A.Issue No. 1
Decision-making is scattered
Every team has its own dashboard, its own spreadsheet, its own version of the truth. Executive committees arbitrate on numbers they have no time to reconstruct.
§ ACT 01
The Valley
Above the fog,the numbersregain clarity.
AI is only worth what its foundation is worth. In most organizations, decision-making is scattered, data is contested, and choices are made blind. Our job: clear the ground before building the summit.¹
BCG, October 2024 - 74% of companies struggle to scale AI and capture tangible value (study "Where's the Value in AI?", 1,000 executives, 59 countries).
Photo: Sasha Kaunas · Unsplash · Racha 2021
§ ACTE 02
The crossing
Before the tool, it's the decision-making that breaks down. Data is scattered, indicators are debated, choices are made blind. Layering AI on top of a fuzzy system produces noise, not value.
B.Issue No. 2
Data is no longer under control
Tools have piled up. Nobody can say where the reference data lives, who produces it, or how it is refreshed.
C.Issue No. 3
AI promises floating above reality
Demos look great, but real usage doesn't follow. We deploy assistants that talk a lot and decide little.
« Before the tool, it's the decision-making that breaks down. »
McKinsey, State of AI 2024- More than 80% of organizations surveyed see no tangible GenAI impact on their EBIT. Fewer than one in five tracks dedicated KPIs for its use cases.
Gartner, July 2024- At least 30% of GenAI projects will be abandoned after proof of concept by end of 2025 - blurry data, imprecise ROI, ill-scoped risk.
§ ACTE 03
First landmarks
We turn executive intuition into systems that measure, decide, and act. First by making decision-making clear. Then by making data rigorous. Last, and only then, by deploying AI that understands its domain - and that you understand back.
§ 01
Decision-making
Clarify what matters. Rationalize indicators. Make performance readable for leadership.
§ 02
Data
Consolidate sources. Make reference data trustworthy. Build a foundation that AI and teams can actually use.
§ 03
Useful AI
Design targeted automation and copilots. Measure value. Hold sovereignty - over the models, over the data.
§ ACTE 04
Rope team partners
We work in direct coordination with leadership. No needless layers, no jargon - just the right question asked at the right moment.
SMB & mid-market leaders
Clarify decision-making, accelerate choices, arbitrate on solid numbers.
Finance leadership
Make reporting trustworthy, industrialize forecasting, lower the consolidation burden.
Transformation leadership
Frame the AI & data trajectory, avoid endless projects, hold commitments.
Operations & data leadership
Rationalize tooling, structure data, equip teams without piling on yet another layer.
Manifeste · Intellencia
« Approaching the summit. The ground before the ambition. »
- Act IV → V · The approach
§ YOU MADE IT - SUMMIT MANIFESTO
We've arrived.
Here is what we stand for.
We name what holds you back. Not what flatters.
One number per indicator. Always fresh, always defensible.
AI that serves. Not AI that rules.
- Intellencia · Code of conduct
§ ACTE 05
The Summit
Every engagement begins with understanding before building. We don't take on work we haven't first proven creates value.
See clearly before you invest.
Decision-making & AI opportunity diagnostic
A short, dense audit to identify friction in decision-making, map the data that matters, and prioritize the automation that will actually create value.
Value curve
Deliverables
- Map of usage and data
- Impact / effort matrix of opportunities
- 6-18 month roadmap
- Executive readout
Timeline
Details 2 to 4 weeks
Trustworthy data, readable indicators.
Decision-making foundation
We consolidate sources, rationalize indicators, and build a robust decision-making foundation - dashboards, reports, and reference data that leadership and operations finally share.
Value curve
Deliverables
- Target data architecture
- Reference data and key indicators
- Executive dashboards
- Trustworthy refresh processes
Timeline
Details 2 to 6 months
AI that decides, not AI that decorates.
Automation & sovereign AI
We design and deploy domain automation and copilots on a controlled foundation: bounded scope, data sovereignty, measured value, room to grow.
Value curve
Deliverables
- Scoped & costed use cases
- Domain copilots and agents
- Document automation
- AI governance & value measurement
Timeline
Details 3 to 9 months
§ ACTE 06
The expedition method
We refuse engagements that start with the tool. Every intervention follows a proven discipline - a short, dense, measurable continuum.
- 01
Listen to leadership
Understand the real stakes, constraints, deadlines. Name what counts - and what doesn't.
- 02
Map what exists
Draw an honest picture of tools, data, and processes. Spot the blind spots and the redundancies.
- 03
Prioritize value
Sort opportunities by impact and feasibility. Refuse the seductive but weak projects.
- 04
Build useful systems
Deliver in short increments, each with a clear measure of value. No over-engineering.
