AI and your operating model: who owns what in your team
By Keith Hamilton · · 4 min read
An AI operating model sets out how work is organised around AI in your business and who is responsible for what. AI projects in small and medium businesses stall when nobody owns them, nobody checks the results, or nobody has explained how people's jobs will change, and a clear AI operating model prevents all three. This guide sets out the roles you need and how to bring your team with you.
Why does AI ownership in a business matter more than the technology?
Without a named owner, a working tool still fails. When a new tool arrives with no clear owner, the same problems appear:
- Staff are unsure whether to trust its outputs.
- Small issues go unreported.
- The supplier is managed by whoever happens to answer the phone.
- After a few months, use drifts and nobody can say whether it is working.
Clear roles fix this. They do not need new job titles or a new department. In a smaller business, one person holds several of these roles at once; what matters is that each responsibility has a name against it.
What AI roles and responsibilities does an SME need?
An SME needs six AI roles and responsibilities covered: an accountable lead, process owners, output reviewers, a data protection lead, a supplier owner and team champions.
An accountable lead
One senior person owns the business's use of AI overall. They agree priorities, hold the budget, report to the board or owners and make the final call when trade-offs come up.
In some businesses this is the managing director. In others it is a fractional leader, as described in our article on whether you need a fractional Chief AI Officer.
A process owner for each piece of work
The person who runs the process being changed owns the AI part of it too. If AI is helping with invoices, the finance lead owns it. They decide how it should work, agree what success looks like and decide when it is ready to extend.
People who review outputs
Named people check what the AI produces, especially in the early weeks and for anything that affects customers, money or personal data. Give them time to do it and tell them what to look for.
A data protection lead
One person knows what personal data each tool touches, why, and on what lawful basis under UK GDPR. In a small business this is normally the person who already handles data protection.
A supplier owner
Each AI supplier has one named person who manages the relationship: renewals, changes, problems and performance against what was promised.
Champions in each team
A few interested people in each team help colleagues, collect feedback and spot new uses. They are the best source of ideas for the next piece of work, because they see the day-to-day tasks.
How does AI change staff roles: tasks or whole jobs?
AI takes over parts of a job, not whole jobs. A customer service role loses some of its routine queries but keeps the complex, sensitive and judgement-based work. An administrator stops keying in data but spends more time on exceptions and checks.
So when you plan a change, map the tasks within each affected role. Note which tasks AI will take on, which stay with people, and which are new, such as checking outputs or handling escalations.
This gives you an honest picture of what changes for each person, and shows where released time can be put to better use. Our guide to blending AI and people in a contact centre shows how this division of work plays out with customer calls.
Where is human review of AI required?
Human review of AI is required wherever a decision has real consequences for people, money or compliance. Write down, for each use of AI, where a person must review or approve before anything happens. Typical examples include:
- decisions that significantly affect a customer or member of staff
- complaints, vulnerable customers and sensitive conversations
- payments, refunds and credit decisions above a set level
- anything regulated in your sector
These rules belong in your AI policy, so staff know them and follow them consistently.
How do you bring your team with you?
Explain early what is changing, why, and what it means for each role. People worry when AI arrives, especially about their jobs, and silence makes that worse. Be honest about what you do not yet know.
Involve the people who do the work in designing the change. They know where the exceptions and awkward cases are, and they spot problems a supplier will miss.
Give them proper training on the tools they will use, and a simple way to report anything that looks wrong.
How often should you review your AI operating model?
Review the roles every three months, alongside your results. Your operating model changes as AI takes on more work.
Check that owners still have time for their responsibilities, that output checks are still happening, and that released time is going somewhere useful.
Where Forwardcycle fits
Our commercial leadership service provides an accountable lead a few days a month, and includes coaching your team to run improvements themselves. To see where AI could help first, take the free AI Readiness Assessment.