This project presents an AI-driven pipeline for subsurface data management, integrating automated classification and metadata extraction to transform large volumes of unstructured technical documents into a structured, searchable digital repository. The classification methodology employs file hashing to detect duplicates and identify master files, followed by AI-driven categorization that distinguishes between Petroleum Engineering and Geoscience assets based on a predefined data hierarchy. Processing an initial 30,545 technical documents, the system identified 17,950 unique master files, removed 12,595 redundant copies, and categorized assets into structured domain and its respective group folders, enabling rapid data standardization within minutes.

Building on this foundation, Large Language Models (LLMs) were applied to automate metadata extraction across thousands multi-format files, including well reports, geological records, and digital logs such as LAS and DLIS formats. Combining Optical Character Recognition (OCR) and Natural Language Processing (NLP), the system converts unstructured technical reports into machine-readable text, with ML models mapping key entities to four OSDU￾compliant templates tailored for distinct engineering workflows. A human-in-the-loop Quality Control (QC) engine allows Data Analysts and Subject Matter Experts to verify and refine AI outputs, resolving misclassifications and metadata inconsistencies. The most impactful outcome is an 80% reduction in processing time, compressing months of manual effort into weeks, redirecting engineers toward high-value analysis and establishing a scalable, high-fidelity foundation for OSDU operations.

This presentation introduces the Data Delivery Plan (DDP), a tool developed by Equinor to improve planning and reporting of wellbore data. The solution provides a structured overview of planned data acquisition aligned with regulatory requirements (Blue Book Table A-1) and the drilling programme, enabling consistent and compliant reporting.

The DDP application allows users to define, review, and export a complete data acquisition plan through a guided workflow covering wellbore information, dataset selection, and final validation. It ensures visibility of mandatory datasets and supports improved completeness, transparency, and communication across operators, service and log QC providers, Diskos and authorities.

Two implementation options are presented: a lightweight version for creating and exporting plans locally, and a more advanced version with user login, storage, and dashboard functionality for managing multiple plans. Overall, the DDP aims to standardise data planning, reduce errors and gaps in reporting, and enable more efficient, consistent, and compliant data delivery to authorities and internal systems.

TGS' internal Knowledge Platform (TKP) is a practical example of how Information Management in the E&P industry can move from fragmented, system-specific work toward a unified, AI-enabled operating model. Positioned internally as "AI for Efficiency and Growth," TKP is designed not as a narrow point solution, but as a truly cross-organizational platform. This presentation will focus on real-world implementations and workflows that have been supercharged with AI to deliver efficiency gains, turning work that previously took days or weeks into something that can be progressed in minutes. In practice, TKP brings together internal knowledge and external sources of information in a manner that is seamless and invisible to the end-user, and without duplicating or transforming data. Access to data and information living in silos created intentionally and unintentionally, such as contracts, transactions, shipments, and financial information, is democratized and augmented with external intelligence such as industry news in one secure framework, allowing users to query, synthesize, and generate reusable outputs through natural-language, agent-powered workflows. We will also touch on the actual infrastructure and implementation behind a genAI-based framework, including its agent orchestration layer.

Capital-intensive projects in the energy industry face constant vulnerabilities due to extreme commodity price volatility. While digital transformation promises advanced risk management through AI and Big Data, its effectiveness is frequently bottlenecked by a critical technological skill gap. To thrive, decision-makers in non-IT areas, such as industrial Project Management, must transcend traditional competencies and develop robust digital literacy.

This presentation explores how fostering a strong data culture empowers professionals to interact effectively with complex predictive systems. While state-of-the-art neural forecasting models excel at capturing market patterns, they often lack contextual adaptability during the result analysis phase, frequently delivering outputs in overly technical formats that hinder strategic interpretation.

To address this, we propose a "Cognitive Information Fusion" framework that formally integrates Human-Computer Interaction (HCI) with AI. Its primary goal is to foster accurate mental models of AI systems among non-technical users. By treating the HCI interface as an active fusion node rather than a passive dashboard, this system translates high-dimensional predictive uncertainty into interpretable visual analytics. This approach champions Professional Development by enabling non-technical teams to align algorithmic outputs with their own heuristic knowledge and critical thinking.

