Home
Products and Services
Delivering Digital at Scale
SLB Software
Techlog Wellbore Software
Techlog Wellbore Software AI import and log prediction

Techlog Wellbore Software AI Import and Log Prediction

Enhance your wellbore data with the power of AI

Techlog
Support Contact Us Related Resources
Techlog Wellbore illustration

Introducing new AI tools for Techlog™ wellbore software, powered by generative AI to significantly enhance the interpretation and application of wellbore data.

By systematically extracting information from unstructured sources—including reports, images, and legacy documents—and automating interpretation processes, these tools streamline workflows and reduce the need for manual tasks.

Built to enhance user productivity and decision making, they unlock untapped value in existing data and provide consistent, high-quality results across wells and projects. Whether deployed on premises or in the cloud, the modules are designed to seamlessly integrate into existing environments, supporting teams wherever they work.

Just imagine if you could…

  • Eliminate manual data wrangling with automated data discovery.
  • Accelerate reviews by instantly unlocking insights buried in legacy reports.
  • Streamline data preparation for seamless quantitative interpretation.
  • Achieve better results with smart workflows and model-driven recommendation.

 

Let AI do the heavy lifting!
Extract calibration points from reports and predict logs effortlessly, anytime, anywhere.

Powering subsurface workflows with automation, geological intelligence, and machine learning

Techlog software AI tools deliver innovative features designed to streamline workflows and unlock greater value:

  • Automated data extraction from unstructured historical reports (free-text, PDFs, scanned documents, or handwritten notes), reducing manual effort, errors, and interpretation delays.
  • Prediction of missing logs using pre-trained foundational models, fine-tuned to your basin or field for data completeness and accuracy.
     
Male office worker using a computer

You’ll immediately benefit from the ability to:

  • Unlock information trapped in unstructured reports and integrate them efficiently using targeted domain-driven agents.
  • Gain deeper insights, faster, by combining AI-led predictions with domain expertise for more confident decision-making.
  • Use historical data to improve quantitative workflows, shortening decision times and minimizing downtime.
  • Deploy seamlessly across on-premises or cloud environments, ensuring flexibility, scalability, and compliance with regulatory requirements.
  • Integrate with existing Techlog software workflows to maintain continuity and implement AI-based improvements.
  • COMING SOON! Customize workflows with ease using Python scripting, no advanced coding skills required
     

Accelerating interpretation—import and predict with AI

Techlog software AI import

AI import

Operators often rely on as little as 10% of the available data when making critical investment decisions for exploration or joint ventures. Most of the information remains locked in unstructured formats—images, PDFs, or scanned reports.

The new Techlog software AI import tool tackles this issue by:

  • Enabling geoscientists to directly import and process unstructured data.
  • Supporting legacy formats such as TIF, LAS, and old images.
  • Streamlining data discovery and significantly reducing manual effort.
  • Automating core tasks like depth regression and quantitative analysis.
  • Making imported data immediately available for quality control and publication to the OSDU® Data Platform.

The result is improved data accessibility and governance across exploration workflows.

Techlog AI import tool
Operators often rely on as little as 10% of the available data when making critical investment decisions for exploration or joint ventures.
Techlog software AI log prediction

AI log prediction

This AI-powered log prediction tool is built on a globally trained log foundation model, developed using curated public data from more than 18,000 wells worldwide. It is lightweight and efficient, requiring no GPU to run.

Designed to accelerate and enhance subsurface interpretation, the log foundation model:

  • Provides a strong baseline that users can locally and privately tune with their own basin or asset data.
  • Enables customization by selecting calibration wells, defining inputs and targets, and running the tuning process directly in Techlog software.
  • Predicts missing logs across other wells with high accuracy and minimal training data once tuned.
  • Improves performance—in a shear velocity prediction use case, it reduced the median absolute error by 50% compared to traditional AI methods.
  • Will support future integration, extending access to customer-tuned models across other Delfi platform apppllications, including Petrel software, for seamless and scalable subsurface workflows.
Techlog AI log prediction
AI-powered log prediction tool is built on a globally trained log foundation model.
Log prediction - Data input and output
Scheme to generate input and output samples when training the pre-trained model. Intervals of size 256 are randomly selected in the wells of interest to be part of the training set or validation sets. Data alteration represents an intensity shift on gamma ray log (dark green).
 Error Statistics  Variable  RMSE  MAE  STD  PSNR
 MLP  DTC_Y  9.13  5.63  8.99  26.82
 Log FM  DTC_Y  6.04  3.54  6.01  30.40
 XGBoost  DTC_Y  6.78  4.20  6.76  29.40

The table compares the performance of three models (MLP, Log FM, and XGBoost) in predicting DTC_Y across test wells using error statistics: RMSE (Root Mean Square Error), MAE (Mean Absolute Error), STD (Standard Deviation), and PSNR (Peak Signal-to-Noise Ratio). Log FM outperforms the other models with the lowest RMSE (6.04), MAE (3.54), and STD (6.01), as well as the highest PSNR (30.40), indicating more accurate and reliable predictions. Compared to MLP, Log FM reduces RMSE by 33.8%, MAE by 37.1%, and STD by 33.1%, while improving PSNR by 13.3%. This demonstrates Log FM's superior performance in predictive accuracy and noise handling.

Related products

Related resources