AI in 60 Minutes — Your First Real Workflow · Lesson 4 of 6
Lesson 4 — Build a daily-use workflow
15 minutes. Now we turn your prompt into a habit.
---
A prompt you used once is a curiosity. A prompt you use every day is a workflow. The difference is intentionality — *when* you open it and *what* you put in.
Workflows that stick are anchored to a specific moment in your day. Vague intentions ("I'll use AI more") die in a week.
Pick one:
- 8:00 AM — first coffee, before the day starts. Triage yesterday's leftover emails.
- End of every customer call — paste the call notes, get back a follow-up draft.
- Friday at 4pm — paste this week's customer interactions, get back a summary for your business partner / spouse / file.
- Every time you get a quote request — open the workflow, paste the request, get the response.
- Monday morning — paste this week's calendar, get back a "what to prepare for" list.
Be specific. Vague is the enemy.
For each workflow, three things have to be obvious:
- What triggers it. A new email lands. A call ends. A timer hits. A request arrives.
- What you paste in. The email body. The call notes. The form submission.
- What you do with the output. Send it. Save it. Edit it. Pass it along.
Write it down. Three sentences.
Example:
> Trigger: Every time a customer email arrives that's not urgent. > Input: The email body, pasted into my saved "customer reply" prompt. > Output action: Read the AI draft, edit anything that sounds wrong, send it via Outlook.
That's a workflow. You can do this in real life.
The biggest reason workflows die is friction. If opening the AI tool takes more than 10 seconds, you'll skip it.
Three friction-reducers:
- Pin the AI tool in your browser — Claude or ChatGPT or Copilot, pinned tab. Always one click away.
- Save the prompt somewhere two-clicks-accessible — Notes app, Notion, Stickies — wherever you can copy-paste it instantly.
- Bookmark a specific "starting" conversation — in Claude, the same project. In ChatGPT, the same custom GPT or saved prompt. So you're not retyping the role and constraints every time.
Pick your moment. Pick your trigger. Pick your input source.
Right now — even if it's not the "right" time of day — run through the workflow once end to end. Open the tool. Paste the prompt. Paste a real input. Read the output. Do whatever you'd normally do with it.
If it worked: you have a workflow. Tomorrow at the anchored moment, do it again.
If it didn't: refine the prompt, or pick a different trigger.
The first week of any AI workflow is awkward. You'll forget to open it. You'll over-edit the output. You'll wonder if it's actually saving time.
Three things to know:
- It takes 5-10 reps before it feels automatic. A new habit always does.
- The output gets better when you edit and re-run. The model learns from how you reshape its drafts.
- The time savings are real but invisible. They show up as "I have more time" rather than a stopwatch number. Trust the process for two weeks.
When this workflow is humming, pick a second one. Same template. Different trigger. Build slowly.
Three good next picks:
- Meeting summarizer. Paste rambling notes, get clean summary + action items.
- Quote request reader. Paste form submission, get a structured response with questions to ask back.
- Customer onboarding email. Paste the new customer's info, get a personalized welcome.
By the end of the month you'll have three workflows running. That's enough to feel like a different operator.
---
One more lesson. Five mistakes I see SMB owners make in their first month with AI. So you don't make them.
→ Lesson 5 — Don't make these mistakes
Previous: Lesson 3 — The first prompt that pays off · Next: Lesson 5 — Don't make these mistakes