Evaluated in a real-world energy sector scenario, our proposed Human-in-the-Loop (HITL) methodology demonstrates that synergizing machine precision with human strategic oversight yields highly resilient decision-making. Ultimately, investing in digital literacy and human-centric design bridges the gap between algorithmic accuracy and practical economic utility, transforming passive users into active, critical thinkers in volatile environments.

REVEAL — "Subsea Revealed" — is the current phase of Aker BP's Subsea Transformation and a fundamental rethink of how subsea integrity is managed. It shifts the model from inspection campaigns and reports toward visible, trusted, decision ready insight: raw data captured by next generation sensor carriers is turned into clear evidence engineers can act on, served directly into a 3D field twin environment, to compe current observations against historic baselines. The aim is to bring hardware, engineering, visualisation, AI, and integrity workflows into a single end to end approach that reduces manual interpretation and prepares the ground for autonomous operations. None of this is possible without a deliberate data foundation. This presentation argues that data, data governance, and data management are the true enablers of REVEAL — the difference between data engineers merely receive and data they can trust and act on. It shows how Aker BP structures core subsea data products, governs geospatial products in ArcGIS, establishes a federated governance model across the GIS, Field Twin, and Subsea Data Management teams with clear owner/steward/custodian roles, and runs a single data issue service loop. Attendees will see why embedding governance from the outset makes a sensor driven subsea transformation deliverable

The growing volume and complexity of geoscience data require scalable solutions for efficient data governance and workflows. This paper demonstrates how data-driven approaches and digital transformation principles can be applied in practice through a data management platform, using Blueback Project Tracker as a case study.

Blueback Project Tracker scans and structures geoscience data, enabling users to identify data duplication, mispositioned assets, and large datasets. Automated project size calculation and integration with visualization tools such as Power BI support data-driven decision-making.

Beyond monitoring, the platform is used in data assessment and consultancy workflows to extract Petrel metadata and support data rationalization. Using command-line (CLI) functionality, standardized workflows are applied, including consolidating data from multiple projects into a single environment, distributing data across projects, and reconnecting or externalizing seismic data to reduce redundancy.

The platform also supports key data management tasks such as project upgrading, archiving, and deletion. Batch processing with optional backup, rule-based validation, and flexible filtering improves efficiency and reduces operational risk. User activity monitoring and access control further support governance and data security.

Overall, this work demonstrates how systematic workflows powered by Blueback Project Tracker can turn data chaos into strategic advantage, eliminating redundancy, unlocking hidden storage savings, and support efficient collaboration across subsurface teams.

Structural interpretations, even of the same seismic survey, commonly reside in multiple siloed data stores. They are often created independently across projects, teams, applications or individuals, with different: goals, perspectives of geological context and, over time, availability of input or calibrating data (e.g. seismic, velocities, wells and production).

Even when discoverable across these silos, a lack of audit trail can result in absence of consistency and confidence in the interpretation data and its downstream use in modeling, resulting in unnecessary rework, inconsistent subsurface models, and delayed decision-making.

In the context of AI training, understanding of data context and quality are essential for good outcomes.

A vendor-agnostic governance framework enables interpretation discoverability with full context: who created it, with which seismic dataset and velocity model, what quality metrics were assigned, and how it was validated. Comprehensive information enables modelers to quickly assess if input interpretations are trustworthy and fit-for-purpose for their current project, eliminating unnecessary reinterpretation work.

We will demonstrate streaming seismic data directly from an OSDU® Data Platform, enabling high-performance visualization and interpretation without data duplication. Resulting interpretations and models are written back to the OSDU® Reservoir DMS with a clear record of inter-relationships and antecedents.

This capability proves particularly valuable when reconciling multiple interpretations created by different teams. Rather than engaging in subjective debates about which interpretation is "better," teams or agents can objectively compare interpretation contexts—examining which used more recent seismic data, which incorporated more well control, or which applied more sophisticated velocity modeling—and make data- and context- driven decisions about which to incorporate into an integrated subsurface model.

