AI Is Reshaping Jobs. Who in Your Company Owns How People and AI Work Together?

This week, the software company SUSE said it is cutting about 10% of its workforce and pointed to AI and automation as a major reason. It won't be the last. Many of the CEOs I talk with are hearing the same question from their boards: if AI makes us more productive, why is headcount flat?

Before you answer that with a reduction, look at two pieces of research released in the last two weeks.

Adoption Isn't the Problem. Ownership Is.

• •Almost everyone has AI. Few get real value. Lever Talent surveyed 484 leaders at companies with 100–999 employees. 97% have adopted AI, but only 28% call it extremely valuable.
• •Nearly half have no owner. 45% have no single person who owns how people and AI work together. Only 21% have someone whose main job is that work.
• •Ownership tracks with results. Companies with a clear owner were 5.5 times as likely to report extreme value as companies with none. Companies that invested only in technology got there 6% of the time. Companies that invested in technology, programs and people got there 50% of the time. That's a correlation, not proof, but the pattern is hard to ignore.
• •Cutting first is getting expensive. Gartner now predicts that by 2029, about 30% of employees replaced by AI will be rehired, often at a higher cost. Gartner's diagnosis is that companies decide how many people to cut before they understand what work needs to get done.

Why Nobody Owns It

I see the same pattern in most mid-market companies:

• •IT owns the tools.
• •HR owns the job descriptions.
• •Department heads own the results.
• •Nobody owns the work design in between: which tasks AI does, which tasks people do, and who checks AI's output.

So employees work it out on their own. Some lean in. Some quietly avoid AI. Some paste customer data into tools nobody approved. RSM found that 85% of middle-market companies say their executives are more enthusiastic about AI than their employees are. That gap won't close by itself.

What the Owner Actually Does

You don't need a new C-suite title. You need one accountable leader with a clear mandate. That can be your COO, your CIO, or a fractional executive. The job:

• •Map the work, not the jobs. Break key roles into tasks. Mark each task: AI does it, AI assists, or people only.
• •Name who checks AI. Every AI-assisted process needs a named person responsible for catching errors. In the Lever data, this alone made a large difference.
• •Set the reinvestment rule. Decide up front where freed-up hours go: growth, service levels, backlog, or a smaller team over time through attrition.
• •Measure it. Record cycle time, error rates and cost per transaction before and after.
• •Own the communication. Tell people what's changing and what isn't. If you stay silent, people fill the gap with the worst assumption.

Cut Last, Not First

I've led workforce integration at scale, including bringing an acquired offshore engineering organization into a larger company. That lesson applies directly to AI: if you change the structure before you understand the work, you lose the people who hold the knowledge, and then you pay to get it back.

A better sequence:

1. Redesign the work for two or three processes.
2. Run the new design for 60–90 days and measure the results.
3. Let attrition and redeployment absorb the capacity you freed up.
4. Make structural decisions only after that, based on your own data instead of someone else's headline.

This is where operational transformation and change management meet. Fix the process first, then make one person clearly responsible for how people and AI split the work.

What to Do Next: A Checklist

• ☐Name one leader accountable for how people and AI work together.
• ☐Pick two or three high-volume processes to start with.
• ☐Break the key roles in those processes into tasks and mark each AI / assist / people only.
• ☐Assign a named reviewer for every AI-assisted output.
• ☐Record baseline metrics before you change anything.
• ☐Write down where freed-up capacity will go.
• ☐Brief employees early, before rumors start.
• ☐Review results at 90 days, before any headcount decision.

Frequently Asked Questions

Related Reading