just frames and shoots: ScanGas reads, validates and moves on to the next meter.
The meter, on screen
Every meter on the route appears in order, with its address, account holder, serial number and previous reading.
Frame and shoot
The camera guide shows where the dial goes. One tap and the photo is taken.
AI reading in under a second
The model locates the dial, reads the index drums and confirms the serial number. Red decimal drums are ignored.
Checked against history
If a reading doubles the previous one or comes in lower, the app warns before saving. The drums can always be adjusted by hand.
Saved. Next.
The reading is stored on the phone, signal or not, and the screen moves on to the next meter by itself.
Animated app demo: the reader photographs the meter, the AI reads the index and the serial number, and the app moves on to the next meter by itself.
03Purpose-built AI
Not generic OCR. A model trained to read gas meters.
We trained it on real photos taken in the field: fogged dials, glare, dirt, low light and awkward angles. The conditions of an actual route.
Photo
The image the reader takes, straight from the camera.
Dial detection
A custom-trained neural network locates the counter within the image.
Reading
A vision model fine-tuned on field photos reads the index and the serial number.
Validation
The result is checked against the previous reading before it is saved.
- 2 stagesdetection + reading
- < 1 sfrom photo to data
- 1 photoserial and reading together
The model keeps improving with new field photos. Manual correction is always available.
Model illustration: it detects the dial, reads the whole-number drums, ignores the red decimals and reads the serial number.
04Meter mapComing soon
Your whole network, on a single map.
Every meter pinned on the map with up-to-date status. Readers know where to go next; coordinators see route progress without making a single call.
- Color-coded status: read, pending or flagged
- Guidance to every meter, even in hard-to-find spots
- Live progress of every route for the office
Meter map preview: a route progresses and meters switch from pending to read.
05Platform
From the route to billing, without retyping a thing.
Everything between the street and the office, handled in the same app.
A photo behind every reading
Every record keeps the meter photo with date and time. When a bill is disputed, the evidence is one tap away.
Alerts before saving
If a reading doubles the previous one or comes in lower, the app warns on the spot. High consumption stays flagged in the list.
Roster import
Thousands of meters from Excel, CSV or TXT, with account holder, address, serial number and previous reading.
Billing-ready exports
Excel, TXT or a PDF report with every meter's notes, ready for your billing system.
Notes that stay with the meter
Notes that carry over from one reading to the next: back-door access, loose dog, damaged dial.
What if the AI can't read a meter?
The reader sets the reading on the app's drums, just like a mechanical counter. The photo is saved anyway, as proof for the record.
Do readers need an internet connection in the field?
Not to record readings: readings and photos are stored on the phone and sync automatically when the signal returns. Automatic AI reading does use mobile data; without coverage, the reading is entered by hand on the drums.
Which meters can it read?
Mechanical gas meters with drum counters, the most common type on residential networks. From a single photo it gets the serial number and the whole-number digits of the index.
How do we load our meters?
Import your roster from an Excel, CSV or TXT file with account holder, address, serial number and previous reading. A single file can hold thousands of meters.
How do readings get into our billing system?
Export them as Excel or TXT (serial number and reading), or as a PDF report with notes. If your system needs another format, we will work it out together.
07Contact
Let's put ScanGas on your next route.
We'll show you ScanGas working with meters from your own network. No strings attached.
Message sent!
We'll get back to you shortly.