Practical lessons from 18 years working with Google, now applied to enterprise AI delivery in Saudi Arabia and Qatar.
Across Saudi Arabia and Qatar, enterprise AI has moved from ambition to mandate. National strategies have set the direction, budgets are approved, and every leadership team has an AI roadmap. Yet ask the uncomfortable question: how much of it is live, in production, creating value? In most organisations, the honest answer is not enough.
This isn’t a Gulf problem. It’s an enterprise problem. But in KSA and Qatar, where Vision 2030 timelines compress everything, the cost of stalled AI is higher, and so is the reward for getting it right.
After 18 years partnering with Google and delivering transformation programmes for global brands, we’ve seen the same four failure patterns repeat, regardless of sector or geography.
The four reasons AI stalls before production
1. Strategy stalls in a deck. Large strategy engagements produce recommendations that are hard to build against: no clear architecture, no prioritised use case, no executable roadmap. The organisation has a vision statement where it needs a delivery sequence.
2. Prototypes never reach the tenant. AI gets evaluated in a lab environment, disconnected from live systems, real workflows and enterprise data. The demo impresses; the deployment never happens, because the demo was never designed to survive contact with production.
3. Governance isn’t ready for the board. This one matters enormously in the Gulf. Without an audit trail, a citation model and a defensible risk framework, AI initiatives cannot clear the AI committee, the CISO or the DPO. In markets where data sovereignty and regulatory trust are board-level issues, nor should they. Governance built in afterward is governance that arrives too late.
4. Ownership becomes fragmented. Oracle, the cloud provider, the AI vendor, internal teams and external consultants are all involved, yet no single partner is accountable for delivery. When everything is everyone’s job, production is no one’s.
What closing the gap actually takes
The answer isn’t a bigger strategy engagement. It’s a delivery model where every layer has a clear owner and the whole thing is accountable to one partner. We structure it in four layers:
- Record: Oracle. Your core transaction data acts as the system of record.
- Refine: Google Cloud. BigQuery and Vertex AI turn raw records into governed, model-ready data.
- Reason: Anthropic. Claude agents reason over refined data, with citation, safety and audit built in.
- Deliver: Chesamel. One accountable partner wraps all three layers, signable on a single PO.
We call it the Triple Stack. The principle behind it is simple: AI creates value when it runs where your data lives. Inside your tenant, against your real workflows, with governance designed from day one. Not in a disconnected demo environment.
Why this fits KSA and Qatar right now
Three reasons this model lands particularly well in the Gulf:
Sovereignty and governance are non-negotiable. Public-sector and regulated organisations in both markets need AI that is audit-logged, citation-linked and defensible in front of governance bodies. That has to be an architectural property, not a policy document.
Speed is expected. Vision 2030 programmes don’t wait for 12-month strategy phases. An Exploratory Session that maps the path to a costed, buildable plan (architecture, prioritised use cases and a 12-month roadmap) matches the pace these markets actually move at.
Proof before commitment. Exploratory session first; no subscription until the plan is proven. In markets where every vendor is making big AI promises, starting small and proving value in-tenant is how trust gets built.
With teams on the ground in Riyadh and Doha, alongside London and San Francisco, this isn’t a fly-in model. It’s local delivery with global standards.
The question worth asking your team
Not “do we have an AI strategy?” You already do. The sharper question is:
How many of our AI use cases are live in production, inside our own tenant, with governance our board would defend?
If the answer makes you pause, the gap between your AI plan and AI in production is exactly the problem worth solving first. And it can be scoped in ten days.
📩 Our team is ready to chat through where AI could take you.
