Where AI shouldn't go in your business

Most AI advice tells you where to add it. The harder question is where to keep it out. Fatima Ait Moulid Deeney on the five places AI does damage.

By Fatima Ait Moulid Deeney, 2026-11-12

Keep AI out of any process that is broken and any decision that nobody owns. Put it there and the problem simply runs faster. Everywhere else, the real question is where AI sits in the way the business is organised.

Most advice on AI is a list of places to add it. Very little of it asks where to keep it out.

The pattern we keep seeing

For more than two centuries, companies have been organised around one problem: how to put divided human labour back together. The department was a solution to the cost of coordinating people. Each department now buys its own AI. Marketing gets a content tool, finance a forecasting model.

Each purchase makes sense on its own. Together, they often produce a more efficient version of fragmentation. The handoffs between departments stay exactly where they were. They simply move faster, and so do the errors that travel through them.

HubSpot's 2026 research found that 98% of marketing teams now use AI in some way, and only 10% report transformational results. The gap is organisational.

Most companies are using AI to make fragmentation more efficient.

Five places to keep AI out, for now

The market is already drawing these lines

This is no longer theory. In November 2026, United Airlines said it would block personal AI agents from booking flights, citing customer privacy and regulation. Delta and eBay have restricted outside agents too. Apple has reportedly capped its engineers' daily AI spending at $200. Each of these companies is deciding where AI belongs.

Where AI should go

AI earns its place where someone owns the outcome and can check it. It earns the most where it removes a handoff altogether. That is the difference between AI as a tool and AI as part of the structure of the company.

At AIT Advisory, our business transformation work starts with a diagnosis of exactly this: where momentum is lost, and which of those losses AI would fix or merely hide.

Frequently asked questions

Where should a company not use AI?

Keep it out of broken processes and out of decisions nobody owns. Be careful, too, with the routine work that trains junior staff.

Why do so many AI projects fail to deliver results?

Usually because AI is added one department at a time. Each tool improves its own function while the handoffs between functions stay unchanged, so the parts speed up while the whole stays slow.

How do I decide where AI belongs in my business?

Start by mapping where work slows down, especially between departments. Remove what should not exist, then place AI where someone owns the result and can check it.

Fatima Ait Moulid Deeney is a transformation expert at AIT Advisory in Dubai and the author of After Human Coordination.

Press & Media Enquiries

For press enquiries, interviews, media opportunities and expert commentary, please contact hello@aitadvisory.com.