Last updated · Peter Baker
Short answer: doing nothing about AI is the right call for a UK service firm when no single repetitive task takes more than a few hours a week. When one does, waiting costs those hours every week. The free Ishigai website report takes two minutes, and the £750 audit tells you which applies, including when the answer is "don't bother".
Side by side
| Do nothing | Find out first | |
|---|---|---|
| Cost today | Nothing | Free report, or an audit from £750 |
| Cost over a year | Whatever the repetitive work already costs, every week | Credited in full against a pilot started within 60 days |
| Risk | Competitors who automate the admin quote faster and reply sooner | Spending £750 to learn the answer is "not yet" |
| What you learn | Nothing new | Where the hours go, ranked and costed |
When doing nothing is the right call
- No single task eats more than a few hours a week. The saving won't cover the build.
- Your data lives in people's heads or inboxes, not in a CRM or shared system. Fix that first; AI needs something to read.
- You're about to change core systems. Automate after the move, not before.
The worked example has a whole section of these calls. In it, four ideas an eighteen-person recruitment agency might reasonably consider get a "don't bother", from automating rejections to replacing the ATS. Read the don't-bother calls.
What waiting actually costs
The cost of waiting is the repetitive work your team already does: CRM notes typed up after calls, the same status update written every week, monthly client reports built by hand. Add up the hours one person spends on one of those each week and multiply by their hourly cost. That number, every week, is what doing nothing costs.
If that number is small, wait. If it isn't, the free website report takes two minutes, and the AI Opportunity Audit from £750 maps the rest.
Measure it this week, for free
You don't need anyone's help to find out whether waiting is costing you. Try this for one week:
- Pick one repetitive task: CRM notes, chasing documents, the weekly status update or a client report.
- Ask the person who does it to note the time each time they do it, for five working days.
- Multiply the weekly total by 46 working weeks and by that person's hourly cost.
- Compare the yearly figure with £2,500, the starting price of a 30-day pilot on one workflow.
If the yearly cost of the task is well under £2,500, doing nothing is the sensible choice for that task. If it's well over, you have your first project, and a number to measure the pilot against.
A worked example, with illustrative numbers. A recruitment consultant spends 4 hours a week typing up CRM notes after calls, and their time costs the agency about £30 an hour. That is 4 × 46 × £30 = £5,520 a year on one task for one person. Across three consultants it is £16,560. Either figure is well over the £2,500 pilot price, so for that agency waiting is the expensive option.
What changes if you wait a year
The tools will be cheaper and better in a year. The hours your team spends on repetitive work this year will not come back. Waiting makes sense when the hours are small or your data isn't ready; it costs you when one task is already eating a day a week.
Frequently asked questions
Is it too early for a small firm to use AI?
Not if one repetitive task takes more than a few hours a week. If none does, waiting is sensible, and the audit will tell you so.
What is the cheapest way to find out?
The free website report takes two minutes and costs nothing. The AI Opportunity Audit from £750 covers the whole operation.
Will an audit just tell me to buy something?
No. A roadmap that recommends everything is a sales document. The worked example shows the "don't bother" calls we make.
How do I know if one task is worth automating?
Time it for a week, multiply by 46 working weeks and the hourly cost of the person doing it, and compare the yearly figure with £2,500, the starting price of a 30-day pilot. Well under that, leave it. Well over, it is a candidate.
What if my data is not ready for AI?
Then fix the data first. If client details live in inboxes and spreadsheets rather than a CRM or shared system, AI has nothing reliable to read. An audit will say so, and that is often the most useful finding.
Related reading
- Pricing: every Ishigai price on one page.
- Worked example: what the £750 audit actually looks like.
- What an AI audit costs in the UK: 2026 market ranges, with sources.
- Ishigai vs a £5,000 consultancy audit: what the price difference buys.
- Ishigai vs building it yourself: Zapier, n8n and when DIY wins.
The weekly note. One short email a week from Akashuin, our research blog: what UK firms are actually doing with AI, the numbers behind it, and one thing you can try that week. Free, and one click to unsubscribe.