Skip to content
AtheronLABS

You're visiting from the United States. Prices are shown in US dollars. Not right?

labs@atheron:~/industries/oil-gas/ai$ train agent --grounded

AI for pipelines

A corrosion model trained on your own images.

An integrity programme produces more inspection images than engineers can look at. A vision model trained on your own labelled images can sort them, so engineers start with the worst. It flags; the engineer decides.

// what matters here

What is different in this industry.

01

Trained on your assets

Coatings, pipe grades and conditions vary by operator. We fine-tune on your own labelled images, which is why the model is custom rather than a general one.

02

Scored before it is trusted

The model is measured on a held-out set of your images, and the score it must reach is agreed before it goes into use.

03

Engineers stay accountable

Every flag is reviewed by an engineer, overrules are recorded, and the model never closes a finding on its own.

04

Your images stay yours

Training and serving run in your own cloud account, and the trained model belongs to you once paid for.

// example projects

Priced examples for this service.

Each one is hypothetical, labelled as such, and priced live from our rate card. Open any of them in the estimator and make it yours.

Example projectWeb appAICustom LLM

Pipeline corrosion model

Hypothetical. Not a client, not a result.

// The problem

Integrity engineers review thousands of photos from digs, drone flights and facility inspections by eye, and the backlog decides which sites are looked at first.

// What we would build

A custom vision model, fine-tuned on the operator's own labelled inspection images, that flags coating damage and corrosion for an engineer to review, served in the operator's own cloud account so the images stay theirs.

// What is in it

  • Images labelled by your integrity team in a simple tool
  • A model fine-tuned on your images and measured on a held-out set
  • A review queue ranked by severity, where engineers confirm or overrule
  • Findings linked to the pipeline segment or facility
  • Overrules saved as the next round of training data

// Stack

  • PyTorch
  • ONNX Runtime
  • Python
  • Next.js
  • AWS GPU instances

// estimate

Build
≈ US$149,000 to US$227,000, delivered within 36 weeksCAD 212,000 to 323,600
Hosting
≈ US$7,020 a monthCAD 10,000 a month
Support
≈ US$2,500 a monthCAD 3,565 a month

Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.

AI route
Custom LLM
Model running cost
≈ US$6,530 a monthCAD 9,300 a month

Timeline by milestone

Discovery
2.9 to 3.4 weeks
Data and evaluation plan
0.4 weeks
Design approved
2.4 to 4.4 weeks
Core features
1.9 to 3.4 weeks
Full build
1.4 to 2.4 weeks
Training
8.9 to 15.4 weeks
Testing and fixes
0.4 weeks
Evaluation and red-teaming
3.9 to 4.4 weeks
Launch
0.9 to 1.4 weeks
Production
0.9 to 1.4 weeks

How it is paid

Deposit 20%
CAD 42,160 to 64,480
Discovery 1.3%
CAD 2,740.40 to 4,191.20
Data and evaluation plan 8.2%
CAD 17,285.60 to 26,436.80
Design approved 1.3%
CAD 2,740.40 to 4,191.20
Core features 4%
CAD 8,432 to 12,896
Full build 2.6%
CAD 5,480.80 to 8,382.40
Training 24.8%
CAD 52,278.40 to 79,955.20
Testing and fixes 1.3%
CAD 2,740.40 to 4,191.20
Evaluation and red-teaming 16.5%
CAD 34,782 to 53,196
Launch 1.3%
CAD 2,740.40 to 4,191.20
Production 8.7%
CAD 18,339.60 to 28,048.80
Holdback, 30 days after launch (10%)
CAD 21,080 to 32,240
GPU time for training and testing, at cost
CAD 1,200. Billed up front, at cost, outside the milestones

// questions

Questions we get about this.

How many labelled images does a corrosion model need?

It depends on how many conditions you want it to tell apart. We start with what you have, measure, and tell you which conditions need more examples.

Can it read in-line inspection data as well as photos?

In-line inspection data is a different problem from photos. We can scope it separately; the example here covers images.

Why not a general AI model?

General models describe a photo well but do not know your coatings or your severity scale. Fine-tuning on your labels is what makes the ranking useful.

// next

Not quite your project?

Tell us what you have in mind. We will come back to you with a range and the questions that would narrow it. Or book a call and talk it through.

AI for oil and gas integrity: custom corrosion vision models | Atheron Network Labs