Hot Hand Media / Foundational Doc

Automation 101

What automation actually is, why a prompt is not a system, and how to know what to automate first.

Most people get sold AI as the answer and then wonder why nothing in their business actually runs differently. The reason: AI is one ingredient. Automation is the kitchen. This doc covers the three things you need to know before you build anything — what to automate, what AI is and is not, and what an automation actually looks like under the hood.

PART01

Map the task before you automate it

If you can't draw it, you can't automate it

Automation does not invent a process. It executes a process you already have. If the process lives in your head, in a sticky note, or in three different brains across your team, automating it will produce three different broken versions of the same thing.

Before you touch a tool, walk the task by hand. Open the tabs. Click the buttons. Send the email. Write down every step, including the ones you do without thinking. That is your map.

Name the outcome. What changes in the world when this is done? Not "send the email" — "the new lead gets the welcome sequence within 5 minutes of opting in."
List the trigger. What single thing kicks this off? Time, event, form submission, status change, manual click.
Write every step in order. Including the ones you assume are obvious. The boring ones are where automations break.
Mark every decision point. Anywhere the path forks based on a condition. "If paid, then…" "If no reply in 3 days, then…"
Note every system involved. Email, CRM, calendar, spreadsheet, payment processor. Each one is a connection you'll have to make.

If you can't draw the task on a napkin, you can't automate it. You can only generate something that looks like the task.

PART02

A prompt is not a system

AI is non-deterministic. Automations are not.

Here's the part most people miss. AI and automation are not the same thing. They do not solve the same problem. They do not behave the same way. And when you mix them up, you build something that works half the time and breaks the other half for reasons you cannot trace.

Automation (Deterministic)

Same input → same output, every time
Runs on rules you write
Predictable, repeatable, testable
Breaks loudly when something is wrong
This is what your business actually runs on

AI (Non-Deterministic)

Same input → different output, sometimes
Runs on probability and patterns
Creative, generative, unpredictable
Breaks quietly when something is wrong
This is an ingredient, not the kitchen

AI lives inside automation. Not the other way around. You build the deterministic system first — the trigger, the steps, the conditions, the error handling. Then, at the specific point where you need creative output (write the email, classify the message, summarize the call), you call the AI. The system catches what the AI produces and routes it like any other piece of data.

A prompt is not a system. A prompt is a step inside a system. Without the system, the prompt has nowhere to go and nothing to do.

PART03

The anatomy of an automation

Trigger → action → result

Every automation, no matter how complicated it looks, is the same three pieces stacked. Once you can name these in your own workflow, you can build them in any tool — Make, n8n, Zapier, GHL, custom code, whatever.

01 / Trigger

The thing that starts it

A new form submission. A scheduled time. A payment received. A row added. A button clicked.

02 / Action

The thing it does

Send an email. Create a record. Update a status. Call an AI to write copy. Move a file. Tag a contact.

03 / Result

The thing that changed

A lead nurtured. A client onboarded. A post published. A report delivered. A decision routed.

Most automations chain several of these together. The result of one becomes the trigger for the next. A form submission triggers a contact creation, which triggers a welcome email, which (three days later) triggers a follow-up check, which triggers a tag change, which triggers the next sequence.

Build one chain at a time. Get it working end to end. Then add the next link. The biggest reason automations break is people try to build the whole machine before any one piece is tested.

Less mess starts here. Map the task. Name what's deterministic. Build trigger → action → result. Everything else is decoration.