I run a small business. Where do I actually start with AI?
Pick the single most expensive recurring task in your week and apply AI to that one. Not everywhere at once.
The biggest mistake small business owners make with AI is trying to "AI the whole business" in a quarter. The result is half-finished experiments and no money saved. The fix is dull and works: pick one task, automate or accelerate it, measure, then pick the next.
A framework that holds up:
1. List your recurring tasks. Anything you or your team does more than once a week.
2. Estimate the hours you spend on each per month.
3. For the top three, ask: could AI cut this in half? Be honest. AI is bad at trust-heavy, judgement-heavy work and good at first drafts, summaries, lookups, and pattern-matching.
4. Pick one and run a two-week pilot. Define success before you start. Did it actually save hours? Did the output quality hold?
5. Either keep it and move to the next task, or kill it and try a different task.
What to avoid: building agents that touch live customers in week one. The risk is real and the upside is small. Build internal tools first. Once the team trusts the system on internal work, then move outward.
Our AI for Revenue Generation cohort walks small operators through exactly this — picking the right task, packaging an offer, and pricing it.