Tela™ agentic-AI assistant catalog excerpt

Connect with an expert

Challenges

• Adapting to new workflows and modules can be challenging for users.
• Interrogating large projects with legacy data types and metadata can be complex and time-consuming.
• Setting up and organizing visual displays often involves lengthy and cumbersome processes.

Solutions

• Instant user guidance, calling on help documentation and subject matter expert content.
• Enables users to explore and analyze data within a Petrel software project through natural, conversational requests.
• Create displays and advanced visualizations with simple prompts.

Petrel software AI assistant

Task

• Visualize and interrogate data
• Learn new workflows and modules

Domain

• Subsurface

Technology

• Embedded in Petrel software
• Generative AI

Benefits

Complexity of original task

  Time saving
Access hundreds of documentation pages instantly, customized to needs.

  Democratize
Interrogate data in meaningful, insightful ways not possible before.

  Impact on decision
Set up visual displays with simple prompts compared to tens of button clicks.

Connect with an expert

Challenges

• Operators use only a small fraction (as little as 10%) of available data to make critical investment decisions.​
• Most data remains locked in unstructured formats such as PDFs, images, and scanned reports, making it difficult to access and utilize.
• Manual data processing is time-consuming and prone to errors, limiting efficiency for use in exploration and joint ventures.

Demo

Solutions

• The Techlog™ wellbore software AI import tool helps geoscientists import and process unstructured data, including legacy formats like TIF, LAS, and images.
• Leverages domain specific processing pipelines, including vision language models and optical character recognition (OCR), to extract information from unstructured sources.​
• Makes imported data instantly available for QC and OSDU® Data Platform publication, boosting governance and accessibility.​
• Unlocks full data value, enabling more informed and accurate investment decisions. Available on premises or in the cloud.

AI import tool

Task

• Import data (core, survey, formation properties etc)

Domain

• Geosciences

Technology

• Techlog software

Benefits

Complexity of original task

  Time saving
Simplifies handling of old and unstructured data formats.​

  Democratize
Cuts down manual work through automated processing.​
Speeds up data availability for validation and sharing.​

  Impact on decision
Supports better decisions by maximizing data usability.

Connect with an expert

Challenges

• Cumbersome analysis.
• Overwhelming documentation.
• Limited insight generation.
• Complex to query results in real time.

Demo: Data interrogation

Solutions

• Provides in-context guidance based on Petrel software and simulator documentation by providing quick access to specific information through natural language queries.
• Understands the simulation outputs for a large ensemble of results and can relate the user query into mathematical concepts.
• Enables the user to provide instructions that can be used to extract custom-selected information from a set of simulation models.

Demo: Workflow guidance

Demo: Results visualization

Petrel software AI assistant for RE

Task

• Reservoir engineering and modelling

Domain

• Reservoir engineering

Technology

• Petrel software

Benefits

Complexity of original task

  Time saving
Smart documentation navigation​.

  Democratize
Real-time insights ​and visualization.​

  Impact on decision
Intelligent model interrogation​.

Connect with an expert

Challenges

• Limited availability of high-quality, basin-specific training data.
• Inaccurate or incomplete logs affecting interpretation quality.
• Fragmented workflows across platforms, limiting model reuse and scalability.

Demo

Solutions

• Uses a pre-trained log foundation model trained on data from more than 18,000 wells.
• Allows users to locally tune the model with their own data for higher accuracy, or train from scratch with best-in-class models like XGBoost.
• Predicts missing logs with minimal training input and strong performance.
• Will integrate with other Delfi™ digital platform applications like Petrel™ subsurface software for broader workflow impact.

AI log prediction

Task

• Geomechanics, petrophysics

Domain

• Geosciences

Technology

• Techlog software

Benefits

Complexity of original task

  Time saving
​Runs anywhere (on premises, cloud).

  Democratize
Adapts to local data for higher accuracy.
Fills log gaps with minimal effort.

  Impact on decision
Scales across platforms for greater impact.

Connect with an expert

Challenges

• Logs often come in multiple formats, across a range of tools and services.
• Manual workflows are prone to human error, and a lack of integrated automation makes scaling log conditioning processes challenging.
• Absence of advanced analytics tools within workflows limits the ability to extract meaningful insights and optimize decision-making.
• Log conditioning remains a labor-intensive task, requiring significant time and effort to clean, process, and prepare data for analysis.

Demo

Solutions

• Discover all the data that is available to you from the OSDU™ Data Platform.
• Access data from the entire basin.
• Automate log conditioning with cloud-based machine learning algorithms that enhance efficiency, accuracy, and consistency in data processing, reducing manual effort and improving overall workflow performance.
• Enable seamless collaboration and data flow by utilizing integrated platforms that facilitate smooth data sharing and collaboration among team members, ensuring efficient and effective project execution.

