Most companies in the Gulf now run some form of artificial intelligence: a chatbot on the website, a forecasting model in finance, a team experimenting with generative tools to draft reports. Few of them can say who decided which projects to fund, how the risks were weighed, or what the business expects to gain. AI often enters a company through enthusiasm at the edges, while the executive team watches from a distance.
That arrangement does not last. As AI moves from pilots into core processes, the decisions it raises belong to the people who run the business. They concern money, reputation, customers and staff. Executives do not need to build models. They need to lead four conversations: value, risk, data and people.
Value: choose where AI earns its place
The first executive task is choosing. Every function can produce a list of AI ideas, and many of them will work in a demonstration. The harder question asks which ones will change a number the business cares about.
A useful discipline is to keep an AI portfolio, in the same way a company manages capital projects. For each use case, the sponsoring executive writes down the problem, the metric it should move, the cost, the data it needs and the person accountable for results. The executive team then funds a small number of cases with clear owners and stops the rest.
Good candidates usually share three traits. The task repeats often, so small gains add up. The company holds the data needed to do it well. And a human can check the output before it reaches a customer or a regulator. Contract review, demand forecasting, customer service triage and internal knowledge search often meet all three tests.
Decide in advance what success looks like and when you will judge it. A customer service assistant might aim to resolve a set share of routine questions without human help within six months. A contract review tool might aim to cut first-pass review time in half. Clear targets let the team stop a project that is not working before it absorbs another year of budget.
Risk: decide what can go wrong and who watches it
AI introduces familiar risks in unfamiliar forms. Models can produce confident errors. They can treat groups of customers unfairly if their training data carries old biases. They can leak confidential information through careless prompts, and they can expose the company to intellectual property disputes.
Executives can manage this without slowing everything down. A simple approach sorts AI uses into tiers:
- Low risk: internal productivity tools where a person reviews every output. Light rules and fast approval.
- Medium risk: tools that support decisions about customers, staff or money. A named owner, testing before launch and regular monitoring.
- High risk: systems that make or strongly shape decisions affecting people's rights, credit, employment or safety. Executive approval, documented testing, human oversight and periodic independent review.
Generative tools need a few rules of their own. Staff should know which tools the company has approved, which information they must never enter, and when a person has to check the output before anyone outside the company sees it. A one-page policy that people read beats a long manual that nobody opens.
The board should receive a short report on the high-risk tier at least once a year, in the same way it reviews other material risks.
Data: the foundation and the obligation
AI performs only as well as the data behind it, and much of that data concerns people. In Oman, the Personal Data Protection Law issued by Royal Decree 6/2022 sets the rules, and its executive regulations followed in Ministerial Decision 34/2024. The transition period ended in February 2026, so companies now operate under full enforcement.
For executives, three questions matter most. Does the company know which personal data feeds its AI tools, and on what legal basis it processes that data? Do staff understand what they may paste into external AI services? And when the company buys an AI product, does the contract say where the data goes and who may use it?
Data quality deserves the same attention as compliance. A forecasting model trained on incomplete sales records will mislead the people who trust it. Executives who fund AI should also fund the unglamorous work of cleaning, labeling and governing the data underneath.
People: the change that decides the outcome
The technology usually works. The adoption is where projects stall. Staff fear for their roles, managers do not change their routines, and a promising tool sits unused after the launch event.
Executives shape adoption through what they say and what they do. They can explain plainly which tasks AI will take over and which new tasks people will own. They can fund training that matches each role, from basic prompting for administrative staff to model oversight for analysts. And they can redesign processes so that the tool becomes part of the normal working day rather than an optional extra.
Executives also teach by example. A chief executive who uses an AI assistant to prepare for a meeting, and says so, gives permission to everyone else. A leadership team that never touches the tools sends the opposite message, whatever the official strategy says.
Oman's National Programme for Artificial Intelligence and Advanced Digital Technologies, led by the Ministry of Transport, Communications and Information Technology, links AI adoption to Vision 2040 and to the growth of local capability. Companies that build AI skills among Omani staff contribute directly to that national direction, and they reduce their own dependence on outside expertise.
Questions for the executive team
An executive team can test its own readiness with a short list:
- Which three AI use cases will make the largest measurable difference to our results this year?
- Who owns each one, and how will we know within six months whether it works?
- Which of our AI uses fall into the high-risk tier, and who reviews them?
- What rules do our staff follow when they use external AI tools with company information?
- Does every AI supplier contract cover data location, data use and liability?
- What training does each role need, and who pays for it?
- When did the board last discuss AI as a strategic topic?
If the team struggles to answer, the gap lies in leadership rather than technology.
Leading AI from the top
AI rewards companies whose leaders make clear choices, set sensible limits and bring their people along. It punishes those that let it spread without direction. The difference rarely comes from the size of the technology budget. It comes from executives who understand enough to ask sharp questions and take responsibility for the answers.
The National Academy of Leadership and Innovation created its Enterprise AI Leadership Fellowship for senior leaders who want to lead that work: choosing where AI creates value, governing its risks and data, and preparing their organizations for the change.
