Big companies are blaming AI for layoffs. What should a small firm actually do?
Not what the big companies are doing. The pattern of the last couple of years is that large employers announce job cuts and name AI as the reason. Some of that is genuine: certain functions really do need fewer people for the same output now. A good deal of it is cover for over-hiring, because an AI restructuring story reads better in an investor call than an admission that the firm added too many seats when money was cheap and is now removing them.
A small firm should read those headlines the way it reads weather reports from another country. You are not carrying thousands of redundant seats. Your problem is usually the opposite one: every seat is visibly busy, administration grows faster than margin, and the owner absorbs the overflow after dinner. For an SME the sensible play is capacity without headcount: let automation take the administrative growth, so the next person you hire is a producer rather than an administrator.
Why are big companies naming AI in their layoff announcements?
Two motives, usually tangled together in the same press release.
The first is real. In functions such as customer support triage, content production, internal reporting and parts of software development, current models genuinely change the arithmetic. A large firm with hundreds of people in those functions can now run the same volume with fewer of them, and some of the announced cuts are exactly that.
The second is narrative convenience. Many of the same firms hired aggressively through the cheap-money years and have been quietly correcting ever since. Attributing the correction to AI converts an embarrassing admission into a story about the future. The announcements almost never separate the two motives, which is precisely why you should not treat them as data about your own business.
Does any of this describe a firm of ten people?
Almost none of it. Walk the logic through your own payroll.
A redundancy programme assumes surplus. In most small firms, and in construction SMEs especially, there is no surplus. ONS construction data describes an industry that has spent years short of people rather than long of them, and CITB has been making the same point about the skills pipeline for longer still. The typical ten-person firm is under-resourced on administration, not over-resourced on production. The quotes go out late not because anyone is idle, but because the person writing them is also running two jobs and the VAT return.
So the correct reading of the headlines is not that AI removes jobs and you should be nervous. It is that the technology which lets a bank shrink a department lets you grow without adding one.
What is the small-firm translation of the same technology?
The table below is the whole argument in one place.
| A large firm | A firm of ten | |
|---|---|---|
| Starting position | Thousands of seats, some redundant | Every seat visibly busy |
| What AI changes | Same output from fewer people | More output from the same people |
| The visible action | A redundancy programme | The admin hire you no longer make |
| The honest risk | Cutting muscle along with fat | Automating a process that was already broken |
| The number that matters | Payroll reduction | Hours returned and response times |
The bottom-right cell is the discipline. If you cannot say which hours came back and what happened to your enquiry response times, you are running on anecdote, which is the investor-call version of the story at a smaller scale.
Which tasks get displaced, and which people?
Be honest about this distinction, because your staff will see through anything less.
Tasks are displaced first, and the displaced tasks are specific: retyping information from one system into another, formatting documents, drafting routine correspondence, chasing people for things they promised to send, filing photographs, assembling reports from scattered messages. List what your office actually does for a week and a visible slice of it is this material.
People are a different question. In a small firm the honest answer is that automation rarely removes a person, because nobody was doing only the automatable work. The administrator whose retyping disappears becomes the person who handles exceptions, chases the genuinely difficult items and applies the judgement the role always needed. But do not pretend the question never arises. A role built entirely on moving data between systems will change, and the fair move is to say so early and retrain deliberately rather than by ambush. Given that the wider industry needs more skilled capacity rather than less, redeployment is a better trade than redundancy for almost every SME that faces the choice.
What should you actually do this quarter?
Three moves, none of which requires an announcement.
First, write down the marginal admin of a new job: every administrative task that arrives when you win one more project. That list, not the headlines, tells you where automation pays.
Second, automate the two worst items on that list before you recruit for them. The real cost of hiring in the built environment is a salary plus recruitment, management, equipment and risk, while an automation that absorbs the same growth costs a fraction of it. The economics of AI in construction now favour exactly this substitution at small scale.
Third, change the next job advert. If administration stops growing with turnover, the next hire can be a fee-earner or a site producer instead of a coordinator.
The large firms are describing their problem, not yours. Denial is the wrong response to this technology, and so is panic. The small firms that get it right will announce nothing at all. They will simply stop making the hire that used to be inevitable, and their competitors will wonder how they carry bigger workloads with the same office.