According to BDO's 2026 mid-market research, AI transformation is 70% people. Not 70% technology, not 70% data, not 70% strategy. 70% people.
And yet most AI initiatives are led by the technology team with the people team brought in as an afterthought, usually when someone realises that the workforce is anxious and engagement scores are dropping.
If you're a Chief People Officer or HR Director, you're probably seeing this play out already. The CEO is excited about AI. The CTO is building a roadmap. And your team is fielding questions from employees who want to know if their job is safe.
The fear is real, and ignoring it makes everything harder
AI will change roles. Some tasks that people currently do will be automated. Some roles will evolve significantly. And yes, in some cases, roles will be reduced. Pretending otherwise isn't kind. It's dishonest, and your people know it.
But here's what the research consistently shows: the businesses that handle this well don't lose their best people. They lose their worst processes. The talent stays because the work gets better: less repetitive data entry, less manual reconciliation, less copying and pasting between systems. The people who were drowning in mundane work get to focus on the parts of their job that actually use their expertise.
The businesses that handle it badly? They announce "AI transformation" without explaining what it means for individuals, watch their best performers leave for competitors who feel more stable, and end up with an AI programme that technically works but can't be adopted because nobody trusts it.
Your people strategy is not a side-track. It is the main track.
What the CPO needs to own in an AI programme
You shouldn't be a passenger in this process. The people dimension needs to be designed from day one, not bolted on after the technology decisions have been made:
Before a single AI tool is deployed, you need a role impact assessment. Map which roles are affected, how they're affected and what the transition looks like. Not all at once, but role by role, process by process, as each initiative is planned. This gives people time to prepare and gives the business time to reskill.
Communication that's honest about change and optimistic about opportunity. The worst thing you can do is say "nobody's job is at risk" when people can see that some tasks are being automated. The best thing you can do is say "these tasks are going to be done differently, here's what your role looks like after the change, and here's how we'll support you through it."
If a claims handler's role shifts from manual document review to exception management and customer interaction, they need different skills, which is why skills gap analysis and reskilling plans matter early. Identify those gaps early and invest in development. This is cheaper than redundancy and recruitment, and it sends a powerful signal about how the business values its people.
Retention risk monitoring. During any change programme, your best performers are the most likely to leave, because they have the most options. Monitor engagement, have direct conversations and make sure the people you can't afford to lose understand their place in the future operating model.
Change readiness as a gate on deployment. Don't deploy AI capabilities to teams that haven't been prepared. If the communication hasn't happened, the training isn't in place and the managers aren't equipped to support their teams through the transition, the deployment should wait. Technology that's ready but the people aren't is technology that will fail.
The "AI anxiety" conversation
Every organisation going through this will face a period where employees are uncertain. That's normal. The question is whether you manage that uncertainty or let it manage you.
Managed well, it looks like: clear communication about what's changing and what isn't, individual conversations about role evolution, visible investment in reskilling and an honest acknowledgment that the business needs to operate differently to stay competitive. Managed badly, it looks like the opposite: vague announcements about "digital transformation," no clarity about individual impact, a perception that AI is being done *to* people rather than *for* the business, and a steady drip of resignations from people who decide the uncertainty isn't worth it.
What a good process looks like
The businesses that get this right start with a capability assessment that explicitly includes workforce impact. For each process being considered for AI-assisted automation, they ask: what happens to the people who currently do this work? Do they move to higher-value tasks? Do they need new skills? Is there a transition period? Who manages the communication?
This isn't extra work. It's essential work. And it needs to happen at the same time as the technology assessment, not after it.
At Oxygen Bubbles, workforce impact is built into our Breathe engagement from day one. When we map capabilities and score AI opportunities, we include a workforce dimension: which roles are affected, what the transition looks like and what the people team needs to plan for. You don't get a technology roadmap that ignores the humans.
A practical first step
If your business is starting to talk about AI and nobody has consulted the people team yet, now is the time to get involved. Don't wait to be invited. The decisions being made now about which processes to automate will directly affect your workforce, your culture and your retention for the next three years.
If you want the structured framework, get in touch and we will share it.
*If your CEO is driving the AI conversation, share this with them: The Blockbuster Question: Is Your Business Model Ready for the AI Era?*
*If your COO is looking at process automation, share this: 5 Signs Your Operations Won't Scale*