Learn AI

Everything you need to know, before you spend a dollar.

Five short reads, written for business owners — not engineers. If you finish these, you'll know more about putting AI to work than most people selling it to you.

READ 01 · 3 MIN

What an AI agent actually is

ChatGPT answers questions. An agent does work. That's the whole difference.

A chatbot waits for you to type, thinks, and types back. An agent is given a goal — "sort today's invoices and flag anything unusual," "turn this voice memo into a week of posts," "answer every booking inquiry within a minute" — and then uses tools to get it done: it reads your email, checks your calendar, writes to your spreadsheet, sends the reply. You review the result instead of doing the task.

A chatbot is an employee you have to talk to all day. An agent is one you brief once.

Agents are built from three parts: a model (the brain), tools (hands — your email, docs, CRM, bank feed), and instructions (the SOP it follows, written in plain language). If your business has a task that follows rules, happens repeatedly, and lives in software, an agent can probably do it.

What agents are not: magic. They're wrong sometimes, they need clear instructions, and they need someone who knows how to wire them into your actual systems safely. That last part is the job.

READ 02 · 3 MIN

Automation vs. agent — which one do you need?

They get sold as the same thing. They're not, and the price difference matters.

Automation is a fixed path: when X happens, do Y. New lead fills a form → add to CRM → send welcome email → notify you on WhatsApp. No thinking involved, which is exactly why it's cheap, fast, and nearly bulletproof. Most businesses should start here.

An agent handles the steps that need judgment: reading a messy email and deciding what the client actually wants, drafting a reply in your voice, categorizing an expense that doesn't fit the rules. Agents go where flowcharts break.

Rule of thumb: if you can draw the task as a flowchart, automate it. If the flowchart keeps sprouting "well, it depends…" — that's an agent.

The best systems mix both: automations move the data, agents make the judgment calls, and a human approves anything that matters. That's how you get speed without handing your business to a robot unsupervised.

READ 03 · 4 MIN

Where AI fits in your business (a map)

Forget "AI strategy." Walk through your week and find the hours that repeat. That's the map. The usual suspects:

  • Inbox & inquiries — drafting replies, routing requests, answering the same 12 questions. Often the single biggest win.
  • Content — turning one idea (a voice memo, a call recording) into posts, emails, and scripts in your voice.
  • Back office — invoices, expense categorization, reconciliation, report prep. Boring, rule-based, perfect.
  • Sales follow-up — the leads that go cold because nobody wrote the third follow-up. An agent never forgets.
  • Research & prep — briefing docs before calls, competitor checks, summarizing long documents.

You don't need AI everywhere. You need it on the three tasks that quietly eat ten hours of your week.

The honest test for each task: does it happen at least weekly, does it follow a pattern, and does it live in software? Three yeses — automate or deploy an agent. A no — leave it human. Plenty of your business should stay human; the point is choosing which parts on purpose.

READ 04 · 3 MIN

What it costs, honestly

Three ways to get AI working in your business, in ascending order of cost and depth:

  • Off-the-shelf tools ($20–100/mo) — ChatGPT, Claude, and the AI already inside your existing software. Great for individual productivity. Limit: nothing is connected, so you're still the glue.
  • Assembled systems (hundreds to a few thousand, one-time + small monthly) — automations and agents wired into your actual tools: your inbox, CRM, calendar, accounting. This is where "I save 10 hours a week" starts being literal.
  • Custom builds (project pricing) — client-facing apps, full pipelines, anything your customers touch. Scoped like any software project, just built dramatically faster than the old way.

The real cost of AI isn't the subscription. It's the ongoing running cost of a system nobody maintains — or the ten hours a week you keep spending because you never set one up.

One more honest note: a system you don't understand is a liability. Whoever builds yours should hand it over documented, explained, and owned by you — not held hostage in their account.

READ 05 · 3 MIN

How to get your business ready

You can do this week's worth of prep before spending anything, and it doubles the value of any AI work you buy later — from me or anyone else.

  • Write down your repeated tasks. For one week, note every task you do more than once. The list is the roadmap.
  • Turn your best process into words. Pick the task you'd delegate first and write how you do it, like you're training a new hire. Agents run on exactly this kind of instruction.
  • Get your information in one place. Agents work with what they can reach. If prices live in your head and processes live in seven notebooks, start consolidating.
  • Pick one metric. Hours saved, response time, leads followed up. One number tells you if the system is working.

AI doesn't fix a messy business. It amplifies whatever structure you give it — so give it a little structure first.

Do these four things and you're ahead of 90% of the businesses "looking into AI." Then, when you're ready, the audit call is free and you'll get a straight answer about what's worth building — including "nothing yet," if that's the truth.

1/137

Next step

Done reading? Let's map your business.

A free 30-minute call. We find the tasks eating your week and I tell you exactly what I'd deploy — and what I wouldn't.

Take the free AI audit Email me for the call →