The continuous accumulation of seismic data resulting from decades of new acquisitions and reprocessing cycles presents significant challenges for seismic data management. As data volumes grow, issues such as duplicate file instances and seismic datasets effectively lost within large‑scale file systems become increasingly common. This presentation demonstrates that, through the use of specialised softwares, these challenges can not only be addressed but also enable the creation of a new interface that changes how geoscientists interact with seismic data.

Over the last 2 years, we have partnered with the NSTA to audit, reclassify, and re-tag nearly 1 million well data items. We will share what we learned about running a mega-scale classification project, including approaches to automation, quality control, and standardising approaches to classification across a diverse project team. During the project, the AI revolution started. We'll also share our thoughts about the impact AI has had, and will have in classification and taxonomy management.

Centralising subsurface data alone is not enough. This presentation shows how data discovery, archive modernisation, and AI-assisted enrichment improve data context, quality, and usability, helping organisations build trusted subsurface datasets across physical and digital archives and prepare them for modern workflows. Subsurface data environments have evolved over decades of exploration and production, creating a complex mix of physical archives, file-based repositories, and application-specific data stores. Seismic, well, and related datasets are often distributed across these environments with limited understanding of their content, condition, and context. Incomplete metadata and missing documentation mean significant effort is spent locating, validating, and preparing data before it can be reused with confidence. This presentation describes a practical approach that connects subsurface data discovery with archive modernisation, supported by AI-assisted techniques. Based on implementations using platforms such as Exploration Archives and Interica DataCentre, it shows how organisations can move from fragmented archives to more structured and usable data environments. The approach begins with metadata-driven discovery to establish understanding across datasets and documentation. AI-assisted techniques support metadata extraction, dataset classification, and identification of gaps, with human-in￾the-loop validation. Archive modernisation activities, including retrieval, quality control, format standardisation, and metadata enrichment, then improve readiness for downstream workflows. By linking discovery, AI-assisted enrichment, and preparation into a continuous workflow, organisations can reduce manual effort, improve access to trusted subsurface data, and increase confidence across interpretation, regulatory, migration, and analytical use cases.

Many organizations dedicate significant effort to defining data strategies and implementing management frameworks, yet a common challenge persists — a lack of confidence in the data itself. Issues such as poor data quality, unclear accountability, unreliable data source and fragmented processes often lead to inconsistent insights and limit the value organizations can derive from their data.

This presentation looks at why this gap in trust exists and explores where conventional data management approaches tend to fall short. It considers how the absence of clearly defined ownership and practical stewardship can weaken even well-intentioned data initiatives.

The session will also show how integrating data stewardship into everyday data management activities can make a measurable difference. It will outline practical ways to establish ownership, clarify responsibilities, and introduce governance practices that ensure data is not just handled, but properly cared for.

Insights and practical ideas will be shared on how to bring together data strategy, stewardship, and operational processes in a way that builds confidence in data. The session aims to provide a clear and realistic path toward creating reliable data environments that can support better decision-making.

Recent advances in agent-based artificial intelligence are transforming how subsurface professionals interact with enterprise data ecosystems, information models, and technical workflows. Instead of manually discovering, validating, and integrating data across multiple systems, intelligent agents act as context-aware assistants that interpret user intent, reason across distributed data sources, and orchestrate end-to-end data management processes.

This work presents a practical demonstration of an agent-driven approach applied to a subsurface use case using an open field dataset. Starting from a high-level objective, the agent discovers and evaluates available data assets across enterprise repositories, summarizes field and reservoir context, inventories wells and associated datasets, and assesses data completeness, consistency, and quality. The agent integrates information from both internal data platforms and external regulatory and public data services to identify and resolve gaps, including missing interpretation data.

Building on this foundation, the agent generates reproducible workflows for data ingestion, transformation, conditioning, and enrichment within a unified environment aligned with enterprise data models. It demonstrates how heterogeneous data can be harmonized through metadata-driven approaches and interoperable services. As part of the workflow, the agent constructs an analytical pipeline, including a machine learning model to estimate missing subsurface properties.

