Artificial Intelligence has moved rapidly from emerging innovation topic to core HR priority. The question is no longer whether to adopt AI, but how far organisations have progressed, what to automate, where human judgement remains essential, and how to protect governance, trust and employee experience.
To explore these themes, I spoke with Hugh Clarke, People AI Lead at Neko Health, an AI-focused healthcare company. With experience across HR, Workday consulting, HRIS leadership and People Innovation, Hugh now designs AI-driven People solutions that automate repetitive work, improve employee support and enhance efficiency while keeping people at the centre of decision-making.
In this interview, Hugh shares his views on AI adoption, governance, automation, future skills and what the next generation of HR technology leaders should prioritise.
HC: I started as a traditional HR generalist before moving into a Workday implementation role at Deloitte, where my HR technology career really began. I worked across Europe and the US, building deep expertise in Workday and the wider HRIS landscape.
From there, I moved in-house as an HRIS Manager, applying that consulting experience to managing internal platforms, building teams and widening my scope beyond Workday to the broader HR technology architecture.
My move into AI came from personal interest. I was an early adopter of ChatGPT and spent time exploring practical applications, so when an AI-focused role became available at Wise, I put myself forward. Wise encouraged an AI-first mindset, and I quickly recognised that AI would become a necessity rather than just an opportunity.
It was exciting and daunting. After years in Workday and HRIS, I still felt new to AI, but my understanding of HR technology and systems architecture helped me identify where AI could automate processes, connect systems and accelerate tool development.
The biggest shift came when I joined Neko Health. At Wise, AI was one part of my role; at Neko, it is my full focus. In a growing organisation with an evolving People function, I have the freedom to experiment, build strong foundations and embed AI properly from the outset.
HC: The role is a mix of strategy, architecture, stakeholder management, experimentation and delivery. When I joined, the priority was laying the foundations: core processes, better data and stronger knowledge bases. Without that, it is difficult to scale AI effectively.
Much of my time is now spent experimenting, thinking like a product owner and building solutions. But architecture underpins everything: we want connected, sustainable solutions rather than standalone tools, so governance and maintenance are major priorities.
Strategy is equally important. We have a huge opportunity to build from a relatively blank canvas, but that means being deliberate about where AI adds value. The focus is sensible automation that frees people up for more strategic work as the business scales.
Stakeholder management remains one of the most enjoyable parts of the role. I work across People teams, IT and Engineering. What has changed is speed: prototypes that once took weeks can now be created in days, giving stakeholders something tangible to react to almost immediately.
HC: Two things stand out, and neither is purely technical. I am not a software engineer; my technical grounding is in understanding the back end of an HRIS and how to build and maintain one. I can hold a technical conversation, but it is not where my edge is.
The first is deep HR domain knowledge. Years as a consultant and then in-house, building and maintaining platforms and understanding what companies need from their processes and data, gave me a real feel for the HR landscape. I have seen which parts genuinely change and which have stayed the same for years.
The second is consulting. It taught me stakeholder management and, above all, translation: taking a less technical person’s requirements, turning them into something a system can do, then translating it back clearly. It is about building the relationship until you land on what they really need, not just what they first asked for.
If I had to rank them, I would put HR and consulting above the technical every time. Translation matters more now than ever because many platform limits have gone. The bottleneck is now how well you understand the real problem and shape it clearly.
HC: At its core, many of the problems are still the same. Good data and properly mapped processes made or broke projects ten years ago, and they still do now. If anything, they matter more because AI lets you build so quickly that shaky foundations surface faster.
HRIS implementations generally move faster than they used to, and better tools now support areas such as data cleaning. But building the right architecture, functional setup and integrations still takes real work and testing. AI can help where it genuinely adds value, but its speed should not encourage teams to rush the parts that need care.
Your HRIS implementation is one of your foundations, affecting payroll, IT and much else downstream, so it has to be done properly. If you are starting an HRIS project, map your data and processes early. If you build AI on weak foundations, you can spend years unpicking the consequences and end up further behind.
The fundamental change is who gets to build. Developers used to be the bottleneck; now, as people become more tech and AI fluent, many more can build solutions instead of waiting in a queue.
On the vendor side, Workday’s dominance has grown over my career, and we are at a pivotal moment. Its agent features look interesting, but whether they are enough remains to be seen.
