Where do I even start with AI?
You've heard the hype. You've probably even tried AI a little. And you're stuck on the same question every owner hits: where do I actually start?
The honest answer is short: you don't start with a tool. You start with one problem that's eating your time, you fix that one thing, and you prove it works before you go further.
Find your biggest time sink about 30 minutes
You're not picking a tool yet. You're picking the one repetitive thing costing you the most hours. Most owners don't know this without looking — that's what this step is for.
How to find it
- List every repeat task. Write down the things you do weekly that are manual and predictable — answering questions, drafting, chasing invoices, booking appointments, typing data around. Five to ten is normal.
- Score each one. Give every task a 1–10 for hours it costs you, and a 1–10 for how much you dread it.
- Pick the top scorer. The task with the highest combined score is your one. If two are close, pick the one with the clearest steps — the easier it is to describe, the easier it is to automate.
Done when: you've written down one task and how many hours a week it costs you.
Match it to a free tool about 1 hour
Match your one pain to the right kind of tool, then set it up and run one real task with it. Signing up is quick — the hour is feeding it your real context.
How to do this step
First, which tool fits your pain:
| Your pain | Where to start |
|---|---|
| Writing, drafting, thinking | A general AI assistant (ChatGPT or Claude, free tier is enough) |
| Answering the same customer questions | A simple chatbot on your site or inbox |
| Invoices / bookkeeping | The AI features already in your accounting software |
| Scheduling | An AI scheduling tool (Calendly, Acuity) |
| Typing between systems | An automation tool (Zapier / Make, free plans exist) |
Then set it up properly — this is where the hour goes:
- Create the account on the free tier. Don't pay for anything.
- Give it real context. The tool works best when it knows your business. Paste a real example of the work — a real email you'd reply to, a real invoice, a real post.
- Run one real task end-to-end. Not a test. The actual task from step 1. Draft the real email, answer the real question, write the real post.
- Edit its output. Treat the first draft as a starting point. Fix what's wrong. That's normal and expected.
Done when: you've used the tool on your real task every day for two weeks.
Measure what actually changed 2 minutes a week
After two weeks, you're answering one question: did this save me time, and how much?
How to measure it
- Count the hours. How many hours a week did this task take before? How many does it take now?
- Judge the quality. Is the output good enough to ship or send? If it needs heavy editing, factor that into the time saved.
- Do the math. If you're saving 3+ hours a week at $75/hour, that's over $11,000 a year back in your pocket.
If you didn't save time, the tool was wrong for the problem — not a sign to give up on AI, just a reason to try a different angle.
Done when: you know the hours saved per week and whether it's worth keeping.
Build outward, one win at a time ongoing
Once one thing works, repeat the same shape with the next pain. Each win proves it and shows you where to go next.
How to keep going
The businesses that get the most from AI aren't the ones with the most tools. They're the ones with a rhythm of small, measured wins. Go back to the list you scored in step 1, pick the next-highest pain, and repeat steps 2 and 3.
Done when: you've added one more workflow that saves you time.
Two things to keep in mind
Your data matters. When you test tools, know where your data goes. Don't feed customer lists, financial records, or anything private into a tool you haven't checked. If you want AI on your own machines instead, that's covered in the data guide.
You don't have to do this alone. If you'd rather skip the trial and error, or you've hit one painful task you want solved properly, that's the kind of thing I build for a fixed fee. The resources here are free for you to use either way.
Audit your biggest time sink → pick ONE tool → test 2 weeks → measure hours saved → expand. That's the whole starting point. No hype, no big-bang plan, no technical degree required.
I'm David. I build AI that runs on your machines.
Twenty years in engineering leadership, now helping small and mid-size businesses put AI to work without handing their data to a cloud. If you'd rather I build it than you learn it, let's talk.
Book a 30-minute call