AXHUB5 cards

Start with one thing
you repeated today

From an owner who runs an online shop solo.
"The start is less grand than you think.
Just delegate to AI one task you repeated today."

We split that one time into five steps.

Combines practitioner notes, automation guides, and the personal axis of a 90-day playbook

Start with one1 / 5

01

Find the repetition

In today's work, find what you did twice or more.
The thing where you thought "here we go again."

The clearer the input and the more the pattern repeats, the better.
It's the first criterion automation guides commonly point to.

Common candidates Meeting-note cleanup · email reply drafts · review responses · quotes · weekly reports · material summaries · inquiry sorting

Source Automation-fit criteria — clear input, regular pattern (work automation guides)

Start with one2 / 5

02

Write down the steps

Pretend you're handing the task to a new hire, and write it in three or four lines.
This memo becomes the instruction itself.

As you write, how you'd been doing this task becomes clear for the first time.
That alone makes it worthwhile.

Example: review-reply rules "These are our shop's review-reply rules. 1) thank them 2) pick up specifically on the menu or situation the review mentioned 3) one line inviting a return visit. For negative reviews, apologize first, no excuses. Tone polite, two or three sentences."

Source The "AI is a new hire" view — the practice of giving context and standards first

Start with one3 / 5

03

Try it with real data

Not practice samples — put in the actual material you got yesterday.
You need about three to see the variance.

Find where what worked on samples breaks on the real thing.
That's as far as this step goes.

Like this The rules you wrote in step 2 + "Below are three reviews I got yesterday. Draft a reply for each."

Source 90-day rollout playbook — the personal version of the build-with-real-data principle

Start with one4 / 5

04

Review the result and fix it

Don't send the draft out as-is.
Check numbers, names, dates against the source.
If the tone is off, show an example and have it fix it.

Same as new-hire feedback — two or three rounds and it improves noticeably.
This "a person sees it last" structure is what split results by double.

Source Stanford Enterprise AI Playbook — review only exceptions 71% vs per-item approval 30%

Start with one5 / 5

05

Save what worked

Paste the instruction that worked into a notes app.
When the same task comes tomorrow, use it again.
A week and it's second nature.

Then, on to the second repetitive task — one each week.
An organization's 90-day playbook is, in the end, these five steps done together by many.

Three lines are enough to save Which task / full instruction text / things to watch (e.g. "if an amount is involved, always check against the source")

Source One per week · turning instructions into assets — 90-day playbook and Stanford (77% of the gap is operations)

AXHUBclosing

Do it once
and you get the feel

Actually delegating three reviews beats reading ten articles.

One thing you repeated today.
How about starting there?

See the source material at axhub.net

Sources: solo-founder and practitioner notes (perspective) · work automation guides · 90-day rollout playbook · Stanford Playbook (figures)

AXHub card No.5 — the instructions in the example boxes are sample text to copy. The figures follow the source cited at the bottom of each card.