The HR leader's guide to AI transformation: What actually works in 2026

AI meets the workforce at a moment when people are on the edge. 

2026 data from Gallup found that employee engagement fell to 20% in 2025 — the lowest since 2020. Mercer's 2026 data found that the share of employees who describe themselves as "thriving" has fallen from 66% to 44% in two years. 

People are already running on low battery. But it’s here in this low-engagement environment that organisations are betting on their biggest transformation yet: integrating AI meaningfully into how work gets done.

The Personio Index Q2 2026 found that AI-related job title changes have increased 4.9-fold from the Q2 2024 baseline, with April alone exceeding all of Q1. Meanwhile, according to 2026 McKinsey data, 88% of organisations are deploying AI in some form — but 86% say they’re not prepared to embed it in daily operations. 

People are feeling the reality of their work change before the organisation is actually ready to make it work.

That makes how HR steps into the frame at this moment especially important. HR is the function that can turn AI into something people can use, question, and trust in practice, and build a people-centred strategy that acknowledges nobody has all the answers.

→ Read: AI is changing jobs — and what HR leaders need to do about it

→ Read: Personio Index Q2 2026

In the AI era, organisations often default to speed first. The fear of missing out spurs a need to recover lost ground — meaning organisations end up introducing AI into workflows before employees understand it, trust it, or have even been asked for their input. 

This can result in a trust “debt”, where trust begins to erode in a way that makes it hard to sustain long-term transformation. Long-term, this loss of trust can be expensive. According to our Workforce Pulse research from 2025, employees who want more support with AI in the workplace are 35% more likely to be job-hunting.

"One of the things that definitely stung was employee sentiment of AI being done to them as opposed to with them. We didn't appreciate that employees were at different stages of that journey — and that definitely shaped the training we did after." — Neil Millen, VP People Business Partners, Personio

We need your consent to load this service!

This content is not permitted to load due to trackers that are not disclosed to the visitor.

More InformationAccept

Part of why this is so hard is because HR teams often can’t supply the clarity and certainty people need, because they don’t have it themselves yet.

Jess Larsen, founder of Thriving Humans and a specialist in AI change management, found this pattern repeating across her network of CPOs and people leaders in 2024. "The reactions I got really surprised me," she said. "These are people who I had a lot of respect for and saw as very capable. But there was a lot of: I haven't really learned enough about AI. I don't really know where to start. And then: my leaders are freaking out. The CTO is leading on it. The CIO is leading on it. And they're asking, well, what's happening for the people? We're not really sure. We're kind of waiting to find out."

The HR leaders who keep employees on-side during an AI transformation focus on trust as the principle that underpins all change — the thing that makes speed sustainable rather than the thing it has to be traded against. In practice, this means three things.

First: making accountability visible. Define explicitly where AI assists and where humans decide. Performance management is a common boundary. "We definitely slowed down where there's a high degree of judgement still needed or subjectivity," said one attendee at Personio's Making AI Work event. "High performance management is a really good example — where we've been really conscious about where we lean on AI and where we protect some human element in the system."

Second: equipping managers with honest language. Employees don’t need certainty — they need honesty and transparency. Managers who can say "I don't know yet, and here's when you'll hear more" maintain trust better than those who don’t say anything at all.

Third: meeting employees where they are. Lenke Taylor, former Chief People Officer at Personio made the case from an employer's perspective: "If I want people in my workplace to be successful workers, it's in my best interest to help them become curious and want to learn these tools — rather than creating a dynamic where people feel that if they learn to use the tool, they're going to automate their own job away. That's not a win-win for either party."

→ Read: Trust during AI rollouts: what HR leaders are actually doing 

→ Download: AI Rollout Trust Kit — a practical playbook for HR teams

The standard framing of the AI skills problem is a gap analysis: here are the skills that exist, here are the skills that are needed, here is the training budget. It’s a useful starting point but it misses the more difficult underlying challenge.

AI doesn’t create a one-time skills gap. Instead, it’s making skills change faster, and apply more broadly across roles in a way that job descriptions and annual performance reviews were never designed to track. The question for HR isn’t "how do we close the gap?" but "how do we build the infrastructure to keep up with something that keeps moving?"

Teresa Rose, a skills-based organisation strategist and founder of ConsultHer, who spoke at Personio's Making AI Work event, framed it this way: "Roles provide structure and legal clarity, but skills make your workforce dynamic. Moving skills fluidly across teams unlocks value — one person with adjacent skills can cover multiple roles, reducing hiring costs and increasing agility."

Applied to real-life hiring, this thinking can extend organisations’ reach for open roles — because organisations aren’t just looking for someone who’s racked up the right amount of experience on their CV. Dominic Joyce, a talent acquisition practitioner and founder of Maverick Otter said he saw a 47% increase in applications for retail roles when applying skills-based hiring.

Ultimately, Joyce said, there’s a stronger case for what a skills record does over time when compared to years of experience.

“If you had a framework in place where you documented someone's progress over the last two to three years — base skills when they joined, skills acquired over time — it builds trust. People know that when they're going to be hired or promoted, it's based on a framework, not based on 'I have three years' experience, I deserve a pay rise.“

Dominic Joyce

Founder and Head of Talent Acquisition, Maverick Otter

→ Read: AI for HR: where it helps, where it doesn't, and how to decide

HR is the only function in the organisation that has a dual vantage point on how change is landing. It’s simultaneously close enough to the people on the ground to spot how they’re feeling and where trust is waning. But it’s also high enough above them to see the patterns in employee data that map to that change — and show their business impact.

Roles, skills, engagement, and retention risk: these are an organisation’s most strategically valuable data points to identify the impact of change as they move through it. Yet in AI-driven transformation, HR is often left handling the communication strategy — and explaining the change — rather than managing the impact of that change.

