Ontology & Semantic Knowledge Graphs Strategic Positioning | TotalEnergies x Forvis Mazars

Stakeholder Interviews

What did internal stakeholders reveal about TSF, SousLeSens, data platforms, LifeX, and AI readiness?

Objective: Capture perspectives from advocates and skeptics to address real adoption barriers.

×

Contents

Stakeholder Interviews

1.Context

2.Executive summary

3.Stakeholder map

4.Knowledge Graph

5.Jean-Charles Leclerc

6.Benoit Guillermou

7.Josephine Boulanger

8.David Teixugueira-de-Castro

9.Bastien Jouret

10.Benoit Soleilhavoup

11.Francois-Xavier Mourot

12.Sonia Vallejo-Vaca

13.Pascal Colombani

14.Elie Maze

15.Pierre Jallais

16.Julien Campan

17.Conclusion

Context

Ontology & Semantic Knowledge Graphs Strategic Positioning

TotalEnergies and OneTech already have a solid semantic foundation through TSF and SousLeSens (SLS). The strategic issue is adoption: leadership needs to see why governed meaning matters, where it creates measurable value, and how it fits the OneTech landscape across platforms, operations, data products, and AI.

The work builds on a first mission conducted from November 2025 to January 2026, the Cross-FPSO Semantic Preparation and Ontology Vetting Program, where Forvis Mazars assembled the team with NCOR, the National Center for Ontological Research. That mission validated TSF's alignment with BFO (Basic Formal Ontology, ISO/IEC 21838-2:2021) through a formal audit of the TE Business Objects ontology, and delivered a machine learning prototype that matched 23,000 Pazflor functional locations to Dalia equivalents to bootstrap cross-FPSO job card transfer for life extension planning. This second mission turns that validated technical foundation into a positioning narrative for TSF and SousLeSens within OneTech.

5 coordinated workstreams and 5 client-facing decks.
12 internal stakeholder interview profiles and 3 Voice of Peers interview profiles.
9 benchmarked tool profiles across semantic control, platform, graph, catalog, and AI layers.
25 strategic intelligence organization profiles backed by 26 evidence notes.
Management assets include the Executive Deck, one-pager, FAQ, use-case showcase, decision framework, RFI/RFP support, benchmark matrix, and evidence registers.

This deck is the internal evidence layer. Use it to understand the 12 stakeholder perspectives behind adoption: where TSF and SLS are trusted, where platform boundaries create friction, and what must be clarified before the semantic foundation can scale beyond expert sponsorship.

Executive summary

Internal stakeholders confirm that TSF needs a clearer operational focus, stronger platform integration, and a LifeX proof point.

What did the stakeholder interviews reveal about adoption, positioning, and value?

Strategic alignment

Stakeholders broadly support the semantic ambition when it is tied to concrete operational decisions.

Main friction

The message is clear technically but still hard to consume for management, platform owners, and operations.

Platform boundary

Cognite, Collibra, Fabric, Search, and AI should consume governed meaning rather than redefine it locally.

Operational anchor

LifeX is the strongest proof point because the value chain runs from field evidence to CAPEX, OPEX, risk, and COMEX reporting.

What enables adoption

Concrete LifeX decision questions.
TSF outputs consumable by Cognite, Collibra, Fabric, search, and AI.
Governance language linked to OPEX, CAPEX, production, safety, and risk.
Clear owner roles across semantic authority, platforms, and operations.

What blocks adoption

Semantic portfolio hard to navigate for sponsors.
Platform teams hear semantic work as competing authority.
Use cases need stronger before / after evidence.
Operational workflows still rely on local tools and manual reconciliation.

Stakeholder map

Each layer sees a different part of the same adoption problem.

Twelve interviewees grouped by organizational layer around the TSF / SousLeSens semantic control layer. Blue boxes are advocates; white boxes are pragmatic platform owners; teal marks the LifeX operational anchor.

Knowledge Graph

What the interviews surfaced

Interview concepts from 12 stakeholder discussions, organized against the seven BFO bucket classes.

Concept key

Process

MAP Semantic mapping

VET Ontology vetting

JCT Job card transfer

DPC Data product creation

RAG Retrieval-augmented generation

Temporal

LXC LifeX decision cycle

Material

TSF TotalEnergies Semantic Framework

SLS SousLeSens

LX LifeX

CDF Cognite Data Fusion

COL Collibra

Site

FPSO Floating production, storage, and offloading

Information

BFO Basic Formal Ontology

CV Controlled vocabulary

DMP Data Management Plan

OWL Web Ontology Language

SKOS Simple Knowledge Organization System

ISO ISO/IEC 21838-2

CFI Capital Facilities Information Handover (CFIHOS)

Quality

DQ Data quality

LIN Lineage traceability

Role

SA Semantic authority

DG Data governance

API Semantic API layer

Stakeholder profile

Jean-Charles Leclerc

TSF / Innovation & Standards, OneTech EP

Signal

TSF is technically mature and externally recognized. The next step is to focus the portfolio on visible business outcomes and operational delivery.