By interacting with data catalogs, services, orchestration frameworks, and computational platforms, the agent highlights the importance of interoperability and standardization. This work shows how agent-based systems bridge the gap between domain intent and governed, reproducible workflows, improving data accessibility, strengthening data quality, and accelerating integrated digital subsurface workflows.

Data exchange between software and companies is a challenge:

Point-to-point connections

Complex to manage Proprietary API’s

Expensive to implement

Differences in data definitions

Hard to map Different approaches to governance

Difficult to get access

The oil and gas industry operates a large portfolio of software applications. These digital solutions help us achieve safer, more efficient planning and operations. As Equinor continues with our digital transformation, we have an increased need for data exchange. Most of our data currently resides in proprietary systems with custom APIs. For Equinor to quickly adopt new technologies, we must find more efficient ways of sharing data.

How do we create value through technical solutions, innovation, and lessons learned? How can you showcase innovative methods, practical experiences, and measurable impact.

OSDU aims to solve data exchange with their standardized data models and platform.

OSDU platform as a master data store, binding point for all applications.

Iterate data exchange more frequently.

Standardized API’s, publicly documented that can be implemented equally for all.

Standard data models, still hard to map, but only needs to be done once.

Easier to cross domain boundaries.

Equinor aims to solve the software interoperability issues by committing to the OSDU platform and standards. We’ve taken a proactive approach by migrating our drilling and well data to OSDU, allowing our partners faster access to our standardized data. We are also collaborating with our partners in making software OSDU compliant, opening up for new and more efficient work processes. Other technologies adopting the OSDU open standards will also benefit from much faster onboarding in our software portfolio.

Effective subsurface data management plays a critical role in optimizing resources and supporting informed decisionmaking in geoscience operations. This study demonstrates a collaborative effort between Cegal and an upstream SEA NOC's innovation team to enhance data integrity and operational efficiency within the subsurface projects environment through the deployment of Blueback Project Tracker.

The initiative begins with a comprehensive evaluation of the existing data landscape, systematically identifying corrupt datasets, obsolete files, and coordinate reference system (CRS) inconsistencies across an extensive portfolio. During the initial phase, more than 5,000 subsurface projects spanning multiple software versions from 2007 to 2024 were assessed. This thorough inventory established the foundation for targeted data optimization activities.

A key focus involves addressing data duplication across well, interpretation, and seismic datasets. By implementing globally unique identifiers (GUIDs) and MD5 hashcode methodologies, redundant data entries were efficiently detected and eliminated. This systematic approach enhanced data integrity while achieving potential storage reduction of approximately 18 TB, significantly improving operational efficiency.

To sustain ongoing improvements, robust monitoring mechanisms were established to track project performance and data quality over time. A critical outcome of this continuous improvement process is maintaining a clean data environment, which enables the establishment of stronger governance protocols.

The results emphasize the value of modern data management methodologies in rationalizing subsurface data, maximizing operational efficiency, and reducing costs. This initiative demonstrates how structured approaches to data governance, combined with appropriate technological solutions, can transform geoscience data environments and support sustainable knowledge management practices.

Is Diskos an enabler for new developments and analysis of the Norwegian Continental Shelf, or are we just more of the same... Status of the now more than 31 years old colaboration project. And, maybe some interesting news on new ways of utilizing the vast amout of data and information that has been collected and shared for decades. Everybody wants to create value - can we help?

Through years of operations, acquisitions and mergers, organizations are often left with a portfolio of outdated applications - each holding critical pieces of asset history that cannot be lost. Yet legacy systems are often kept alive far longer than necessary, not for their functionality, but for the data they contain. This session explores how Aker BP separates data from their systems, decommission legacy applications, and still retain access to trusted information across the full asset lifecycle. By consolidating fragmented data landscapes into a structured, searchable archive, Aker BP reduces technical debt, supports compliance, and unlocks new value from historical data. Because the real challenge is not retiring systems, but ensuring the information they contain remains available.

