From Jobs to Skills: Insights From Our Executive Breakfast

The way large organisations think about their workforce is quietly, fundamentally shifting. Not from a press release or a strategy deck - but from conversations happening right now in leadership teams across financial services, healthcare, professional services, and beyond.

We recently hosted a breakfast roundtable with senior HR, reward, technology, and workforce planning leaders from some of the UK's largest organisations. The discussion, held under Chatham House rules, demonstrated a clear and consistent signal: the old way of seeing the workforce - through headcount, job titles, and org charts - is no longer enough.

From jobs to skills: a shift in the unit of analysis

For decades, workforce planning has been anchored to roles: you hire for a job, you design pay bands around a job, you train someone into a job. The job is the unit.

That’s changing. Across the table, leaders were describing - at different stages of maturity - a move toward skills as the fundamental unit of the workforce. Not the role or function, but the capability.

This shift is partly driven by AI. But as several voices in the room were quick to point out, the need for agility predates the AI moment. The pandemic demonstrated that organisations could adapt at pace when they had to. Skills-based thinking is how you build that capacity without a crisis forcing you to.

What AI has done is put a rocket under the urgency. When roles are changing faster than job descriptions can keep up, and when the half-life of a skill is compressing year on year, organisations that plan around fixed roles are planning to be slow.

Human at the centre - not just in the loop

One of the most thought-provoking reframes of the morning came around language. The phrase "human in the loop" has become shorthand for responsible AI deployment - a way of saying that a person is still involved. But several leaders pushed back on the framing.

If AI is the system and humans are inserted at checkpoints, you've already made a choice about who is running things. The more useful model, they argued, is human at the centre, AI in the loop - where the human owns the outcome, the strategy, the judgement call, and AI is the capability layer that makes that judgement faster, more informed, and more scalable.

This distinction matters enormously for how you design roles in an AI-augmented workforce. A role built around oversight of AI outputs is different from a role where human expertise is the primary value and AI is the tool that enhances it. Both exist. But conflating them creates skill gaps, performance measurement problems, and - ultimately - a workforce that doesn't trust the technology it's supposed to be working alongside.

The skills map problem

Nearly every leader in the room was grappling, at some stage, with the same challenge: they don't have a skills map. Or the one they have is out of date. Or it exists in spreadsheets. Or it was built for a world that no longer exists.

Several organisations are starting from zero - conducting qualitative interviews, walking factory floors, reviewing process documentation - doing it the old-fashioned way first, because there isn't yet an AI that knows what the specific work in your specific organisation actually looks like. The data has to come first.

What was consistent, though, was the aspiration: to move from that qualitative foundation into a dynamic, AI-inferred view of skills that can be continuously updated. The challenge isn't building a skills taxonomy for the organisation as it is today. It's building one that's alive - that can tell you, in near real-time, what skills are emerging, which are declining, where you have latent capability you're not using, and where the gaps will be in eighteen months.

One leader described it plainly: "We don't do workforce planning because we don't have the data and insight to think about it. Where we do, it's ad hoc, based on hunches, or something blatantly obvious. If we just had structured data, we could make such big advances - let alone when we start using AI."

That's the prerequisite. Structured, trusted, up-to-date data on what your workforce can do and what your organisation needs done.

The agentic shift

Copilots and assistants were already old news in this room. The conversation had moved to agentic AI - AI that acts, not just advises. And with that comes a new management challenge: how do you manage a workforce that includes both humans and agents?

Questions that don't yet have clean answers: How do you set objectives for an AI agent? How do you measure its performance? How do you allocate tasks between a human and an agent based on what each is genuinely better at - not just what's cheaper or faster?

The prerequisite, in every case, is task-level visibility. You can't allocate what you haven't mapped.

One framing that landed well: treat AI agents as co-workers. Not tools, not automations - co-workers with a defined role, a skill profile, and a relationship to the human doing adjacent work. That reframe changes everything from how you design jobs, to how you think about accountability, to how you measure team performance.

And then there's the scaling problem. Gartner data cited in the room suggests that 95% of AI implementations and POCs have yet to scale. The capability exists. The data quality, the skill to drive it, and the organisational design to support it - those are where most organisations are stuck.

What good looks like

When we asked leaders to describe the two biggest problems they'd solve if they had full visibility of their operating model - skills, tasks, automation propensity, redeployable capacity - the answers converged on a few themes.

Agility: The ability to move from strategy to execution faster. To know, six months in advance, what skills a new product or initiative requires - and to already be building the team rather than scrambling to hire it.

Empowerment: The ability to give employees visibility of their own skills, their own development pathways, and their own place in the organisation's future - not just their current role.

Redeployment over redundancy: When AI-driven automation reduces demand in one area, the question isn't how many jobs disappear - it's how many people have transferable skills that can be redirected. That calculation requires data. Most organisations don't have it yet.

Where Beamery fits

The conversations from that morning are precisely the territory Beamery was built for.

Our Work Intelligence capability makes the work visible - not just the workforce. Starting from a standard HRIS extract, we can infer which tasks are performed in which roles, at what volume, and with what propensity for AI augmentation or automation. That's the foundation that turns "we think 25% of roles could be affected" into "38% of Finance task-hours are automatable - here's what that means for redeployment and reskilling."

Our skills graph connects internal employees and external candidates on one governed model - so you can see not just who you have, but who you can reach. And our agentic AI layer, Ray, helps workforce leaders act on that intelligence in real time, not after another six-month consulting engagement.

If the questions from that breakfast table are ones your organisation is asking - we'd be glad to show you what the answers could look like.

Beamery is the AI platform for workforce transformation. To learn more or speak with one of our team, get in touch.