Last checked: 7 November 2026
In short: An AI operating model is how a business organises people, decisions, budgets and processes to build and run AI. It sets who owns AI work and how it is funded and governed.
Table of Contents
How it works
- Choose a structure. The business decides whether AI skills sit in one central team, in each department, or a mix of both.
- Set decision rights. It agrees who approves projects, budgets and risks, and who owns the results once a tool is live.
- Build shared services. Common data, tools, rules and training are set up once and used by every team.
Where you’ll meet it
Consulting reports and organisation charts use the term. It shows up when firms appoint a chief AI officer or merge teams.
What it means for your business
- Decide central or local. Small firms usually do best with one small central team that serves everyone.
- Put owners in the business. Each AI tool needs a manager in the team that uses it, as well as technical support.
- Revisit the structure yearly. As AI spreads, skills often move from a central team into every department.
Watch out for
- Two bosses. Staff who report to both a central AI team and a department can get mixed orders.
- Structure before value. Reorganising before any AI tool works adds cost with no gain.
Example
Moderna. In November 2024 Moderna gave its human resources chief, Tracey Franklin, charge of its digital technology team too, with the title chief people and digital technology officer. In May 2025 she told the Wall Street Journal that the merged function plans work by deciding which tasks people should do and which AI should automate. Roles were being created, cut and redesigned as a result.
AI operating model vs AI centre of excellence
| AI operating model | AI centre of excellence | |
|---|---|---|
| What it is | The whole set-up for running AI across the business | A central team of AI experts |
| Covers | Structure, decision rights, funding, rules and skills | Standards, expert support and shared tools |
| How they relate | May include a centre of excellence | One part of an operating model |
Frequently asked questions
Does a small business need an AI operating model?
Yes, in a light form. Name who approves AI tools, who owns each one, how spending is agreed and which rules staff follow. One page is enough.
Should AI sit in IT or in the business?
Both. IT runs the systems and security, while business teams own the tasks, the results and any staff changes. Leaving AI to one side alone usually fails.
When should the operating model change?
When AI use spreads beyond a few teams. A central team that suited early projects becomes a bottleneck, so skills and decisions move closer to each department.