When someone asks "should my business be using AI?", I ask back: "What's the recurring task you dread most every week?" That's the starting point.
This is the hub post of the business automation series. Each topic gets its own deep-dive; this post sets the overall order.
How to pick what to automate
Not every repetitive task is a good first target. My three criteria:
- Does it recur on a schedule — daily or weekly work
- Can the judgment be written down — "when X, do Y" expressible in plain language
- Is a mistake recoverable — a human can review before it goes out
Hit all three and it's a candidate. The usual suspects:
- Review replies — answering Google reviews one by one. (Automating Google review replies covers the how.)
- Content drafts — first drafts of social posts and blog articles
- First-touch inquiry responses — drafted answers to the questions you get every week
- Report assembly — collecting the same numbers into the same summary every Monday
The rollout order — small and reviewable
Failed automations share one trait: "everything at once." The order I recommend:
- Pick one task. The most frequent, most dreaded one.
- Start with draft generation. AI writes, a human approves and sends. Fully automatic publishing comes after trust is earned.
- Run it for a week, then adjust. Collect the awkward outputs and fold them into the instructions.
- Then expand to the next task.
What it costs
Agencies quote thousands for "AI adoption consulting." The actual structure is usually:
- AI subscription: $20–200/mo
- The automation script: $0 if you build it yourself (and you can have AI build it)
- Optional glue tools: tens of dollars a month
The core expense is one AI subscription. If a quote says otherwise, ask what the difference pays for. Sometimes it's legitimate (complex system integration) — often it isn't.
Managing expectations
Being honest:
- What works: big cuts in repetitive-task time, faster response times, higher content output
- What doesn't: judgment calls on exceptions, fully unattended "it just handles everything" operation
- The surprise benefit: building the automation forces you to write down how the work is done — long-term, that documentation is the bigger asset
New experiments and their results land in the AI This Week series and the newsletter.