HC: The focus should be quick wins: plugins, skills or automated workflows that remove repetitive work where your team spends most of its time. You should not try to build your own HRIS; you still need robust core platforms. AI is most useful as a layer on top, supporting areas such as payroll processing or employee queries. That repetitive work is where the opportunity is.
Whatever you build, keep the human in the loop. It is called Human Resources for a reason. The human should always be there somewhere, whether that is checking the output or stepping in to support.
Every organisation is different, so start with your own priorities: the biggest pain points, and which are quick wins or easy deployments. That requires experimentation, which is why people need time and encouragement to test ideas on an AI platform.
This should not sit only with People Tech. Everyone needs AI fluency because people know their own processes best. If teams can build solutions to their own problems, things move faster than if everything funnels through one team.
On hype, fully autonomous AI recruiters are the clearest example. You may save time, but candidates often get a worse experience, which can damage their view of the company. That is where the human belongs.
Instead, we automate the manual parts of job creation, such as job specs, interview questions and sourcing strategies, freeing recruiters to focus on finding candidates and executing the strategy.
HC: Employee trust is essential, and trust depends on data quality. If you build a knowledge base, you need regular checks on its content and on the responses your bot gives employees. You cannot ask people to trust answers from a stale or messy source.
Governance needs to be deliberate. That means working closely with legal and privacy teams, understanding your AI platform contract, training employees on what they should and should not put into these tools, and being clear about how sensitive HR data is handled.
With anything we build, we put hard gates in place. In the case of our bot, it does not invent anything, it does not pull from external sources, and if it is not highly confident in its response, it brings a human in. It is Tier 0 support, not a replacement. The human is never out of the picture, and we bring one in quickly when needed.
HC: Tech and AI fluency matter, but so does understanding what the business and stakeholders really need. You can create impressive-looking AI tools quickly; the discipline is asking whether they solve the problem or simply add another tool to the stack. Strategic thinking is the difference.
The other point people miss is what happens after automation. When you remove part of someone’s work, what does the team do with the time returned? Deleting work is the easy half; making sure that time goes to higher-value work is harder and more valuable.
The AI Lead does not own every individual’s development, but they should make sure the conversation happens: if ten hours a week come back to a team, that time needs a plan. The failure mode is automating work without thinking about the gap it leaves behind.
HC: Prompt design is less of a distinct skill than it was a year ago, because the models have got much better at working out what you mean. Coding, as I have found myself, you do not strictly need to be an expert in.
There are two things that actually compound. The first is simply building things. Fluency does not come from a course, it comes from picking a real problem you own and making a scrappy tool for it. That is how you learn what these models genuinely can and cannot do, which is the model limitations point, except learned in your hands rather than in theory.
The second is problem framing. Knowing what is actually worth solving and being able to define it clearly. That is the part that does not get automated, and it is the same translation muscle that has served me throughout my career.
So the honest answer to which skill from that list is that they are all downstream of just starting. Pick something that annoys you in your own job and build a version of the fix.
HC: Everyone needs a level of technical skill, but that means something different now. You do not need to code; AI can handle much of the coding and technical explanation. In some ways, that side is easier than ever.
What you do need is to understand the technology architecture and landscape. Not to build the integration yourself, but to know why it matters and how the pieces connect. Alongside that, you need enough AI fluency to speak the language.
The bigger point is looking outward. HR has often worked through a narrow lens, but in the AI era that is risky. Different functions may solve the same problem separately, creating duplicate solutions. HR should understand what other departments are building and where it can adopt, join or reuse existing work. Organisational awareness matters more than any single technical skill.
HC: Get in there and experiment. Play around, be creative, be adventurous and keep looking for workarounds, while being careful about what data you use and where it goes.
Early in your career especially, there is a real opportunity to be the AI champion in your organisation or your department. Be the person who knows how these tools work and who comes in with fresh ideas.
That fresh perspective is the asset. HR problems often persist because long-standing processes become normalised. If you can question them, learn the business properly and understand stakeholders, that becomes your edge.
So: learn the business properly, build constantly and be brave about it.
HC: The goal was never to remove the human from HR. It is to give them their time back for the parts that actually need a human. The leaders who get this right will be the ones who automate in order to make HR more human, not less.
Annabelle Yee is a Consultant in the HR Infrastructure team. Get in touch to find out more or continue the discussion.
