Naijm: How Oussama Ammar built a Dubai AI company that only earns when the client saves

Most enterprise software is sold the same way everywhere, and the Gulf is no exception. A licence is signed, an integration is invoiced, a pilot is celebrated, and the question of whether anything actually got cheaper is settled much later, by the buyer alone. A Dubai company is selling the opposite arrangement.

                                                 Photo: STEVE, CC BY 3.0, via Wikimedia Commons

It is called Naijm. Its founder, the entrepreneur Oussama Ammar, announced publicly that he had raised three and a half million dollars to build it, backed by the business community in the Emirates, and said he would lead it himself. His summary of the deal leaves little room for interpretation: if the client does not win, the company does not eat.

One business, stated in three lines

The company’s own front page refuses to decorate the proposition. Inside every large organisation, one process consumes more money than any other. Claims. Underwriting. Collections. Customer operations. Naijm takes that process inside a market leader, rebuilds it as a system, and runs that system itself. It is then paid a share of what the rebuilt process saves. That, they write, is the whole business.

The contract therefore has two components. A base fee for doing the work properly, and a share of the savings demonstrated against the client’s own prior numbers. The firm is explicit about where the money actually sits: most of it is in the share.

Learning the job before writing a line of code

The method follows from the pricing. If payment depends on a measurable reduction, the work cannot begin with a slide deck. It begins by sitting with the people who perform the task until the team can perform it themselves.

What they are looking for is listed plainly: the exceptions, the workarounds, the unwritten rules, the forms nobody reads. Anyone who has worked inside a large operation recognises the list immediately. These are the things that never appear in a process manual and that quietly determine whether the operation functions at all.

Only afterwards is the process rewritten as software, and then operated inside the client’s own walls for as long as it continues to save money. One technical commitment is stated as a condition rather than a feature: the data never leaves.

The second principle says the same thing in four words. Logins, not decks. And the request is concrete: credentials, a desk on the floor, and a seat at the morning meeting where the work is handed out. It is an unusual thing to ask of a large company, and an unusual thing for a large company to grant.

Long contracts, and one machine per client

The third principle is to build and then operate. The team stays to run the system in production, and the contract is long by design, because savings are measured across years rather than across a demonstration.

The fourth is the most commercially counterintuitive. Each system is built on a single operation’s own data and exceptions, which makes it, by the company’s own admission, useless anywhere else. That is the exact opposite of the software publishing model, whose entire economics rest on selling the same product repeatedly.

The client list is accordingly short and openly described as such. A handful of market leaders, in construction, retail and education, who opened their operations: credentials, floor access, and their own baseline figures.

The number, and how it is framed

On an operation representing one hundred million dollars in costs, the company displays a range of twenty to thirty per cent in savings. It takes care to label this its working thesis rather than an achieved result, and that is how it should be read.

Framed honestly, the figure is still useful. If the main revenue is a percentage of that reduction, anyone can work out what the provider stands to earn, and the client knows precisely which number the invoice will be built on.

The same rule applied to salaries

The most revealing detail sits in the terms offered to the company’s own hires rather than to its clients. Naijm is recruiting ten people in Dubai across four roles, and publishes its compensation without hidden bands.

The bonus is described as transparent and uncapped, calculated on the same savings figure the client pays on. Their phrasing is exact: you and the client win on the same line of the invoice.

So the provider, its employee and the buying organisation all watch a single number. That is the portable idea here, and it requires no artificial intelligence project at all. Arrange matters so that what enriches the supplier is precisely what enriches the customer.

What the model asks of the buyer

There is a difficulty in all this, and it falls on the client rather than the vendor. Agreeing to pay on results requires an undisputed starting figure. For a great many organisations, producing that baseline is already the hardest part of the project.