Decide before you deploy

A Hunter BI scoping engagement lasts four to eight weeks depending on the size of the organisation, and produces four deliverables. A portfolio of use cases: identified with the business lines, quantified by impact and effort, prioritised — because the first seam of value is almost never the one the executive committee had imagined. A platform arbitration: OpenAI, Anthropic, a multi-model or sovereign approach, argued on your real use cases and not on generic benchmarks — our membership of both partner networks guarantees that the recommendation follows your interest, not a catalogue. A budget trajectory: licences, engineering, training, with the assumptions made explicit. A governance framework: who decides, who controls, how you measure.

The method is deliberately concrete: interviews with the business departments, workshops on real material — your contracts, your complaints, your reports — and instrumented tests of the platforms on your representative data. The executive committee receives a plan it can weigh line by line, not an expert's conviction.

Three questions come up in almost all our engagements. 'Build or buy?' — the answer depends on the use case: general-purpose assistants are bought (ChatGPT Enterprise, Claude Enterprise), business agents are built on your systems, and the boundary shifts every quarter, which argues for reversible architectures. 'How much, and in which currency?' — licences and inference are billed in foreign currencies, which raises very concrete questions of budgeting in dirhams, foreign-exchange regulation and rate exposure; we build these parameters into the budget trajectory, together with your finance department. 'What is the competition doing?' — we set the record straight: what is genuinely in production at comparable players in the Moroccan market, what is merely communication, and where the window of advantage still stands open in your sector.

A fourth question deserves to be asked even when it is not: 'what fails?'. The failures observed in Morocco as elsewhere have stable causes — licences with no use case, pilots with no metrics, absent governance — and they are avoided at the scoping stage, at a cost that is trivial compared with that of a failed deployment.

The national Digital Morocco 2030 strategy is not merely a public-policy document: it reshapes the competitive environment of every Moroccan company. The digitalisation of public services — hence of your company's interactions with the administration — a massive rise in digital skills — hence a shifting talent market — and stated ambitions in artificial intelligence and offshoring — hence competitors that are gearing up.

We align our clients' AI roadmap with this context: drawing on national training momentum to build the internal teams, anticipating the growing digital expectations of public-sector partners, and positioning the company within the ecosystems that matter — AUSIM for CIOs, APEBI for the technology sector, Gitex Africa for continental visibility. For Moroccan groups present in sub-Saharan Africa, the roadmap takes on an added dimension: Morocco as a regional deployment base for AI, with architectures and governance designed from the outset to be replicated across subsidiaries.

AI governance is not one more document: it is the answer to questions your organisation will ask itself in any case, usually at the worst possible moment. Who authorises a new use case? Which data is allowed into which tool? Who answers to the CNDP in the event of an incident? What do you tell a customer — or a regulator — who asks whether a piece of work was assisted by AI?

We put in place a mechanism proportionate to your size: an enforceable usage charter drafted in French and adaptable into Arabic for field teams, an AI committee with a real mandate — to arbitrate use cases, not to produce minutes — a processing register aligned with Law 09-08, and metrics tracked by the executive committee. For organisations aiming for certification, we structure the mechanism in line with ISO/IEC 42001, the international standard for AI management systems. The goal is not compliance for its own sake: it is to make it possible to move fast safely — organisations that govern well deploy more, not less.