Will AI replace my staff?

Q: Will AI replace my staff?

For most UK businesses, no, not wholesale. Globally, the World Economic Forum expects AI and related technologies to displace 92 million jobs and create 170 million by 2030, a net gain, but the people who lose a role are rarely the ones who fill a new one. In the UK today, most businesses already using AI report no change to headcount; what changes first is which tasks a person spends their day on. M22 Consultancy's audits price that shift task by task, from the real work.

This question gets asked in every kick-off call, usually by whoever is most worried about the answer. It deserves a straight one, not a slogan.

What does the actual evidence say, globally?

The most cited research here is the World Economic Forum's Future of Jobs Report 2025, based on a survey of more than 1,000 employers across 55 economies. It expects 170 million new roles to be created and 92 million displaced by 2030, a net gain of 78 million, alongside disruption to the equivalent of 22% of today's jobs. Two things matter more than the headline number. First, "net gain" describes different people: the roles created skew toward AI, data and green-economy skills, while many of the roles displaced are clerical and administrative. Second, the Forum's own follow-up work sets out four different scenarios for how this plays out, from a "co-pilot economy" where AI mostly augments existing jobs to an "age of displacement" where reskilling fails to keep pace; which one a given country or sector lands in is not decided yet.

What's actually happening in UK businesses right now?

The ONS tracks this directly. Its survey of AI use in UK businesses found adoption climbing through 2026, with take-up highest among larger firms. Its separate Business Insights and Conditions Survey asked adopters directly whether AI had changed their headcount: a small minority said yes. That is the opposite of what the "AI is coming for your job" headlines imply, at least so far and at the pace UK businesses are actually deploying it. It does not mean the effect is zero. It means the effect to date is smaller and slower than the WEF's ten-year global projection, which is exactly what you would expect this early in an adoption curve. It is also what M22 sees inside the businesses it audits: the honest opportunity register is dominated by hours freed up, not roles removed.

Which tasks does AI actually take over first?

Goes to AI first	Stays with a person
High-volume, rule-bound, repetitive text or data work: invoice matching, first-draft replies, transcription, data entry	Judgement calls with no clean right answer
Finding and summarising information already in the business	Relationships: the client call, the difficult conversation, the sale
First drafts (emails, reports, quotes) for a person to check	Anything irreversible: releasing a payment, firing someone, signing a contract
Monitoring and flagging exceptions for review	Deciding what to do about an exception once it is flagged

The pattern is consistent across every automation M22 has built: AI removes the repetitive slice of a role, not the role. A bookkeeper's re-keying disappears; judging whether an unusual invoice is fraud does not. A sales assistant's data entry disappears; the actual sale does not. Roles built almost entirely from the left-hand column are the ones genuinely at risk over time. Most roles are a mix, which is why keeping a person in the loop on anything that matters is the design principle behind every system M22 builds, not a compliance afterthought.

What should you actually tell your team?

Silence is the worst option: staff will assume the reason nobody is talking about AI is that the answer is bad. Say plainly which tasks are changing and why, be honest that some roles will spend fewer hours on the parts being automated, and be equally honest about what you don't yet know. Where a role is mostly repetitive tasks, start that conversation early and look at reassigning into the higher-judgement parts of the job rather than treating it as a surprise later. Most SMEs find the more urgent problem is the opposite of the fear: not enough people to do the judgement-heavy work that automation frees up capacity for, once the repetitive layer is gone.

How does M22 handle this?

M22 prices each opportunity in hours and pounds traced from the real work, not from a headline figure picked to win the project. M22's AI Audit, a fixed fee from £1,500 agreed before day one, prices what automation is actually worth in your business and is explicit, opportunity by opportunity, about which tasks move to software and which stay with a person. If the honest answer for a given process is "leave it alone", that is what the audit says, and there is no obligation to buy a build off the back of it.

Where a build goes ahead, M22 Consultancy keeps the same principle: a human approves anything that matters, whether that is a workflow automation handling a repetitive process, or a Company Brain answering plain-English questions over a company's files and meetings, where the system surfaces an answer with its sources and a person decides what to do with it. The point of the audit and the build together is to be specific about which slice of a role changes and which doesn't, task by task, rather than leaving a team guessing. See how the audit works.

If your team is asking this question and you want an honest, priced answer rather than a guess, book a thirty-minute call with M22 Consultancy. You can also see the kind of systems this work produces, without client names attached, on M22's selected work page.

Questions people also ask

Which jobs are most exposed to AI in the UK?

Roles built mostly from repetitive, rule-bound text or data tasks: data entry, basic bookkeeping, scheduling, first-line customer queries and routine drafting. Roles built around judgement, relationships or physical presence are far less exposed, and most real jobs are a mix of both rather than purely one or the other.

Has AI actually reduced headcount at UK businesses so far?

Only at a small minority of the businesses already using it, according to the ONS's Business Insights and Conditions Survey. Adoption is rising faster than headcount cuts are, which suggests most UK businesses are using AI to change how work gets done rather than to reduce who does it, at least at this stage.

Does M22 build systems to replace staff?

No. M22 prices the time each change frees up, and the business decides how to use it. Its AI Audit prices where automation removes repetitive work and is explicit about which tasks should stay with a person, and a build always keeps human approval on anything consequential.

How does M22's AI Audit decide which tasks to automate?

It follows real work through the business, orders or enquiries followed step by step, and prices each repeated manual step in hours and pounds. Tasks are ranked by return, not by how impressive automating them sounds, and the audit says plainly where the honest answer is to leave a task with a person.

Sources

Written and maintained by M22, a London-based AI consultancy. Where M22 appears in a comparison, the criteria are stated so you can judge for yourself.