AI for SMEs: where to start without burning your budget
Most small businesses look at AI with a mix of fascination and anxiety. Fascination because results are real. Anxiety because the market sells rockets when there isn't even a runway.
The most common mistake
Starting with the model, not the problem. Hiring an "AI consultancy" before mapping where operational friction actually lives. Result: 30k spent on a chatbot nobody uses.
Where to start
- Map repetitive tasks. Anything done more than 10 times a week is a candidate.
- Measure the real cost. Time × salary × frequency. If under 5k a year, skip it.
- Pick one case. One. Not three.
- Prototype in two weeks. If it doesn't fit, the case is too big.
What to automate first
- Repetitive email responses
- Lead qualification
- Internal reporting
- Meeting summarization
What to avoid (for now)
- Autonomous agents without supervision
- Replacing humans in complex sales conversations
- Training your own models without data volume
How to measure
Two metrics: time saved per week and errors avoided. If neither moves after 30 days, kill the project.
AI isn't magic. It's leverage. And leverage only works with a well-chosen fulcrum.

Writes about applied AI, operations, GEO/SEO and how to turn companies into machines that keep running even when no one is watching.
