Key points
- Running a small UK business, you hear AI could save hours or cut costs. The question is finding an adviser who helps rather than sells.
- You may not need outside help yet, since tools you already pay for may include AI. Paid help makes sense when you know the problem but have no time or expertise to fix it, when the job touches several systems, or when an error would be expensive.
- Before any meetings, describe the problem on one sheet: the job, its cost now, and the one measure to improve. Do this yourself, since the first adviser you talk to may bend it towards whatever they already offer.
- Give each candidate an identical brief, and rate answers only on whether they could shift your measure. Ask each to show how they would handle a small, genuine task of yours, not a demo of their own.
- A good adviser begins by asking about your work, the outcome you need and what the problem costs, not by pushing a tool. An adviser who says part of the fix is changing how you work, not AI, is showing judgement worth paying for.
- Begin with a brief, paid project: one everyday back-office task, your actual data, and a measure and finish date set in advance. Unpaid trials tend to wander; a set price and deadline keep everyone on course.
- By the end, expect a clear decision to keep going, change tack or stop; a colleague able to tweak the system; and a copy of the data and settings. Agree in writing early which changes your team must make alone. That protects you if you part company, and is the best guard against being stuck with one firm.
- For each step in more detail, read the full guide to choosing an AI consultant on AI Consultancy Finder.
Transcript
Running a small UK business, you hear AI could save hours or cut costs. The question is finding an adviser who helps rather than sells. You may not need outside help yet, since tools you already pay for may include AI. Paid help makes sense when you know the problem but have no time or expertise to fix it, when the job touches several systems, or when an error would be expensive.
Before any meetings, describe the problem on one sheet: the job, its cost now, and the one measure to improve. Do this yourself, since the first adviser you talk to may bend it towards whatever they already offer. Give each candidate an identical brief, and rate answers only on whether they could shift your measure.
Ask each to show how they would handle a small, genuine task of yours, not a demo of their own. A good adviser begins by asking about your work, the outcome you need and what the problem costs, not by pushing a tool. An adviser who says part of the fix is changing how you work, not AI, is showing judgement worth paying for.
Begin with a brief, paid project: one everyday back-office task, your actual data, and a measure and finish date set in advance. Unpaid trials tend to wander; a set price and deadline keep everyone on course. By the end, expect a clear decision to keep going, change tack or stop; a colleague able to tweak the system; and a copy of the data and settings.
Agree in writing early which changes your team must make alone. That protects you if you part company, and is the best guard against being stuck with one firm. For each step in more detail, read the full guide to choosing an AI consultant on AI Consultancy Finder.