Every week seems to bring another AI product.
For an SME, that creates a difficult question. Where should you actually start?
Our recommendation is simple. Do not start with AI. Start with a problem.
Ask your employees:
Those answers can reveal useful opportunities.
Your first AI project does not need to transform the business.
Choose something that has:
Then test whether AI genuinely improves it.
Suppose an employee spends three hours preparing a recurring report. Record that before making changes.
If AI helps reduce the task to one hour, you now have evidence of an improvement. But go one step further.
Time saved only becomes business value when you understand what that time enables.
You may examine a process and discover that simple automation would solve the problem. Or Microsoft 365 already has the feature you need.
Perhaps the process itself is unnecessarily complicated. That is a good outcome.
The objective should be to improve the business, not find somewhere to use AI.
Before using AI with business information, understand what information will be involved and who should have access.
For businesses using Microsoft 365 Copilot, Microsoft states that existing identity and permissions continue to apply to what Copilot can access.Â
This is one reason good information management needs to accompany AI adoption.
If your first project works, decide what comes next.
At Commuserv, we believe this is a more practical path for SMEs.
Understand it.
Choose the right technology.
Test the improvement.
Measure the outcome.
Then build from there.
Choose a repeated, time-consuming task with a clear owner, suitable information and a measurable outcome. Examples can include producing a first draft, summarising internal material or preparing a recurring report.
Understand the current process first. If a simpler workflow change, existing Microsoft 365 feature or basic automation solves the problem more reliably, AI may not be necessary.
Start with a limited pilot rather than a company-wide rollout. The appropriate budget depends on licences, integration, information readiness, risk and training, so define the problem and expected value before setting the investment.
Consider the sensitivity of the information, who can access it, the impact of an incorrect answer, privacy, security and how employees will check outputs before using them.
A focused pilot often needs enough time for employees to move beyond initial experimentation and use the tool in real work. Define the use cases and measures first, then review adoption and results at agreed intervals.