“If we as a function aren't owning this, what's going to play out in organisations? Are we these admin people who just manage processes, or do we jump feet first into the strategic value? We sit on a landscape of data. Start building that vision.“

Jess Larsen

CPO and Founder, Thriving Humans

That vision can only only hold if HR comes in at the decision-making stage and shapes the messaging — not when the tools, rollout timelines, and workforce impact have already been rubber-stamped. 

2026 research from Boston Consulting Group found that when organisations focus on the human part of the change above all else, it can increase the likelihood of it sticking by up to 90%. Leave HR out of the conversation, and that long-lasting impact might vanish along with the promised ROI. 

But brought in early, HR can shape AI talent strategy in a way that provides clarity and honesty over how skills and roles may shift, what you know and don’t know, and critically, what that means for employees.

"The real value of these tools is — what if I could tell my managers in real time what's happening in their organisation?” Taylor said. “Who just resigned, what the hiring pipeline looks like, where the gap is going to be. A human simply cannot do that. But if you set up these tools, there's a future where you could proactively take that to the business and say: we're definitely going to be behind on hiring for software engineers, let's do something about it now. We don't talk about the tools that way right now. We talk about them like: automate something."

AI tools are only as good as the data they draw on. And in HR, that data is often not good enough. Employee records are often incomplete across systems, skills ontologies don’t get updated for months, and performance data lives in one place — but compensation management lives in another. 

Disconnected processes and incomplete data make it hard for any AI to make reasoned insights or analysis — it can also introduce bias. In a workforce context this can be especially damaging, because it gives the illusion of evidence-based decision-making to processes that impact real people. 

“If you're only looking at one piece of the data, you're losing all of the connected value and the ability to make really interesting and rich decisions. What we're really trying to do is think about all of the ways that data is interconnected — recruiting, payroll, time and absences, performance — and if you're only looking at one piece, you lose the ability to make the decisions that matter.“

Alex Bannon

Engineering Manager, Personio

According to 2025 research from Gartner, the primary cause of AI failure in enterprise contexts is the absence of the right data foundations — with the consultant estimating that organizations will abandon 60% of their AI projects through 2026 due to a lack of AI-ready data. 

But the good news, then, is that perfection isn’t the bar. "Good enough, accurate enough, and connected enough to support the specific decisions we need to make" is the right standard. To evaluate the quality of your data estate, ask:

• Where does our core HR data live?

• How complete or accurate is our data?

• How many HR platforms do we currently use — and are the definitions the same across each platform?

• Do we have any processes where people are still using manual workarounds for data — such as spreadsheets or files?

• Where would poor quality data create a risk to the business?

→ Read: Connected data beats clever AI

Once you’re in the demo, it’s easy to get dazzled by what AI tooling can do — that’s the point of a sales pitch. But when evaluating AI tools for HR, you need to be able to get beyond the spiel, and ask questions that unpick how the tool uses your data, how it makes decisions, and how the tool does (and doesn’t) support HR processes.

Ask these four questions:

1. Where does it get its data? A tool drawing on your own HR system of record — recruiting, payroll, performance, absences, all connected — is fundamentally different from one generating outputs from general training data or a single source. The connected context is what makes AI insight useful rather than generic. Vendors also need to be able to explain how their tool uses data to make decisions. 

2. Where does the human stay in the loop? For any decision affecting employment, compensation, performance, or career opportunities, the human decision point should be explicit and non-negotiable. If a vendor cannot clearly articulate where a human approves, overrides, or decides, that is a red line. Under the EU AI Act, this is also a legal requirement for high-risk AI systems.

3. What happens to your data? All data should be processed and stored within the EU. Your customer data should not be used to train the vendor's models. These are baseline requirements under GDPR, not premium features. You should also ask how the tool protects sensitive data and controls access.

4. What won’t the tool do? Vendors with genuine principles should be as clear about the limits of their tool as they are about its capabilities. You’re looking for clear decision gates that show what it will not automate and where it doesn’t make decisions: "We don't run payroll autonomously — that requires 100% accuracy and a human check." "We don't identify a ranked list of underperformers — that judgement belongs with managers."

“The minute that we burn trust with our customers by shipping something that is just not good enough — it's not worth the trade-off.“

Camille Merritt

Product Manager, Personio

→ Read: A builder's view — what good AI looks like in HR software

The organisations that make the most progress on AI in HR start with clarity about what problem they are trying to solve. Before investing in any AI HR tools, get three things straight:

• Use case: Where are HR and managers making decisions with inadequate information? Where is significant time going on data collection and summarising? These are your highest-value AI use cases — not the most technically interesting ones.

• Data foundation: Is your data complete, connected, and accurate enough to support AI? Where are there gaps that need to be fixed?

• Human accountability: Which HR decisions and processes always require a human? Which AI can assist with — and which AI can mostly handle with human review?

AI in HR refers to the use of artificial intelligence technologies to support HR processes — including recruiting, performance management, workforce analytics, employee engagement, and operational administration. This ranges from automating repetitive tasks (scheduling, data entry, document summarisation) to providing analytical insights that support HR and management decision-making.

The main risks are trust erosion (when AI is perceived as opaque or unfair), poor data quality producing misleading outputs, and over-reliance on automation in situations that require human judgement.

Start with your data and your most valuable decision points, not with a platform selection. Understand what people data you have, where it lives, and where the gaps are. Define which decisions you want to improve, and what information you would need to improve them. Then evaluate tools against that specific brief.

No. The most effective AI in HR removes the operational burden that prevents HR professionals from doing their highest-value work: strategic advising, employee experience design, difficult conversations, and organisational development. The shift is from subject-matter knowledge holder to strategic partner — enabled by AI freeing up the time and providing the data that strategic work requires.