Do you need to hire someone for AI? Almost certainly not yet

Once a business decides AI matters, the next question is who does it. The instinct is to hire. In a firm of twenty to fifty people, that is usually premature and expensive.

The case against hiring first

A capable AI or data person commands a salary that competes with a senior engineer, and in a smaller firm the role has a structural problem: there is not yet enough defined work to fill it.

What happens next is predictable. They arrive, spend three months discovering the data is messy, build something useful, and then run out of problems that have been specified well enough to work on. Within a year they are doing general IT, or they leave — and both outcomes are expensive.

Hire when you have a queue of defined work, not to create one.

The case for buying tools

For a large share of SME use cases, the right answer is a subscription to something that already exists. Document handling, transcription, drafting, bookkeeping automation, CRM enrichment — these are solved problems and building your own version is a poor use of money.

The failure mode is different here: buying tools with no owner. Software arrives, two people try it, it renews silently for three years. Whoever buys it must be accountable for whether it is used.

The case for fractional advice

The middle option is the one most firms do not consider: buy senior judgement for a small number of days, not a full-time salary.

It fits the actual shape of the problem. The genuinely hard questions — which problem to tackle first, whether the data supports it, whether a project will pay back, whether a vendor's claims are credible — are intermittent and high-stakes. The work in between is mostly execution your existing people can do, once someone has scoped it properly.

It is also reversible. If it turns out you need a permanent hire, you will know precisely what the role should do, which makes the recruitment far more likely to succeed.

How to tell which you need

Buy a tool when the problem is common, the outcome is obvious and no integration is required.

Get advice when you are unsure which problem to attack, when a vendor is quoting more than £20,000, when the project touches your ERP or production data, or when you have already had one pilot fail and do not know why.

Hire when you have completed two or three projects, have a backlog of specified work, and the constraint is genuinely delivery capacity rather than direction.

The funded route

Worth knowing before you spend anything: if you are a Midlands manufacturer, some of this is already paid for. Made Smarter offers fully funded impartial advice and digital roadmapping, a free leadership programme, and free technical courses covering data analytics and AI applications — alongside match funding up to £20,000.

For a firm at the "we do not know where to start" stage, taking the funded advice first costs nothing and makes any subsequent spending better targeted.