The HAND Framework for Monday Automation

Doing Monday automation by feel gives uneven results. One week you hand off a job and it works. The next you hand off another, skip the testing and it quietly goes wrong. A fixed routine fixes that.

This post gives you one: HAND. It is four letters, four moves, and you can repeat it for every weekly task. The name is just a memory aid, not a method with research behind it.

Lesson 6.6 walks you through it step by step: Hand off the job you do every Monday.

HAND stands for Hunt the job, Annotate the steps, Nail the first three runs, Deliver a note. Do them in that order, every time.

H is for Hunt

Hunt for the right job. Before touching AI, list every recurring Monday task and score each from zero to three on four questions: is it on a fixed rhythm, are the steps the same each time, is a mistake easy to catch, can you check the result in a minute.

Pick the highest score. If you tie, pick the one you dislike most. Write the winner at the top of a fresh page. This step takes ten minutes and prevents the most common failure, which is automating something that was never stable.

Output of H: one named job and a one line reason.

A is for Annotate

Annotate means writing the job as numbered steps in plain words, as if training a new assistant. Include the trigger, the inputs, the finished result and the stop rules.

A sample, for illustration. Say you send a weekly status note to a client:

1. Open last week's notes folder.
2. List what was finished, what slipped and what is next.
3. Write five lines in a friendly tone.
4. Leave it as a draft. Never send.

Four lines is enough. The stop rule in line four is as important as the steps. Annotation also exposes the habits you can drop, and any decision hidden inside the job that only you can make.

Output of A: a numbered note you could hand to a stranger.

N is for Nail the first three runs

Nail means: do not trust the job until it has survived real weeks. Run it by hand with AI once, then let it do the first three scheduled runs while you check each one against your own version.

For each run, log the differences in a simple table.

The table above is an invented example of the format. Your rows will differ. The point is that every difference becomes a new line in the note, so the note improves with each run.

Output of N: three logged runs and a note that has been corrected.

D is for Deliver

Deliver has two halves. First, deliver the schedule: put the job on a rhythm so you stop starting it. A repeating Google Calendar event is the lightest option, since Google lets an event repeat on chosen days. A scheduler that runs the job itself is the heavier option. The Hermes Agent documentation, for instance, describes recurring scheduled tasks that send results back to your chat.

Second, deliver a note. The last step of every run should send you a short completion message: what ran, what looked odd, what needs you. No note means no visibility, and no visibility means you will not find a failure until someone else does.

Output of D: a schedule and a completion note you read every week.

Running the loop weekly

HAND is one pass for one job. Once a job is delivered, a weekly review takes five minutes.

• Read the notes from the last week. Did it run?
• Did you change anything by hand afterwards? If yes, fix the note.
• Did any step become unnecessary? Remove it.
• Is the job ready for a bigger scope, or is it fine as is?

Then go back to H for the next job. Do one job at a time. If you want examples of what to pick, see twelve Monday automation examples.

Common ways to break the framework

• Skipping H. You pick a job because it is easy to describe, not because it is stable.
• Skipping the stop rules in A. The AI does more than you intended.
• Skipping N. The first real problem lands on a customer.
• Skipping the note in D. The job fails and nobody sees it.

A full pass, start to finish

Here is HAND applied to one invented job, so you can see how the pieces connect. Say you are a consultant and every Monday you prepare a one page brief for each meeting that week.

Hunt. The job is on a fixed rhythm, the steps are stable, a mistake is easy to see and the stakes are low because you read the brief before the meeting. It scores well.

Annotate. You write seven numbered lines: open the calendar, list the week's meetings, find earlier notes for each attendee, summarise the last conversation, list open questions, write the brief in under 150 words, save it in one folder. Stop rule: do not contact anyone.

Nail. The first run misses a meeting with no earlier notes. You add the line "if there are no earlier notes, say so." The second run is too long, so you add a word limit. The third run is clean.

Deliver. You set a weekly rhythm and add a final line: "Tell me how many briefs were made and which meetings had no notes."

Total change to your Monday: you read a short note and the briefs instead of building them. The details are invented, but the sequence is the part to copy. Notice that most of the effort sat in A and N, not in the scheduling at the end.

Use it this week

Print the four letters, pick a job today and finish H before lunch. For the guided version, with the prompts, see Lesson 6.6: Hand off the job you do every Monday. It pairs well with Lesson 6.1, plan your day in five minutes.

Questions people ask

Is HAND a proven method?

No. It is a simple checklist built from common sense about testing and monitoring. Treat it as a routine, not as research.

How long does one pass take?

It depends on the job. The checking stage runs across three weeks by design.

Can I use it for jobs that are not on Monday?

Yes. Any recurring task fits the same four steps.