The honest answer is: it depends entirely on what your business does day to day. Not on AI in general. Not on what you have read in the press. Not on what your competitors are supposedly doing.
That is not a dodge. It is the only correct starting point. AI is a set of tools, some of which are genuinely useful for certain kinds of work and genuinely useless for others. Working out which side of that line your business sits on takes about an hour of honest thinking, and this article will help you do it.
When AI tends to be worth the investment
AI performs well on tasks that follow a pattern. Feed it enough examples of a thing, and it can usually handle new versions of that thing faster and more consistently than a person can. That is not magic. It is pattern matching: the same task, done faster and more consistently across larger volumes.
In practical SME terms, this shows up most clearly in:
- Document processing. Reading invoices, job sheets, purchase orders, applications, and extracting the relevant data. If your team currently opens a document and types information from it into a spreadsheet or system, that is a strong candidate.
- First-draft communications. Quotes, follow-up emails, tender responses, reports where the structure is consistent and the facts change. AI produces a workable draft in seconds; a person reviews and adjusts. The time saving is in the starting, not the finishing.
- Data extraction and summarisation. Pulling key points from contracts, meeting notes, maintenance logs, customer records. Tasks where a person currently reads something in order to produce a shorter version of it.
- Scheduling and routing workflows. Allocating jobs, matching resources to requirements, flagging anomalies in data. Anything where the decision follows a defined set of rules most of the time.
The common thread is that these are pattern-based tasks. Someone in your business currently applies a known rule to new information. AI can often do that faster, with less fatigue, and at lower cost per item.
When AI probably is not worth it yet
There are two categories here, and they are different problems.
The first is tasks where the value comes from human judgment, relationship, or craft. A skilled estimator who knows a client's risk appetite and reads the room on a site visit is not performing a pattern-based task. A sales conversation that requires reading subtle signals and adjusting in real time is not pattern-based. Neither is managing a difficult supplier relationship or making a judgment call on a complex job. AI can support these tasks around the edges, but it cannot replace the core of them, and trying to make it do so usually creates more problems than it solves.
The second is a readiness problem, not a technology problem. AI systems need to learn from your data, your processes, your examples. If your processes are not documented, if the same task gets handled differently by different people depending on who is available, or if your records exist mainly in people's heads and on scraps of paper, then AI will either fail to work at all or automate the inconsistency rather than the process.
If your processes are not documented and the same task is handled differently by different people, AI will automate the inconsistency, not the process.
This readiness problem is more common in SMEs than people admit, and it is fixable. But it needs to be fixed before you introduce AI, not after.
The question that cuts through the noise
Here is the most useful question I know for working out whether AI is likely to be worth it in your business:
What does your team do manually each week that follows a rule?
Not interesting work. Not creative work. Not relationship work. Specifically: tasks where, if you wrote the rule down, someone could follow it without needing to ask questions.
If you can list three or four of those tasks, and together they account for more than a few hours a week, then exploring AI is worth your time. If you genuinely cannot list any, that is useful information too. It probably means your business runs on judgment and relationships more than on repeatable process, and AI will have limited impact until the nature of the work changes.
At Vanda Coatings, the answer to that question covered most of our administrative function. Every job generated the same kinds of paperwork, the same data entry, the same cost tracking: all done manually, all repeated as information moved between different parts of the business. When I measured it properly, I found around 30,000 points of duplicated data across our processes. That was the number that confirmed the manual approach was costing us more than we had accounted for.
We built a bespoke ERP system around our actual workflows and added AI features progressively to handle the more repetitive admin tasks. The measurable result is at least five hours a week saved across the business on administration. The people who were spending those hours on admin are now doing other things: marketing, client work, tasks that move the business forward.
The less visible benefit has been in decision quality. We now have live cost visibility across every job. Before, you would find out whether a job had gone over budget after it had finished. Now you can see it while it is still running. That kind of information does not show up as hours saved on a timesheet. It changes the decisions you make in the moment, which has a different order of impact on the business.
If you want to work through this question for your own business, see how the consulting engagement works or book a free call. No pitch, no commitment. Just a straight conversation about your specific situation.
Sources
- British Chambers of Commerce, "The Turning Point for SMEs", September 2025
- DocuSign Digital Maturity Report, 2024
- OECD, "Generative AI and the SME Workforce", 2025