We believe AI will fundamentally change the people function in the years to come, in recruiting, in HR, and in how companies build and develop their teams. We also believe that is good news. Used well, the technology takes over much of the repetitive groundwork and gives people more time for the work that actually requires them: understanding organisations, advising leaders, and making good calls about people.

We want to help shape that change rather than react to it, so we are building Amby to stay ahead of the curve, and we invest real time and resources in getting there. This post explains what that looks like in practice: what we believe, how we work with AI today, what we have learned so far, and what it means for the people who work here.

What we believe

Technology should do the pattern work, and people the judgement

Our work in recruiting and HR has always involved two very different kinds of tasks. The first follows a pattern: researching a client before a meeting, reading the talent market, building a list of target companies, summarising interviews, and drafting job ads. The second does not: telling a client that the role they have described is really two roles, noticing what a candidate is not saying, or recognising that a hiring problem is actually a leadership problem.

AI is already very good at the first kind of work and nowhere near good enough at the second, and practitioners seem to agree. In our survey of HR and talent professionals this year, almost everyone who uses AI uses it to write job ads, while almost nobody lets it interview candidates. We think this is the right place to draw the line, and we design our work around it on purpose.

We want to be ahead of the change, not catch up with it

The tools will change, what our clients need from us will change, and so will what good looks like in our field. We would rather change early and on our own terms than late and because we have to, so we treat our own ways of working the way we would treat a client’s: something to test, question, and keep improving.

Everyone builds with AI

AI at Amby is not something a few enthusiasts use on the side, and it is not a project that a small team builds and then hands over to everyone else. The people who do the work know best how it should be done, so they are also the ones best placed to shape how AI supports it. In practice, that means our consultants test new setups on their own projects, point out where AI could help, and tell us when something does not work. It is also what makes new ways of working stick, because people are far more likely to use something they have helped build.

How we work with AI today

AI is part of the everyday work

Everyone at Amby works with the same AI model, and it is part of how we research clients and markets, prepare for meetings, take structured interview notes, and produce first drafts. We chose one model for the whole company rather than letting everyone pick their own, because going deep with one tool has given us far more than spreading ourselves across many, and we set clear rules for data from the start: candidate data stays in our applicant tracking system and never goes into the model.

We invest in testing and learning

A lot of our time goes into trying things out: testing new tools on real work, and looking for new ways to solve problems we have solved the same way for years. Much of it does not stick, which is part of the point, because what does stick has been properly tested before it becomes part of how we deliver.

We learn from people outside our own bubble

It is easy to become convinced you are good at something when you only compare yourself with yourself. So we regularly consult AI experts, and we use AI-native startups and scaleups, both within the people space and well outside it, as sparring partners. Many of them have built their companies around AI from day one, and seeing how they organise their work, what they automate, and what they deliberately keep human has shaped a lot of our own thinking. It also keeps us honest, because it is much harder to believe you are ahead when you regularly spend time with people who are further along than you.

We map first, then start where it matters most

Before automating anything, we mapped our full recruiting process, including how our people actually do each step rather than how it looks on a slide. We then evaluate each step by how feasible it is to support with AI and how much value it would add, start where both are strongest, and work our way along the process from there. Not every step will need AI, and where it does, the right solution will not always look the same.

One of the tools we use is what we call a skill, which is a written playbook for one recurring task that the AI follows every time. Our first one covers the research we do before meeting a client about a new role: the company, the role, the talent market, and where the right candidates are likely to be. It took more than thirteen versions, shaped by colleagues running early versions on real searches and showing us what was wrong, before it was genuinely useful. We have since built similar support on the HR side, for example for salary benchmarking.

The AI drafts, and people decide

The AI never delivers a final answer. It delivers a draft, and everything it cannot verify is flagged for a consultant to check, whether it is something only the client can confirm, data that needs to be looked up, or an assumption that needs testing. Our people start from a solid foundation, and the judgement stays with them.

What we are learning along the way

The biggest surprise has been where the work actually sits. Almost none of the improvements we made came from making the AI smarter, and most came from writing down our own standards more precisely and getting the AI to follow them.

Building good AI support has forced us to become much clearer about what good work looks like at Amby.

We have also learned that the hardest thing to write down is judgement. The steps are easy to describe, while the calls an experienced consultant makes without thinking are not, which is exactly why those calls stay with people.

Finally, as with most technology changes, and AI is no exception, building is the easy half and adoption is the hard one. A new way of working is worth nothing if nobody has time to take it on, which is why we have made AI part of how we work rather than something people try when they have time.

What this means for the people at Amby

For the people who work here, this changes the job in a good direction. Less time goes into repeatable groundwork, and more goes into what drew most of us to this work in the first place: the conversations with clients and candidates, and the calls that require real experience.

It also changes how we learn. Know-how that used to be passed on slowly, from one project to the next, is now written down in a form everyone can use, so newer colleagues start from a much stronger foundation and experienced colleagues get to shape how the whole firm works.

We are introducing an AI standard at Amby, where AI proficiency becomes part of our performance reviews and a requirement for moving up in seniority. We expect everyone to get good at working with AI, and we help them get there. And since the way we work will keep changing, we look for people who find that exciting rather than tiring.

Where the people function is heading

In the years ahead, we expect technology to take over more and more of the repeatable work in recruiting and HR. We do not see that as a story about machines replacing people. We see it as the people function finally getting the time to do more of what it has always wanted to do, which is helping leaders build strong organisations and helping both people and companies grow. Recruiters and HR partners will increasingly work as advisors, and the value of good human judgement will go up, not down.

The people function is changing, and we want to help drive that change.