The conversation around artificial intelligence in the workplace has been dominated, for too long, by a single anxious question: will it take our jobs? It’s an understandable fear. But it’s also, I’d argue, the wrong question — and the frame it creates is actively unhelpful to organisations trying to figure out how to move forward.
A better frame, one I keep returning to, is the idea of the AI exoskeleton. Not a replacement. Not a competitor. A wearable — something that fits around the worker, amplifies what they can already do, and takes the physical (or cognitive) strain off the parts of the job that wear people down.
What is an AI exoskeleton, exactly?
The physical exoskeleton has been around in industrial settings for years. Factory workers wear them to lift heavy objects without destroying their backs. Surgeons use them to hold instruments steady for hours. The exoskeleton doesn’t replace the worker — it removes the physical limits that would otherwise cap what that worker can achieve.
AI tools, used well, do the same thing for knowledge work. They remove the cognitive limits: the time it takes to synthesise a hundred documents, draft a first version, spot the anomaly in the data, translate a concept into five different languages, or simply remember everything relevant from the past three years of a project.
The worker still brings the judgement, the relationships, the ethical sense, the contextual understanding, the ability to read a room. The AI brings scale, speed, and consistency in the areas where humans are slow, inconsistent, or simply constrained by the hours in a day.
What this looks like in practice
Let me be concrete. Here are three roles where I’ve seen this exoskeleton effect play out clearly — and where it’s changing expectations in ways organisations haven’t fully caught up with yet.
The analyst
A mid-level analyst used to spend a significant portion of their week pulling data, building reports, and writing up findings. AI tools now handle most of the assembly work in minutes. The analyst’s week hasn’t disappeared — but its shape has changed completely. The job is now almost entirely interpretation, stakeholder conversation, and strategic framing. That is a better job. It is also a harder job, in a different way.
The developer
A developer with AI code assistance isn’t writing less code — they’re often writing significantly more, faster. But the balance has shifted. More time goes on architecture decisions, code review, specifying intent clearly, and catching the subtle errors that an enthusiastic AI will confidently produce. The craft shifts upstream.
The manager
This is the one organisations are most unprepared for. AI assistants can now draft performance reviews, summarise one-to-ones, flag patterns across team communications, and surface risks in project timelines. The manager who uses these tools well can hold more context, across a larger team, with greater consistency.
The gap organisations are ignoring
Here is where most organisations are currently failing. They’re deploying the tools. They’re even seeing productivity numbers move. But they’re not asking the harder question: what does it now mean to be good at this job?
An exoskeleton changes the performance ceiling. But it also changes the floor. A worker who can’t use the exoskeleton effectively is now at a disadvantage they didn’t have before.
This isn’t a technology problem. The tools are, broadly, quite good. It’s a change management problem dressed up as a technology rollout. And the organisations that treat it as the latter will keep wondering why their AI investment isn’t delivering what the vendor promised.
What good looks like
The organisations getting this right share a few characteristics. They’re not necessarily the ones with the most sophisticated AI deployments. They’re the ones that have taken the human side of the change as seriously as the technology side.
They involve workers early — not to get sign-off, but to genuinely understand which parts of the job AI should be augmenting and which parts it should stay away from. They rewrite role expectations honestly, reflecting what the job now requires.
The human cost of getting it wrong
For workers who have built their careers on deep process expertise — on knowing the system, knowing the manual, being the person who remembers how something was done — AI can feel genuinely destabilising. Not because their jobs are disappearing overnight, but because the thing that made them valuable is now less scarce. That is a real loss, and the appropriate response is not to minimise it.
Where does this go next?
The exoskeleton metaphor has limits, and it’s worth being honest about them. Physical exoskeletons assist but don’t replace human movement — the human is always in control. With AI, that boundary is blurring in ways we haven’t fully worked out yet.
But as a starting frame — as a way to push back on the replacement narrative and redirect attention to the adaptation challenge — I think it holds up. The question isn’t whether to put the exoskeleton on. It’s whether you’re going to design how it fits.
