Illustrative image: a tablet on a geologist's desk, beside a hammer, a compass, a hand lens and core samples; on its screen, a short summary of core porosity and permeability by sample.

Digital Geology: Cross-Checking Historical Well Data

Digital Geology: Cross-Checking Historical Well Data

Video for illustration only.

Video for illustration only.

The challenge

Decades of well data sit in scans and paper files, and documents for the same well often disagree.

The challenge

Decades of well data sit in scans and paper files, and documents for the same well often disagree.

1

1

Read

AI agents read well files of different ages and formats: wellhead coordinates, well deviation surveys, formation tops and core descriptions.

1

1

Read

AI agents read well files of different ages and formats: wellhead coordinates, well deviation surveys, formation tops and core descriptions.

2

2

Cross-check

Values for the same well are compared across documents, and every conflict is flagged.

2

2

Cross-check

Values for the same well are compared across documents, and every conflict is flagged.

3

3

Decide

The geologist decides which document is right.

3

3

Decide

The geologist decides which document is right.

1

1

Read

AI agents read well files of different ages and formats: wellhead coordinates, well deviation surveys, formation tops and core descriptions.

2

2

Cross-check

Values for the same well are compared across documents, and every conflict is flagged.

3

3

Decide

The geologist decides which document is right.

The work

On a producing-field project in the Volga-Ural basin, I built a small team of AI agents that consolidated and cross-checked historical well data: one read the documents, another compared the values, and every conflict came to me for a decision.

The pilot runs within the project's data rules: the models run locally, and any external model sees only anonymised data. Every value stays traceable to its source document.

What it showed: preparation of historical well data was cut from months to weeks (expert-validated). AI reads and compares well; the decision on which document is right stays with the geologist.

The next steps of the same work: a company library that answers, and documents turned into video. All eight tools and projects with their stages are in Tools and projects.

Context

Producing-field project, Volga-Ural basin

Step · Year

Data · pilot

Category

Digital Geology

Focus

AI agents · Historical well data · Cross-checking

AI agents · Historical well data · Cross-checking

The work

On a producing-field project in the Volga-Ural basin, I built a small team of AI agents that consolidated and cross-checked historical well data: one read the documents, another compared the values, and every conflict came to me for a decision.

The pilot runs within the project's data rules: the models run locally, and any external model sees only anonymised data. Every value stays traceable to its source document.

What it showed: preparation of historical well data was cut from months to weeks (expert-validated). AI reads and compares well; the decision on which document is right stays with the geologist.

The next steps of the same work: a company library that answers, and documents turned into video. All eight tools and projects with their stages are in Tools and projects.

Context

Producing-field project, Volga-Ural basin

Step · Year

Data · pilot

Category

Digital Geology

Focus

AI agents · Historical well data · Cross-checking

The workflow
From a paper archive to a single database of all wells, checked by a geologist.
01
Sources
Document types, scan quality and target fields reviewed.
02
Reading
Agents extract values from each document.
03
Cross-check
Values compared between documents; conflicts flagged.
04
Geologist review
Each conflict resolved by the geologist.
A single database of all wells, with conflicts resolved by a geologist
Built on
ScansWell filesAI agentsGeologist review
The workflow
From a paper archive to a single database of all wells, checked by a geologist.
01
Sources
Document types, scan quality and target fields reviewed.
02
Reading
Agents extract values from each document.
03
Cross-check
Values compared between documents; conflicts flagged.
04
Geologist review
Each conflict resolved by the geologist.
A single database of all wells, with conflicts resolved by a geologist
Built on
ScansWell filesAI agentsGeologist review

Proposal · PDF

The method in two pages

Tell me what is in your archive and what you need from it.

First page of the proposal: Digital Geology, cross-checking historical well data

Proposal · PDF

The method in two pages

Tell me what is in your archive and what you need from it.

First page of the proposal: Digital Geology, cross-checking historical well data