In Collaborative Well Planning (CWP), data delivery speed is often assumed to be an integration challenge. In practice, the dominant constraint is human behaviour under time pressure: last minute data submissions, informal handovers and unclear ownership that undermine trust and traceability. These behaviours routinely stretch data preparation from hours into days—and carry forward into downstream workflows. This presentation describes how a shift towards small, user prepared data packages reduced CWP data delivery from 1.5 days to as little as 1.5 minutes, without adding heavyweight process or central bottlenecks. The change was enabled by empowering users to take control of their data and thus bypassing the formal data preparation and readiness checks that were needed to ensure successful well planning sessions. The human dimension is at the core of this, and why users disengage with the prescribed methods, even though they are straightforward and simple. By designing data workflows around real behaviour—rather than perceived ideal behaviour—governed data packages become an enabler of speed and not an obstacle.

As part of Equinor’s initiative to increase data governance maturity, an updated enterprise data strategy has been developed to address both evolving external demands and internal capability gaps. The existing data strategy, established in 2019, laid the foundation for a federated data governance model and the development of initial data capabilities. While progress has been achieved since then, maturity and adoption remain uneven across the organization. At the same time, rapid advancements in the digital landscape—including increased use of artificial intelligence alongside stricter regulatory and operational requirements—have created a need to reassess and strengthen the strategic direction.

This presentation describes the development of the updated data strategy, created through a six-month co-creation process involving key stakeholders across the enterprise and guided by Equinor’s standard strategy development framework. The resulting strategy defines three enterprise-level strategic focus areas, each supported by clear ambitions and goals that aim to drive consistency, scalability, and value realization from data.

The presentation will outline the development process, including the applied framework, stakeholder engagement approach, and key work methods, as well as challenges encountered along the way. It will also introduce the strategy on it's current state and its delivery package, including a proposed path forward for implementation. Finally, lessons learned from the process will be shared to provide insights for similar initiatives.

Energy companies hold large volumes of operational and subsurface data in table-based formats such as spreadsheets, relational databases, and flat-file exports. Ingesting this data into an OSDU platform today often relies on bespoke, one-off solutions that are difficult to maintain, inconsistent in governance, and hard to reproduce across teams or data domains. A standardised, repeatable pattern is needed to turn bulk ingestion into a governed, scalable activity that any OSDU adopter can apply.

This work proposes a pattern that standardises bulk ingestion from table-based sources into OSDU by combining OSDU Data Definitions with the platform's native services.

The pattern is built around five elements:

1. Source dataset: the table-based data to be ingested.

2. Data mapping file: the core artefact of the pattern, defining how source columns map to a target Well-Known Schema or Custom Schema.

3. Data partition: determines tenancy and isolation within the OSDU platform instance.

4. Ingestion engine: an execution component that reads the mapping file and runs the transformation and loading algorithm against OSDU platform services.

5. Governance configuration: ACLs and Legal Tags applied to every ingested record, ensuring compliance from the moment data enters the platform.

By packaging these elements into a single repeatable pattern, organisations can onboard new table-based datasets with minimal custom development, while keeping governance and schema alignment consistent across data domains. The approach is platform-agnostic at the mapping layer and can be adopted by any OSDU-compliant deployment, helping the community reduce duplicated engineering effort around bulk ingestion.

Seismic data is one of the most important technical assets in upstream oil and gas, but in many organizations it is still managed mainly as large SEG-Y files stored in archives, project folders, or application-specific repositories. This makes the data difficult to discover, slow to access, hard to govern, and not ready for analytics or AI workflows.

This presentation proposes a greenfield architecture for a modern seismic data platform that transforms seismic from a file-centric archive into governed, AI-ready data products. The design preserves SEG-Y as the original archive format, while introducing MDIO/Zarr as a modern multidimensional working representation for seismic volumes. By storing the data on S3-compatible on-premise object storage, the platform enables scalable access while supporting organizations that need to keep sensitive subsurface data within their own environment.

