Sonata is built around one founder and a working network of specialists assembled per engagement. The structure is deliberate. The kind of work we do is careful, measured, and paid back in months rather than quarters. It does not get better by adding headcount. It gets better by matching the right people to the right problem.
Ingemar started as a developer in 1996. Almost three decades later, he is still building. The first half of his career was spent inside financial markets, leading capital-markets engineering teams at Lehman, Barclays, and Bank of America Merrill Lynch, then running technology as CTO of SunGard Financial Systems. Equity derivatives, structured credit, capital-markets technology: domains where the gap between a demo that works and a system that runs without losing money on a Friday afternoon is enormous, and where the consequences of getting that gap wrong are immediate.
The second half has been in AI. He founded his first AI startup in 2015, eight years before ChatGPT made AI a boardroom conversation, back when the work meant classical machine learning, hand-tuned pipelines, and explaining to investors what a neural network was. The decade since has been a string of founder-and-CTO roles across AI, fintech, and Web3 ventures, leading teams of data scientists, machine-learning engineers, and ML-ops practitioners. The technology has turned over twice in those ten years. The questions that decide whether a project works have not.
The discipline that comes out of financial-markets engineering is the discipline Sonata applies to AI work now: measure before you build, instrument everything, write down what you would do if it broke before you turn it on. It is what was missing from a surprising number of the AI projects the firm gets called in to clean up.
“‘Can AI do this?’ is rarely the right question. The right questions are whether it should, what it would cost, what it would replace, and what would tell you it has stopped working. A surprising number of AI projects skip them.”
Sonata exists because the SMB market is where the gap between what AI could do and what is actually deployed is widest. Businesses with 10 to 200 people, real revenue, real operations. Enterprises have transformation budgets and dedicated teams. Solo operators have ChatGPT and a willingness to experiment. The middle is where careful, measured, accountable work pays back fastest and is least available. Sonata is built to close that gap. The choice to base the firm in Dubai is deliberate, not biographical: MENA is where the underserved demand sits.
Most consultancies grow by hiring. The shape of work changes as headcount rises: more layers, more handoffs, more time spent managing the firm and less spent on the engagement. Sonata is structured differently because the work we do rewards a different structure.
The person you meet in the diagnostic is the person leading the Application Map, present in the build, and accountable for the kill criteria afterward. No junior team running the work while a partner checks in monthly. The principal stays close to the actual problem.
Each engagement draws on a working network of senior practitioners (engineers, domain specialists, AI researchers, designers) selected to match the specific shape of the work. The right combination for an operations-heavy build is different from the right combination for a sales-focused one. We scope the team to the problem.
Where the right answer is a third-party tool we have a relationship with, we use it. Where the right answer is to build, we build. The decision is made on the merits, not on which vendor has paid us. We disclose every commercial relationship that touches an engagement, in writing, at scope.
Every contributor follows the same engagement principles: measure first, ground answers in real evidence, write kill criteria before writing code, keep humans in the loop where judgment matters. The network is not a referral list. It is a working group with shared standards.
The model has a constraint. Engagement capacity is finite. We turn away work that does not fit, or that arrives in months we are already booked. The alternative is to overstaff, dilute the senior bench, and become the kind of firm that bills hours regardless of impact. We have chosen not to be that firm.
When we tell a prospective client we cannot start until the following quarter, we mean it. When we say a problem is outside what we should be hired for, we mean that too. A smaller pipeline of work we are well-suited for beats a fuller one we are not.
The order matters. Sonata is built in Dubai because the MENA region is where SMB-scale AI consultancy is most underserved relative to the appetite for it. Family businesses, regional operators, fast-growing groups across the Gulf, Levant, and North Africa face the same gap between what AI could do and what they have actually deployed. Most of the available consultancy is either enterprise-priced or thin.
We work in English and Arabic. We understand the operational realities of family-owned and family-led businesses, the regulatory variation across jurisdictions, and the particular shape of how decisions get made in this region. Generic Western frameworks are not transplanted; they are translated.
We also take work outside MENA, where the fit is right. The engagement principles are the same whether the client is a European logistics group, a North American specialty manufacturer, or an Asian services firm. Geography is a focus, not a fence.
If you are weighing AI for a business with real customers, the right next step is a structured conversation. We will ask what your operation actually looks like, what the gap is, and whether anything in our catalog fits. If it does not, we will say so.
Start the diagnostic