AI changes tasks long before it changes jobs. Almost every role in your team contains some tasks AI can now help with, some it can largely take over, and many it cannot touch. The managers who handle this well do not ask "which jobs will disappear?" They map tasks, decide deliberately what changes, and redesign roles around the time that is freed up.
Why job titles are the wrong unit
Headlines talk about whole occupations being "at risk". Inside a real team, that framing is unhelpful. A marketing coordinator, a finance analyst and an operations manager each do dozens of different tasks. AI might draft half of the coordinator's routine emails, but it will not manage the supplier relationship or decide which campaign to cut. Looking at roles as bundles of tasks gives you something you can actually act on.
The task-mapping exercise
Take one role and list its main recurring tasks, roughly ten to fifteen. Then sort each task into one of three groups:
| Group | What it means | Typical examples |
|---|---|---|
| Assist | AI makes the person faster or better, but they stay in charge | Drafting reports, researching, summarising meetings |
| Automate | AI can do most of it, with a human checking | Formatting documents, first-line email triage, standard replies |
| Human | Judgement, relationships, accountability or physical presence dominate | Negotiation, coaching, handling complaints, final decisions |
Do this with the people in the role. They know the tasks better than you, and involving them turns a threatening exercise into a shared design job.
Which roles typically shift most
Roles shift most when a large share of their time sits in "assist" and "automate" tasks: work that is text-heavy, repetitive and rules-based. In many teams that includes administrative coordination, first-draft content work, routine reporting and standard customer correspondence. Roles built mainly on relationships, judgement and accountability shift least, although they still gain from AI assistance.
That said, the impact varies hugely between organisations. A finance analyst in a firm that already automated reporting will see less change than one still building spreadsheets by hand. Map your own team rather than relying on general lists.
What to do with the freed-up time
This is the leadership decision that matters most. If AI saves a coordinator eight hours a week, you can:
- Redirect the time to higher-value work they currently never get to.
- Expand the role, for example adding analysis or customer contact.
- Absorb growth without hiring, so the team handles more volume.
- Reduce hours or roles, which is a legitimate choice but has consequences for trust.
Whatever you choose, say it openly. Teams that do not know what will happen with saved time tend to assume the worst and quietly resist.
Talking to your team about it
Be honest that tasks will change, specific about which ones, and clear about what you intend to do with the time saved. Offer training before changes land, not after. Our article on whether AI replaces jobs or tasks explores the wider evidence, and how to use AI without being replaced is a useful read to share with your team.
Leading this redesign is a core management skill now. Our Leading AI Innovation & Organisational Transformation course includes mapping your own team's tasks, finding where AI helps, and talking honestly with people about what it means for their roles.
Frequently asked questions
Should employees be involved in deciding which tasks AI takes on?
Yes. They know the work best, they spot risks you will miss, and involvement is the single best way to reduce resistance.
Will AI reduce the number of people we need?
It depends on your choices and your growth. Many teams use AI to absorb more work with the same people rather than to cut roles. Decide deliberately and communicate clearly.
How often should we redo the task map?
Every six to twelve months, because AI capabilities change quickly and tasks that were "human" last year may now be "assist".