The architecture covers the full seismic lifecycle, from SEG-Y ingestion and validation through MDIO/Zarr conversion, catalog publication, and API-based access. It is organized around Bronze, Silver, and Gold data zones, with metadata, lineage, quality, and entitlement treated as platform layers rather than afterthoughts. AI-ready products such as seismic slices, attributes, and 3D patches are exposed as governed datasets that can support machine learning for fault detection, facies analysis, and subsurface data discovery.

The presentation is based on an ongoing architecture and prototype effort using openly available seismic data such as Volve and F3. The expected contribution is a practical reference design for E&P data professionals who want to make seismic data more accessible, more governed, and ready for analytics and AI, built on infrastructure that organizations can realistically own, without depending solely on public cloud platforms.

This presentation shows how large language models (LLMs) can be used to navigate complex data on the Norwegian Continental Shelf (NCS). Central to this is the NOD FactPages, available through REST APIs/Data serivce, which provide structured data.

The data service offers rich metadata describing the data itself and rich metadata about the relationships between tables and data is provided. This makes it possible for LLMs to act as an interface, where users can ask complex questions and get answers based on AI combining relevant data through "navigating" the APIs.

Sodir, together with FabriqAI, has applied this approach in the Knowledge Navigator. This solution uses LLM agents to retrieve and combine information from all public NOD data sources, both structured and unstructured, Tthe FactMaps API and data service forming a key backbone to navigating . To the right unstructured data/information, if the Factpages is not itself sufficient to provide an answer.

Good metadata provided in a format that is AI-native through well-described APIs is grounds for value creation by navigating and receiving relevant information using LLMs.

Reliable labor‑effort estimates are critical for asset management and cost control in offshore maintenance. In a large energy company, the lack of standardization and the reliance on free‑text descriptions and decentralized records made it difficult to consolidate man‑hours (HH) indicators, reducing visibility of effort in work orders and preventing consistent comparisons across operating areas. To address this information‑quality and governance challenge, we developed and deployed (since 2025) the HH Calculator: an analytics solution powered by generative AI that integrates the corporate maintenance system (SAP Plant Maintenance), an enterprise data & analytics platform (Databricks lakehouse), and interactive Power BI dashboards.

The solution uses a Large Language Model (GPT‑4.1) to interpret free text from technical inspection reports and work orders, automatically extracting standardized attributes such as “Action” (e.g., replace, swap, inspect) and “Object” (e.g., valve, piping, pump), and estimating the associated labor effort. GenAI functions as a semantic enrichment and historical standardization layer, without requiring changes to legacy systems or redesign of existing workflows.

Results show the approach is both feasible and valuable: >94% accuracy in attribute extraction (Action 94.96%; Object 92.31%; Tag 91.83%), 100% coverage of work orders with HH estimates, and >99% reduction in processing cost (BRL 795 vs. BRL 446,783 for 7,007 records). Lessons learned include: AI‑driven standardization exposed systematic master‑data inconsistencies and triggered improvements in data capture at source; continuous validation with domain experts and close collaboration across business, engineering, and IT were essential to sustain accuracy, traceability, and auditability. The case illustrates how generative AI can strengthen information management in maintenance—connecting data, information, and people—and can be replicated across the energy industry.

Many organizations are rich in data, systems, and digital tools, yet struggle to realize their full potential. This presentation introduces the concept of the “gap to potential” — the difference between current ways of working and what becomes possible when data is effectively utilized.

Drawing on experience from Digital Twin implementations and Information Management in operations, the session highlights how existing data foundations — structured engineering data, historical records, and integrated platforms — already enable significant value creation. It shows how these foundations support new ways of working, reduce manual effort, and shift focus toward better decision-making.

The presentation emphasizes that high-quality, governed data is the key enabler for AI. Without it, AI amplifies inefficiencies; with it, AI can connect fragmented information, reuse historical data, and deliver insights at scale. The key message: AI creates value by closing the gap to potential — enabled by strong Information Management, robust data foundations, and a focus on scalable impact.