Service line — Data & AI
Turn Data into Decisions.
Turn AI into Business Value.
We help organisations turn their data into a reliable asset and put artificial intelligence to work — from strategy through execution, so that Data and AI ambitions become measurable, durable results.
Speak with a partner
The problem is not the technology
The gap between experimentation and production is now the single largest source of value destruction in AI programmes. Studies converge on the scale of it, even where their methods differ.
The reported causes are not algorithmic: unprepared data, success criteria never defined, and change-management debt. A successful prototype then operated as if it were a system is the most expensive form of this failure.
Where we work
Four domains on data, two on artificial intelligence. They are not ordered separately: AI built on ungoverned data produces decisions that are fast and wrong.
- 01
Data Governance
The foundation of trust, before anything else.
Roles and accountability over data, quality, lifecycle, compliance. Without governance, a modern platform merely industrialises distrust — distributing, faster, figures nobody believes.
- 02
Data Platforms
An architecture that carries today's load and tomorrow's uses.
Design and delivery of scalable data platforms, with architecture, cost and sovereignty trade-offs made explicit — not inherited from a vendor's catalogue.
- 03
Data Management
Flows that are controlled, normalised, automated.
Integration, transformation and reliability from source to point of use. The goal is not the pipeline: it is that a business director stops rebuilding the figures in a parallel spreadsheet.
- 04
Data Analytics
Data made decidable.
Reporting, steering and analysis in service of the decision. An indicator that changes no decision is a cost, not an asset.
- 05
Artificial Intelligence
From experiment to industrialised use.
Identifying use cases with demonstrable value, measured prototyping, governed deployment, and adoption. The difficulty is never the model — it is the business process that has to be transformed end to end.
- 06
Advanced Analytics
Anticipate rather than observe.
Predictive modelling and advanced analysis applied to the sector's real problems: demand forecasting, risk, maintenance, operational performance.
A clarification
"Agentic": what the word means, and what it commits you to
The term is used everywhere and rarely defined. The distinction is not commercial but operational — and it changes the nature of the risk.
Generative AI
A reactive system. It produces content — text, image, code — in response to a prompt. It pursues no objective of its own and does not act on your systems.
Risk: informational. Factual error, bias, misleading phrasing. It is managed through review and editorial control.
Agentic AI
A system given an objective that determines the steps itself: it plans, calls tools, observes the result and adjusts. The generative model is only one component of a larger loop.
Risk: operational. The system acts on live systems. It is managed through human-in-the-loop thresholds, provenance logging and strict tool-access control — designed in from the start, never added afterwards.
This difference in kind explains why scaling rates remain low: a March 2026 survey of 650 enterprise technology leaders found that only 14% had deployed an agent across their organisation. This is not a model problem — it is a problem of governance, observability and integration.
Our method
Launch · Accelerate · Steer
Three movements, applied equally to a data programme and to putting AI into production. Each transition is an explicit decision, not a drift.
- I
Launch
Frame the use cases, decide what deserves to exist, set the target architecture and the compliance frame. You leave this phase with an investment decision defensible before a committee, not with a list of ideas.
- II
Accelerate
Build and deploy. A prototype measured on real data, then industrialisation: security, monitoring, recovery, documentation. Going to production is an explicit decision against criteria set in advance.
- III
Steer
Keep it alive: adoption, team capability, impact measurement, governance over time. This is the phase most programmes omit, and the one that decides whether the investment produces anything.
The Moroccan context
Sovereignty and compliance are not optional
Morocco does not yet have legislation specific to artificial intelligence, unlike the European Union and its AI Act. The CNDP, the supervisory authority under law 09-08, has begun work on regulating AI drawing on the European framework; sectoral recommendations are expected.
In practice: designing today for the current state of the law alone means rebuilding tomorrow. We set traceability, transparency and non-discrimination requirements at design time — because they will be required, and because they are already conditions of trust.
The infrastructure is moving fast. The April 2026 announcement of the Nexus AI Factory at Nouaceur — 12 billion dirhams, led by Nexus Core Systems with the Ministry of Digital Transition — changes the hosting and sovereignty trade-offs that applied until now.
The national strategy rests on three pillars: infrastructure (data centres, sovereign cloud, local compute), data governance (CNDP framework, alignment with 09-08, a responsible-AI reference) and human capital. Those three pillars are precisely where a data programme fails or holds.
Let's discuss your use cases
A first conversation with a partner, without commitment, to qualify what deserves to be launched — and what does not.
Speak with a partner