Automated log conditioning

Task

• Data discovery
• Log conditioning
• Mapping

Domain

• Geosciences

Technology

• Data preparation workflows
• Wellbore Interpretation

Benefits

Complexity of original task

  Time saving
Automating data preparation for thousands of wells.

  Democratize
Time taken to prepare data reduced from months to days.

  Impact on decision
Working with all the data available, from different vintages.

Connect with an expert

Challenges

• Wasted hours hunting data—engineers spend too much time digging through multiple systems instead of solving problems.
• Slow, siloed decisions arise when critical answers are trapped within disconnected workflows and isolated expertise.
• Missed opportunities from buried insights—valuable patterns stay hidden in complex datasets, delaying action and lowering impact.

Solutions

• In the Lumi data workspace, conversational insights transform the way users interact with data by leveraging generative AI to enable natural language queries.
• This powerful tool allows knowledge workers to ask questions and receive precise, actionable insights directly from their data, combining AI with trusted data governance.

Lumi data workspace—conversational insights

Task

• Ask in plain language, get instant answers
• Find insights across domains in seconds
• See clear summaries instead of raw data

Domain

• Subsurface analysis
• Drilling operations
• Production optimization

Technology

• Lumi data workspace

Benefits

Complexity of original task

  Time saving
Cut decision cycles from weeks to hours with instant answers in plain language.

  Scalability
An AI assistant powered by the most reliable and current energy data.

  Impact on decision
Data delivered instantly so work gets done faster.

Connect with an expert

Challenges

• AI adoption remains fragmented—data scientists, engineers, and domain experts often work in isolated tools, making it difficult to collaborate or scale AI efforts.
• Slow, complex model deployment—moving from prototype to production requires multiple handoffs and technical expertise, delaying value creation.
• Limited visibility and governance—models, datasets, and outcomes are often disconnected, increasing risk and reducing trust in AI-driven decisions.

Solution

• Lumi AI workspace enables teams to build, train, and deploy AI solutions faster within a unified, governed environment.
• This powerful tool connects data scientists, engineers, and domain experts in a shared workspace—accelerating innovation and ensuring consistency from concept to deployment.
• Fully integrated with trusted data foundation of the Lumi platform, it bridges domain expertise with scalable AI, ensuring every model is grounded in reliable data.

Lumi AI workspace

Task

• Build and train AI models using low-code or pro-code tools
• Automate repetitive workflows with intelligent agents
• Monitor, evaluate, and improve models continuously

Domain

• Any petrotechnical discipline
• Data scientists
• AI experts

Technology

• Lumi AI workspace
• Integrated with Lumi platform AI foundation

Benefits

Complexity of original task

  Time saving
Cut AI project timelines from months to weeks through seamless data-to-deployment workflows.

  Scalability
Scale from single-use pilots to enterprise-level AI with full traceability and compliance.

  Impact on decision
Drive measurable impact by embedding AI directly into domain workflows—where decisions are made.

Connect with an expert

Challenges

• ESP wells often run below optimal efficiency due to fixed setpoints, delayed adjustments and manual tuning.
• Manual surveillance creates operational delays, inconsistent decisions, and limits scalability across large well portfolios.
• High power consumption and frequent failures drive up operating costs and increase downtime.
• Disconnection between surface systems, subsurface sensors, and cloud analytics prevents real-time, closed-loop optimization.

Solution

• Delivered in OptiFlow™ production assurance solutions leveraging edge computing to stream real-time ESP data to the cloud.
• Analyzes performance trends, identifies optimal operating conditions, and generates control recommendations.
• Autonomous or human-in-the-loop execution adjusts ESP setpoints.
• Continuous feedback loop between edge, cloud, and agent ensures adaptive and safe optimization at scale. .

ESP optimization agent

Task

• ESP monitoring and optimization
• Automated setpoint adjustments
• Optional human-in-the-loop notifications

Domain

• Production assurance
• Well and lift optimization

Technology

• OptiFlow solutions
• Lumi™ data and AI platform
• Edge AI and IoT

Benefits

Complexity of original task

  Time saving
Maximizes uptime and efficiency through autonomous, real-time ESP optimization.

  Democratize
Reduces power consumption and costs while minimizing failures and operational risks.

  Impact on decision
Scales across entire fields, automating workflows and freeing engineers for strategic tasks.

Connect with an expert

Challenges

• Manual tracking of energy data across multiple subsystems.
• Delayed detection of inefficiencies and energy deviations.
• Limited visibility into real-time optimization opportunities.

Solutions

• Use large language model (LLM)-powered agents to analyze real-time energy use and detect anomalies.
• Integrate simulation insights with LLM reasoning for optimization suggestions.
• Automate reporting and recommendations through contextual, natural-language outputs to operators.