- 05
Hand over mastery
Document, train, make teams autonomous. We exit when the teams are steering on their own.
Manifeste · Intellencia
« Above the fog, clarity returns. »
- Act VI → VII · Cases in production
§ ACTE 07
In production
Three real engagements in production. A recent sovereign AI, a corporate BI mission inside a public operator, and an NLP module running for five years. Consumer AI, data engineering, proprietary R&D - the full triptych.
Case 01 · Tourism · Regional network
A sovereign digital concierge for regional tourism.
Client · Club Hôtes Vienne (a French regional Airbnb hosts network)
Context - The hosts network wanted a hyperlocal conversational agent able to inform visitors in French, without depending on a public model, and GDPR-compliant.
Intervention - Design of the territorial knowledge base, selection of a France-hosted model, RAG on institutional sources, configurable conversational interface, GDPR / AI Act compliant deployment. We also built an Analytics Dashboard to track usage: volume of questions, recurring topics, geographic zones, resolution rate - essential instrumentation to keep the knowledge base alive.
Results - Concierge in production since June 2026, hyperlocal, customizable per host or territory, embeddable in 30 seconds. Full data sovereignty. Analytics dashboard live to steer content over time.
High-pressure gas transport operator · FranceCase 02 · BI Mission · Public energy operator
A Data governance across SAP S/4 and SAC Planning, at the scale of a national operator.
Client · NaTran · €2B revenue 2024 · French high-pressure gas transport operator
Context - NaTran is the main high-pressure gas transport operator in France. As a public service mission, the company guarantees continuity and safety of gas supply and supports the evolution of the energy system. The business teams (Controlling, Finance) and the technical team (SAC Planning) needed a shared, trustworthy, governed SAP data foundation.
Intervention - Co-creation of the Core Team BI to govern SAP data across Controlling, Finance, and the technical team. Architecture of the S/4 CDS Views and the SAC Planning models (Data Actions, Stories, scripts), SAP audit then remediation plan, Agile facilitation with the Product Owners. A bridge stance between business and technical - governance before development.
Results - Governance installed, business ↔ technical alignment achieved, CDS Views foundation structured to serve both SAC Planning and the DataLab. The kind of hard foundation that makes downstream automation and reliable AI possible.
● In service since 2020
Case 03 · R&D · Vertical web application
A custom NLP tokenizer for proprietary semantic matching.
Client · Web2Vi · Software for construction professionals
Context - A vertical web application needed to associate items by meaning, without depending on public models - full confidentiality constraints on the data processed.
Intervention - Custom NLP tokenizer built from scratch, proprietary semantic matching algorithm, full-stack module integration into the existing application, tests on construction-domain corpus.
Results - Module in production since 2020 - over five years of service. Zero dependency on OpenAI, Anthropic, or any third-party model. Full confidentiality preserved.
§ And upstream
Fifteen years of data architecture in consulting firms, in industry, and inside the finance departments of international groups. That's the rigor that makes AI reliable.
§ ACTE 08
Our voice carries
Thinking useful AI, four years before the hype.
In 2021, Julien Bédouret stepped onto the TEDxBlois stage to defend a simple idea: combine progress, ethics, and timeless humanism. That conviction still sits at the heart of Intellencia. (Talk delivered in French.)
Archive · 14:32 · Public talkYouTube ↗
§ ACTE 09
Rope team commitments
Standards curve
live2010 ---- 2026
§ 01
15+
Years of data architecture
§ 02
100%
Sovereignty of client data
§ 03
48h
Maximum response time
§ 04
0
Commitments without a diagnostic
A
We speak business first.
Before talking technology, we understand how you read performance. Tools come after - that is where we make the difference.
B
We refuse hollow projects.
If a project has no measurable value, we say so. Our credibility is worth more than a contract signed for convenience.
C
We frame before we build.
A short diagnostic saves months of wandering. We start with clarity, and only then with delivery.
D
We ship with discipline.
Short increments, regular reviews, knowledge transfer. The agency leaves when the teams are steering on their own.
E
We defend your sovereignty.
Your data stays your data. Our default architecture: control of the models, control of the hosting, control of the code.
F
We come from data rigor.
BI architecture, data engineering, applied AI - demanding environments that forge discipline. Rigor here is not a posture, it's a habit.
§ ACTE 10
The descent
One hour together is often enough to clarify the trajectory. We will tell you plainly what deserves to be structured - and what can wait.
Office
Poitiers (86)
Working across France
Availability
Diagnostic within 15 days
Poitiers · FR · SIRET 943 852 731 · 2026
Clearer vision. Sharper decisions. Useful automation.
Build 04.06