Main processes

Maintains TotalEnergies Semantic Framework since 2021.
Represents TotalEnergies in CFIHOS, JIP36, PCA, AFNOR, and ISO work.
Supports SousLeSens and semanticization of industrial standards.
Worked on LifeX process semanticization and BFO alignment.

Pain points

Clarify where TSF fits beside Cognite, Collibra, and Fabric.
Move from semantic assets to a concrete V2 production deployment.
Structure the semantic portfolio in business terms.
Connect semantic outputs to operational workflows.

Implications

Focus TSF on a small number of mission-critical use cases, starting with LifeX, and package outputs so others can reuse and explain them.

Stakeholder profile

Benoit Guillermou

Data Officer OneTech EP, Data Governance

Signal

The semantic vision is understood, but it needs simpler packaging, concrete ROI, and scoped deliverables business teams can reuse.

Main processes

Drives data governance alignment across OneTech EP.
Coordinates data-domain, platform, and governance messages.
Frames the value of semantic knowledge graphs for leadership.
Connects Collibra, Cognite, and TSF positioning.

Pain points

Prove ontology value through practical business cases.
Make TSF understandable for management audiences.
Deliver finite outputs instead of open-ended research.
Anchor governance in inventory, quality, and observability.

Implications

The communication kit must translate semantic capability into decision language: where it applies, what it changes, and how value is measured.

Stakeholder profile

Josephine Boulanger

Transverse Data Modeler, Data Exposure & Product

Signal

Conceptual modeling captures business understanding, but much of that meaning is lost when delivery ends as SQL-like artifacts.

Main processes

Builds business-led conceptual models across branches.
Reviews standards including CFIHOS, PPDM, OSDU, and Allotrope.
Validates outsourced models and cross-branch consistency.
Bridges business language and data-platform implementation.

Pain points

Increase modeling capacity and reduce single-person dependency.
Mandate neutral models across branches.
Capture semantic knowledge alongside data exposure deliverables.
Prioritize business-driven use cases.

Implications

Stakeholder interviews confirm that semantic value depends on operating model, ownership, and tooling together.

Stakeholder profile

David Teixugueira-de-Castro

Head DEP / Cognite Programme Manager, OneTech

Signal

Cognite delivery needs semantic clarity to improve contextualization, but TSF must be positioned as a capability inside the data-platform reference model.

Main processes

Leads Cognite programme across EP surface assets.
Coordinates data products with branches and data officers.
Manages Quantum to Cognite migration and source contextualization.
Works with modeling teams on standards and implementable models.

Pain points

Position TSF as a capability, with modeler ownership.
Show before / after value through LifeX.
Integrate semantic work into the modeling lifecycle.
Benchmark semantic maturity against industrial peers.

Implications

The platform story should be cooperative: Cognite connects operational data; TSF governs meaning consumed by the platform.

Stakeholder profile

Bastien Jouret

Lead Data Modeling, Cognite EP, DEP

Signal

Semantic modeling will be heard when it directly reduces manual mapping work and supports business use cases inside delivery constraints.

Main processes

Designs Cognite enterprise data models and LPG structures.
Contextualizes multiple operational sources.
Co-builds models with business teams and decision logs.
Applies standards such as CFIHOS, SIFOS, and ISO 14224 where relevant.

Pain points

Start from business-driven cases.
Automate translation of non-aligned object descriptions.
Create time to assess TSF / SKG utility outside project pressure.
Improve source completeness and harmonized instrument standards.

Implications

The TSF to Cognite bridge should solve concrete mapping pain, then prove value through a narrow LifeX workflow.

Stakeholder profile

Benoit Soleilhavoup

Data Product Architect, DEP

Signal

Ontology adoption depends on proof of money, time, or safety gain. Product teams need process-level value stories before they invest.

Main processes

Architects governed data products on Cognite and Databricks.
Works on work orders, tags, equipment, documents, sensors, and time series.
Links use cases to Collibra governance.
Avoids duplicate pipelines across data platforms.

Pain points

Demonstrate concrete value before investing.
Tell process-level value stories.
Use ontology when the use case requires it.
Expose semantic assets through APIs for downstream checks and agents.

Implications

The decision framework should clearly separate SKG cases from cases better served by BI, catalog, RAG, or property graph approaches.

Stakeholder profile

Francois-Xavier Mourot

Data Governance / Collibra, DSC OneTech

Signal

Collibra needs a clear link to TSF business objects and a defined perimeter between catalog governance and semantic modeling.

Main processes

Facilitates data-domain and affiliate governance contacts.
Migrates Data Management Plans into Collibra.
Works on business objects, lineage, and ownership in the catalog.
Supports MDM facilitation for shared reference concepts.

Pain points

Mandate shared tool conventions.
Clarify shared reference ownership.
Define the perimeter between catalog and SousLeSens.
Import TSF business objects into Collibra with agreed mapping.

Implications

TSF should feed Collibra with governed concepts; Collibra should expose ownership, lineage, and reporting readiness.

Stakeholder profile

Sonia Vallejo-Vaca

Head Data Intelligence & Strategy, OneTech

Signal

Semantic governance must be practical, measurable, and connected to search, EDM, Cognite, and AI agents.