Energy optimization agent

Task

• Process engineering

Domain

• Facilities operations

Technology

• OptiSite solutions
• Physics-informed AI
• Symmetry process twins

Benefits

Complexity of original task

  Time saving
Real-time visibility of energy performance.

  Democratize
Quick detection of inefficiencies.

  Impact on decision
Lower energy use and operating costs.

Connect with an expert

Challenges

• Traditional machine learning models require manual tuning for each asset, restricting their ability to scale effectively.
• Data fragmentation across systems and formats make it difficult to build continuous, context-aware models.
• Rule-based detection relies on static thresholds fail to adapt to evolving operational patterns.
• Existing TSFMs lack multi-variate capabilities and field specific training data to reflect real-world behaviour.

Solutions

• Multimodal modelfor forecasting, anomaly detection, fault diagnosis, remaining-useful-life (RUL) prediction and more.
• Trained on diverse, real-world datasets, capturing complex temporal dependencies with multivariate capabilities.
• Unifies fragmented data sources (e.g., logs, manuals, events) into context-aware, actionable insights.
• Seamlessly integrates into pre-existing solutions like dashboards, agentic workflows and real-time systems.

Time series foundation model

Task

• Forecasts performance and operational trends
• Detects anomalies and diagnoses faults

Domain

• Asset performance
• Asset performance management. (APM)

Technology

• OptiSite facility, equipment, and pipeline solutions
• Lumi™ data and AI platform

Benefits

Complexity of original task

  Time saving
Reduces manual work by eliminating the need for asset-specific model tuning and maintenance.

  Democratize
Delivers high-accuracy forecasts and detection at scale, adapting to evolving field conditions and equipment behavior.

  Impact on decision
Improves engineer productivity and confidence in predictive insights.

Connect with an expert

Challenges

• Limited visibility into production optimization opportunities makes prioritization difficult and inconsistent.
• Siloed and manual workflows hinder data integration and make interpretation inefficient.
• Reactive, delayed decisions result in untimely interventions and missed opportunities.
• Slow, disconnected expert-driven analysis delays actionable insights and reduces operational agility.

Solutions

• Unified and context-aware intelligence enabled by a dynamic view across wells, reservoirs, and networks, integrated with OptiFlow™ production assurance solutions and OptiSite™ facility, equipment, and pipeline solutions.
• Proactive opportunity discovery through an evergreen register of gaps and optimization opportunities.
• Autonomous analysis and simulation performed at scale in a connected, real-time environment.
• Decision-ready recommendations prioritized by impact, cost, and constraints for informed decision-making.

Wells reservoir facility management agent

Task

• Discover optimization opportunities proactively
• Deliver prioritized, decision-ready recommendations

Domain

• Production assurance and optimization
• Asset performance
• Facility managementt

Technology

• OptiFlow solutions and OptiSite solutions
• Lumi™ data and AI platform

Benefits

  Time saving
Enables autonomous orchestration across the production system for seamless optimization.

  Democratize
Integrates openly with data and workflows, minimizing effort and filling analysis gaps.

  Impact on decision
Scales across platforms to maximize impact and operational efficiency.

Connect with an expert

Challenges

• Evaluation of assets through acquisition, farm-in, or screening with government agency involves huge data sources of over 100 gigabytes, with both structured and unstructured data sources.
• Multiple data sources and data types such as reports, OFM™ well and reservoir analysis software, and Petrel™ subsurface software projects etc.
• 50-60% of the time is spent in data wrangling.
• It takes ~10-12 weeks for a quick evaluation.

Solution

• Generative-AI enabled extraction of data from structured and unstructured data sources into a structured database.
• Classification of documents for each of the domains e.g., production, reservoir, geology etc.
• Enables natural language interaction with asset data for intuitive asset evaluation.
• Uses agentic reasoning to answer relevant question from different domains based on the available data.
• Supports interactive conversation.
• Simplifies data complexity, focusing on insights and conversational interface.

Asset Evaluation Agent

Task

• Rapid asset evaluation
• Insight generation from different domains.
• Multi-agentic interactive asset assistant.

Domain

• Surface and sub-surface domains
• Data intelligence

Technology

• Conversational AI
• Agentic reasoning

Benefits

Complexity of original task

  Time saving
Accelerates asset evaluation through intuitive multi-agentic workflow.

  Scalability
Improves decision-making with contextual, data-driven insights.
Reduces manual effort in data extraction and interpretation.

  Impact on decision
Rapid access to all available insights for a user in one day.
Scalable solution for different assets, adaptable to multiple data environments.

