Talent Intelligence is getting smarter. But we may be looking at the wrong thing.

Why understanding talent is only half the puzzle in the age of AI.

Yesterday I spoke at the Global Talent Strategy & Intelligence Conference in Amsterdam. Two days surrounded by people from organisations including NATO, Microsoft, Booking, Coca-Cola HBC, Philips, HelloFresh, Riot Games and ServiceNow, all trying to understand how work and talent are changing.

I came to talk about From Planning to Matching and Agentic HR. But while listening to the other speakers, I realised that many of us were circling around the same, slightly uncomfortable conclusion: we are getting incredibly good at understanding talent. But do we understand the work itself well enough?

That question matters more than it may seem.

The job is starting to hide more than it reveals

For more than a century, we have organised companies around jobs. We take a collection of activities, bundle them together, give the bundle a title, salary and place in the hierarchy, and call it a job. It has been an incredibly useful abstraction. But it is still an abstraction.

Two people with exactly the same job title can spend their weeks doing very different things. And according to one of the figures shared at the conference, only around 44 per cent of the work people actually do is captured in job descriptions.

Then AI enters the picture. AI does not neatly replace a box in the organisational chart. It writes part of the report, analyses the dataset, answers the customer question, prepares the meeting, searches for candidates or generates the first version of the code. In other words, AI enters through the work, not through the job title.

Lisa K. Simon from Revelio Labs shared another fascinating finding: around 90 per cent of current work change is happening inside existing occupations through changes in tasks. That stopped me. Because if the change is happening inside jobs, why are we still doing so much of our workforce planning at job level?

Skills helped us open the box

The skills-based movement has already made an important correction. Instead of assuming that someone’s job title tells us what they can do, we are starting to look at their actual capabilities.

I strongly believe in that shift. It is also a big part of what we have been working on at 8vance. Good skills data can reveal talent that traditional HR systems simply do not see. Someone in Finance may have capabilities that are desperately needed in Operations. A nurse may have skills that open a completely different career path. Someone who looks like a poor match based on their CV may be remarkably close when you look at what they can actually do and learn.

But lately I have started to think that skills-based organising may be a transition rather than the destination. Skills help us understand people much better. They do not necessarily tell us what the work should look like. For that, we need to open the other box too.

What if we started with the work?

This was the idea behind my talk in Amsterdam. Imagine that instead of starting with a vacancy or job description, we start with a much simpler question: what actually needs to get done?

We can break that work into meaningful building blocks. Not endless lists of tiny tasks, but useful atoms of work: activities, responsibilities, outcomes, knowledge, skills, tools and context. Once you can see those atoms, you can start recombining them.

Some work may fit perfectly with someone already inside the organisation. Some could become a project for someone from another department. Some may require external expertise. Some could be performed by an AI agent. And quite a lot of it will probably be done by humans and AI together.

Suddenly, the question changes from who do we need to hire for this job to what is the best way to get this work done. That sounds like a subtle difference. I think it is a fundamental one.

Talent Intelligence needs Work Intelligence

Talent Intelligence has made enormous progress on one side of this equation. We can infer skills, analyse labour markets, understand career moves, discover adjacent capabilities and see patterns that were almost invisible a decade ago.

And the field itself is growing up. Sam Fletcher from Coca-Cola HBC and Eva Crampton from Booking.com showed how Talent Intelligence is moving beyond a specialist recruitment discipline. Dedicated TI job titles are growing, while analytical expectations are increasingly becoming part of broader recruitment and HR roles.

Toby Culshaw captured another important conclusion from the conference beautifully: the job title is dead as a unit of analysis. If that is true, we need to think seriously about what replaces it.

My answer would be that we need Work Intelligence alongside Talent Intelligence. Talent Intelligence helps us understand what people can do, what they could learn and where capabilities exist. Work Intelligence helps us understand what needs to be done, how that work is changing and which capabilities it requires. Put the two together and something much more interesting becomes possible: you can match work to capability continuously.

Maybe workforce planning should not be about planning the workforce

This is the part I find most exciting. Strategic Workforce Planning traditionally asks questions such as how many people we will need, in which roles, with which skills, and where the gaps will be in three or five years.

Those are useful questions, but AI makes the underlying assumptions increasingly unstable. If the activities inside a role change faster than our planning cycle, predicting the number of people required for that role gives us a false sense of precision.

Philips, through Monique Bijl and Babette van Biljouw, shared how it has already moved away from fragmented, FTE-based spreadsheet planning towards an integrated position-based approach, with continuous monthly planning as the next step. I wonder whether there is another step after that.

What if workforce planning becomes less about predicting the future organisational chart and more about building the capability to continuously respond to changing work? You still need strategy. You still need scenarios. You still need people making deliberate choices about where the organisation is going. But underneath that strategy sits a much more dynamic mechanism: continuously understanding the work, understanding available capability and bringing the two together. In other words: less planning the boxes, more matching the work.

This should not become Taylorism with better technology

There is an obvious danger here. Once you start talking about breaking work into atoms, it is easy to imagine an organisation in which humans themselves become resources to be endlessly optimised. Every minute measured, every activity allocated by an algorithm. That is not the future I am arguing for.

Dr Dieter Veldsman raised an important warning at the conference about extracting tasks from roles until what remains becomes a meaningless collection of leftover work. He also raised the question of what happens to junior development when AI takes over many of the activities through which people traditionally learned their profession. Those are not side issues. They are design questions.

If we can redesign work, we also get to decide what kind of work we want to create for humans. Jess Von Bank used a phrase I loved: moving from human in the loop to human in the lead. The closing discussion returned repeatedly to agency, empathy and hope as essential ingredients of this transition. Technology gives us more choices about how work gets done. The interesting question is what we choose to do with those choices.

From planning to matching

I left the conference feeling that a number of developments we often discuss separately are actually beginning to converge: skills-based organising, Talent Intelligence, Strategic Workforce Planning, internal mobility, AI agents and workflow redesign. They all point towards an organisation that understands work in much more detail and can continuously connect that work to the right capabilities.

For me, the journey looks something like this. From jobs to skills. From skills to work. From Talent Intelligence to Work Intelligence. And from workforce planning to continuous matching. Not because planning becomes irrelevant, but because planning alone is too slow for a world in which the work itself keeps changing.

A huge thank you to Toby Culshaw, Jess Von Bank, Geert-Jan Waasdorp and the Intelligence Group team for creating a conference where people genuinely shared what they are discovering, including what is not working yet. That openness was probably the most valuable part of the two days.

I went there to share my thinking about Agentic HR. I came home with a bigger question: what if the future of HR is not primarily about finding the right people for our jobs, but about continuously finding the best combination of human and artificial capability for the work we want to get done?

If that is where we are heading, understanding talent is only half the puzzle. We need to understand the work just as well.