Every week, another SME owner tells us some version of the same thing: “I know AI is important, but I don’t know where to start.”
It’s a reasonable position. The landscape is noisy. The vendors are overselling. And the stakes feel high — no one wants to spend money and management time on something that doesn’t stick.
Here’s how we approach it.
Start with pain, not possibility
The worst AI projects start with the technology. “We should use ChatGPT for something” leads to generic experiments that don’t connect to business outcomes.
The best projects start with a specific, felt pain:
- “We spend 40% of our admin time on ___”
- “We lose deals because our quoting process is too slow”
- “Our team can’t find the information they need without asking someone”
- “We answer the same customer questions 50 times a week”
These are operational problems that AI can solve. The question is which one to tackle first.
The four-quadrant filter
When we assess AI opportunities with clients, we map them across two axes:
Impact — how much time or money does solving this save? How much better is the outcome for staff or customers?
Effort — how much data preparation, integration work, or change management is required?
High impact, low effort wins. Every time.
The common mistake is chasing the impressive project — the AI that completely transforms a core product or replaces a complex team function. Those projects take 12+ months and carry high risk. They’re often the right eventual destination, but wrong as a starting point.
What makes a good first project
A good first AI project for an SME typically looks like this:
- Narrow scope. One task, one team, one workflow. Not “AI across our business.”
- Clear success metric. You can measure it before and after. Time saved, volume handled, errors reduced.
- Low-risk data. It doesn’t touch your most sensitive customer data or your most critical business process.
- Visible to the team. People can see it working. Trust builds from visible wins.
Document generation, internal Q&A tools, and automated response drafting consistently meet these criteria for SMEs in service industries.
The 90-day test
If a first project can’t deliver something measurable within 90 days, it’s probably too ambitious for a starting point.
This doesn’t mean the project itself needs to be complete — it means there should be a working prototype with real users, generating real signal about whether the approach is right. Early feedback is worth more than a perfect system you haven’t tested.
A note on the tools
The right tool depends entirely on the problem. We’re not wedded to any platform — our clients use a range of AI tools, and the best one is the one that solves the actual problem without unnecessary complexity or cost.
That said: don’t let tool selection be a blocker. Pick the simplest thing that might work. You can always upgrade later. Done and imperfect beats planned and never shipped.
If you’re trying to identify your own first project and want a sounding board, get in touch. We do a free 30-minute call for SMEs in the early stages of figuring this out.