Connect with an expert

Challenges

• Geoscientists must manually sift through vast seismic datasets, limiting analog search scope and scalability.
• Analog selection can be biased, reducing prediction accuracy.
• Inconsistent acquisition parameters across surveys make direct comparison of geological features difficult.

Solution

• The seismic foundation model (SFM)-based task transforms analog hunting with engineered AI that can quickly identify and isolate geological features.
• Browses huge volumes of seismic cubes at a rapid pace while providing control to the end user.
• Results are delivered as a high-resolution heatcube, directly imported into Petrel™ subsurface software for interpretation and model building.

Geological Analog Search

Task

• Geological analog search
• Geological similarity ranking
• Heatmap generation

Domain

• Seismic interpretation
• Prospect evaluation
• Exploration geoscience

Technology

• Seismic interpretation
• Petrel™ subsurface software
• Geological indexing engine

Benefits

Complexity of original task
 

  Time saving
Accelerates analog hunting from weeks to minutes.

  Scalability
Eliminates cognitive bias with statistically optimal analogs.

  Impact on decision
Reduces uncertainty in reservoir prediction and improves exploration outcomes and prospect ranking.

Connect with an expert

Challenges

• Manual production modeling is slow, fragmented, and requires deep technical expertise.
• Running simulations, sensitivity studies, and nodal analysis are repetitive and error prone.
• Siloed systems and complex coordination slow down decision-making and reduce agility.
• Simulation tools are powerful but difficult to access and scale across teams.

Solution

• Agentic AI workflows automate steady-state multiphase flow simulations and streamline production.
• Enables natural language interaction to run simulations, generate plots, and automate reporting through a web interface.
• Python Toolkit (PTK) AI assistant builds and executes Python-based workflows using knowledge graphs and embedded files.
• Provides contextual guidance and personalized recommendations using an intelligent knowledge assistant.

Pipesim simulator agent

Task

• Production simulation
• Nodal analysis
• Knowledge assistance

Domain

• Production engineering
• Flow assurance
• Well performance

Technology

• Lumi™ AI workspace
• Pipesim™ steady-state multiphase flow simulator

Benefits

Complexity of original task
 

  Time saving
Reduces manual effort and lowers the barrier to entry for advanced simulation tools.
 

  Scalability
Improves consistency, accuracy and decision speed.
 

  Impact on decision
More collaborative production modeling through agentic automation.

Connect with an expert

Challenges

• Decline curve analysis (DCA) is often manual, difficult to integrate with other workflows or data systems, and time-intensive to manage across large well portfolios.
• Iterating through different decline models is complex and prone to errors.
• Inconsistencies in execution lead to unreliable reserve estimates and production forecasts.

Solution

• The solution automates DCA using machine learning and historical data to detect outliers, carry out segmentation, and enables curve fitting to generate forecasts.
• It fits automatic packet reporting system decline models to segment data outputs and forecast parameters with time series datasets.
• An integrated interactive application enables users to forecast settings intuitively.

Production forecasting advisor

Task

• Decline curve analysis
• Production forecasting

Domain

• Production engineering
• Reservoir management

Technology

• Lumi™ AI workspace
• Intervention opportunity manager (IOM)

Benefits

Complexity of original task
 

  Time saving
Faster and more scalable production forecasting through intelligent automation.
 

  Scalability
Reduces manual effort and improves consistency across wells and fields.
 

  Impact on decision
Minimizes error margins and enhances reliability of reserve estimates.

Connect with an expert

Challenges

• Manual safety monitoring is slow, inconsistent, and difficult to scale across industrial sites.
• Traditional computer vision systems require large, labelled datasets, and retraining for each new use case.
• Safety oversight remains fragmented, reactive, and costly, with limited ability to adapt to site-specific procedures.
• Manual inspections struggle to keep up with changing safety guidelines and operational complexity.

Solution

• Combines video feeds with SOP documents using multimodal AI pipelines.
• Visual language models (VLMs) interpret both images and text to detect non-compliance in real time.
• Runs on cloud and on edge devices for low-latency, cost-effective deployment and privacy protection.
• Automatically extracts and applies SOP context, logging compliance status, and generating reports automatically.

Standard operating procedure (SOP) compliance agent

Task

• Real-time safety compliance monitoring
• Event logging and reporting
• Multimodal sensor integration

Domain

• Industrial safety
• HSE (health, safety, environment)

Technology

• Visual language models (VLM)
• Multimodal AI pipeline
• Edge computing
• Cloud dashboard

Benefits

Complexity of original task
 

  Time saving
Faster, more accurate safety oversight with automated, real-time compliance monitoring.
 

  Scalability
Reduces manual inspection and infrastructure costs by leveraging existing cameras with lightweight edge devices.
 

  Impact on decision
Scalable across multiple sites, supporting coaching, and hazard prevention.