Main processes

Leads operational data governance priorities.
Works on reference data, quality rules, and lineage.
Connects Smart Search, RAG, and ontology snippets.
Coordinates agentic data roadmap and multi-branch programmes.

Pain points

Tie governance to production, OPEX, emissions, and business priorities.
Define governance maturity metrics.
Explain integration with search, EDM, Cognite, and agents.
Specialize LLMs by industrial domain over time.

Implications

The semantic layer must become an integration service consumed by search, data products, and AI workflows.

Stakeholder profile

Pascal Colombani

Data Officer EP, TotalEnergies

Signal

Surface data maturity varies across affiliates. Data ownership, DMP completeness, and quality controls are prerequisites for scaling semantic outputs.

Main processes

Oversees data governance across EP affiliates.
Supports Data Management Plan completeness and compliance.
Drives acculturation and training for data discipline.
Assesses Cognite readiness across affiliate maturity levels.

Pain points

Complete DMP foundations before scaling dashboards.
Clarify ownership by affiliate and domain.
Use quality tooling before executive reporting.
Clean migrations and improve cross-platform interoperability.

Implications

Semantic value will scale faster when source ownership, quality, and affiliate readiness are explicit.

Stakeholder profile

Elie Maze

Data Science SME, RAG / AI, DataTech

Signal

AI teams need domain ontologies as queryable services. The practical gap is access to semantic assets through usable APIs.

Main processes

Builds NLP and LLM pipelines for unstructured subsurface data.
Runs automated workflows on internal GPU infrastructure.
Measures quality on labeled datasets.
Collaborates with Smart Search and AI teams.

Pain points

Expose domain ontologies through APIs.
Use formal definitions to accelerate prompt design and QA.
Fine-tune domain models with structured semantic input.
Embed semantic knowledge in copilots and skills.

Implications

TSF should be accessible as a service so AI and modeling teams can consume governed meaning directly.

Stakeholder profile

Pierre Jallais

SSE Smart Search / GenAI RAG, DDT

Signal

Search quality depends on domain vocabulary, metadata, and discoverable semantic resources, especially after the initial GenAI excitement fades.

Main processes

Operates enterprise search over large document collections.
Works on RAG and agentic search architectures.
Uses controlled vocabularies and taxonomies for domain jargon.
Exposes search capabilities through API and MCP patterns.

Pain points

Centralize discoverable semantic resources.
Sustain controlled vocabulary investment.
Improve metadata quality at document source.
Develop multimodal search capabilities.

Implications

RAG and SKG are complementary: RAG retrieves; semantic resources improve domain grounding and relevance.

Stakeholder profile

Julien Campan

LifeX Asset Manager, EP Angola, Block 17

Signal

LifeX is the clearest operational proof point: a funded, high-value, weekly decision cycle with fragmented evidence and direct CAPEX / OPEX exposure.

Main processes

Manages the LifeX programme for Angola FPSOs.
Updates investment plans based on equipment condition.
Coordinates inspectors and evidence capture.
Produces Investment-ready reporting with CAPEX / OPEX implications.

Pain points

Embed domain experts at process level.
Create a V2 data product connected to the weekly decision cycle.
Capture field evidence through OCR, Mistral agents, and structured knowledge.
Build a blueprint that can scale through the Digital Factory.

Implications

LifeX should become the first semantic data product because it links field observations to asset analogues, risk, cost, and management decisions.

Conclusion

Everyone agrees shared meaning matters. What is missing is the one operational case where its absence made a decision slower, costlier, or wrong.

Twelve interviews across four organizational layers produced a single coherent finding: there is no disagreement on whether shared semantics are valuable. Standards teams, platform architects, AI engineers, and field operations all recognize that absent governed meaning creates friction in decisions, integrations, and reporting. Advocates want to accelerate. Skeptics want evidence. Both groups are asking for the same thing: a concrete operational case where the absence of shared meaning produced a traceable cost. TSF should be communicated as a governed semantic control layer, with LifeX as the first production focal point.

LifeX is that case. Julien Campan described a weekly CAPEX and OPEX decision cycle fed by fragmented field evidence: inspection reports, equipment condition data, and analyst judgment that are never semantically reconciled before they enter the investment plan. The before-and-after story exists in the workflow. A semantic data product that names the gap, quantifies it, and shows how governed meaning changes the quality of the decision is what makes TSF unavoidable rather than aspirational. Platform teams want governed meaning delivered as a service; AI and search workflows are already constrained by vocabulary fragmentation; and operating model ambiguity over who owns which concept blocks adoption more than any technical gap.

Make TSF consumable. Stakeholders need semantic outputs as services, mappings, APIs, and decision-ready assets consumed by existing platforms.
Use LifeX to prove value. LifeX links field evidence to similar equipment, risk, schedule, cost, and investment-ready decisions.
Keep the decision framework practical. Use SKG when mission-critical decisions need memory, rules, interoperability, and auditability. Use simpler tools for simpler questions.

Use these internal stakeholder profiles as an appendix to the executive pack and as evidence for the FAQ, use case showcase, and decision framework. The interviews show who needs proof, who needs integration, and who can sponsor adoption.