What happens without an owner
Every department has AI, and nobody owns it. Sales bought a website bot, accounting a subscription for document recognition, marketing a neural network for texts. Each pays for its own tool, and nobody knows the total.
Costs grow quietly: one subscription and one contractor at a time. This is the first reason in Gartner’s forecast.
The value stays unclear because nobody set a metric. Each department judges its own service, usually with the word “convenient”.
Nobody controls the risks: which information went into which services, who has access to the CRM, what the agent may promise a client. When something goes wrong, the whole project is closed, even if it had useful parts.
Another picture: the pilot went well and the report was delivered, but nobody is there to take the solution to other departments. The contractor left, the department went back to the old way of working, and six months later the company launches the same pilot again.
BCG links about 70% of AI adoption difficulties to people and processes, 20% to technology and only 10% to algorithms. The person responsible for the result works on exactly those 70% that no model will solve.
Why not the head of IT
The IT department makes sure systems run without failures. AI changes how sales, procurement, accounting and production work. The head of IT cannot decide how the sales team’s work should change: he has neither the knowledge nor the authority.
The IT department is still needed: it connects AI to the CRM, the accounting system and the phone system, grants access and watches security. But it does not decide what to change and why.
There is also a conflict of interest. The IT department is judged by system stability, while AI adds changes and risks in the first weeks. It is hard to answer both for results and for zero failures.
The head of IT can own AI if he has the time and the authority to change the processes of departments. More often he is busy with support, and questions about results stay without an owner. More on what the IT department lacks is in a separate analysis.
Who should own it
The company’s CEO makes the decision. The CEO sets a goal in money and appoints the person who answers for the result in all departments at once.
The owner answers not for the technology but for a number: how much the company saved or earned with each solution. Technical questions go to the IT department and contractors.
In a mid-size or large company this is an AI department: people and rules that tie the departments into one process. The AI director answers for the solution map and the result, department heads for their numbers, the IT department for connections and access, and lawyers and security for personal data, 152-FZ (Russia’s personal data law) and vendor contracts.
In a small business, the AI department is one person in charge plus an AI director for a few hours a week. No separate unit is needed.
The owner must have authority: to switch off a service that gives no numbers and to stop a contractor who misses a deadline. Without authority, the owner becomes the project’s secretary.
The owner has a list of all AI services and agents with their access rights, a metric for each, rules for client information and a plan for when a service becomes unavailable. What the law and the risks require of a company is covered on our page on who owns AI in a company.
Once a week the owner looks at the metrics and decides: what works, what to switch off, what to roll out next. If it does not work, we switch it off. Without this rhythm, a project lives until its first mistake.
A fractional AI director
A full-time AI director is rarely affordable for a mid-size business. The option for such companies: an outside AI director, 5 hours a week, from $999 a month, with no new hire.
He runs AI the way an in-house head would: builds the solution map, chooses contractors, tracks metrics and risks, and reports to the CEO.
He does not replace department heads. Each of them stays the owner of their process and their number, and the director ties them into one plan.
The IT department and the lawyer stay in their places. The director gives them AI tasks and puts their parts together into one whole.
If you already have AI projects under way, the director starts with a check: what works, what you pay for out of habit and what is missing for a result. Terms and price are on the AI director page.
You can start with a 30-minute review: we look at which AI services the company already has, who pays for them and who answers for the result.