Learning Path
AI Automation Agency Path
An AI automation agency does not sell software — it sells a workflow a small business no longer has to do by hand. This path teaches the agency business in order: what the service is, how the three automation tools compare on cost and limits, how to pick one niche and one workflow, build a demo, price a paid pilot, and win a first client without doing free work.
How this path works. Each lesson below is a short summary of the topic it links to, with an exercise that proves you can do the work. Work through them in order, do the exercise, and mark each lesson complete when you have read the guide and done the work. The two-question check under each lesson is for yourself — it is not scored and it is not a certificate. Your progress is stored only in your own browser, in local storage — no account, nothing sent to us. Completing this path gives you no certificate and no qualification, and it promises no income: the proof is the finished work from the capstone lesson, not a certificate.
What you will learn
- What an AI automation agency sells, and what it does not.
- n8n vs Make vs Zapier on cost, limits, and which to learn first.
- Pick one niche and one workflow — the only starting position that works.
- Build a demo workflow as the proof asset.
- Price a paid pilot, find a first client, and count the running costs.
Your progress
0 of 7 lessons complete.
The lessons
Read each lesson summary, then open the full guide it links to. The summaries are written for this path — they do not repeat the guides, they prepare you for them.
Lesson 1: What an AI automation agency actually sells
An AI automation agency sells a business outcome, not a tool. A small business does not buy n8n or a language model — it buys the invoice that processes itself, the inbox that answers itself, the lead that gets followed up without anyone remembering. Your product is a workflow that takes a real process off a real person's desk, and the proof is that it works on their data, not on a demo slide.
This lesson defines the offer precisely: what counts as an automation service, what belongs in the first package, and what does not sell — vague AI promises, dashboards nobody opens, and tools the client has to learn. You will learn why agencies that sell workflows last, while agencies that sell models vanish when the model changes.
Exercise: Write your agency offer in one sentence: the process, the business, and the result.
Check yourself — two questions on this lesson
1. What does the agency sell?
A workflow that takes a real process off a real person's desk — an outcome, not a tool.
2. What does not sell?
Vague AI promises and tools the client has to learn.
Lesson 2: n8n vs Make vs Zapier — cost, limits, which first
The three tools do the same job with different costs and limits, and the choice matters because your pricing depends on it. n8n is the flexible, self-hostable option with a steeper learning curve. Make is the visual middle ground. Its Core plan costs $12 a month on annual billing or $16 a month monthly, for 10,000 credits, last checked 15 November 2026. Zapier is the simplest to start with and the most expensive as volume grows. No dedicated comparison existed in the observed market when this path was mapped; the comparison is the differentiator.
This lesson compares the three on the numbers that decide your first year: monthly cost at real volumes, operation limits, how each one scales, and what each one is bad at. You will learn the rule — pick one, learn it properly, and let the client's volume decide whether the choice changes later. Two tools half-learned are worth less than one tool you can run in your sleep.
Exercise: Price one workflow on all three tools at your expected volume, and choose your first tool.
Check yourself — two questions on this lesson
1. What is Make Core's price, last checked 15 November 2026?
$12 a month annual billing, or $16 monthly, for 10,000 credits.
2. Why learn one tool properly first?
Two tools half-learned are worth less than one you can run reliably.
Lesson 3: Picking one niche and one workflow to offer
An agency that serves everyone serves no one. One niche — estate agents, dentists, local shops — and one workflow inside it is the only starting position that works, because the same problem repeated is a product: you build it once, learn its edge cases, and sell the same fix to the next business with the same pain. A general agency has to rebuild its knowledge for every client; a niche agency reuses it.
This lesson picks your niche and your workflow: which businesses have a repetitive process they will pay to remove, which workflow you can build today, and how to say no to the work that pulls you back to general. You will learn the test for a good niche — ten businesses with the same problem within reach of your outreach — and the test for a good workflow — you can demo it on real data in a week.
Exercise: Name your niche and your one workflow, and list ten businesses that have the same problem.
Check yourself — two questions on this lesson
1. Why one niche and one workflow?
The same problem repeated is a product — you build once and sell it again.
2. What is the test for a good niche?
Ten businesses with the same problem, within reach of your outreach.
Lesson 4: Building a demo workflow — the proof asset
Nobody buys an agency from a promise; they buy it from a working demo. Your demo workflow is a real process, built end to end, running on realistic data — an inbox triaged, a form followed up, an invoice chased — that you can show a prospect in five minutes. It is the proof asset the whole path is built around: without it you are selling a claim, with it you are selling evidence.
This lesson builds the demo properly: pick the workflow from the last lesson, build it on the tool you chose, run it until it breaks, fix the breaks, and package it so a prospect can see the before and after in one screen. You will learn what not to demo — a toy example with fake data, a workflow you have never run — because the first question a good prospect asks is what the demo does when it fails.
Exercise: Build your demo workflow end to end and run it on realistic data until it breaks once.
Check yourself — two questions on this lesson
1. What is a demo workflow for?
Proof — evidence the workflow works on real data, not a claim.
2. What should a demo never be?
A toy example with fake data that you have never run.
Lesson 5: Pricing a paid pilot — not free work, not a big contract
A paid pilot is the agency's first sale: a small, fixed-scope build at a real price, with a clear result and a clear end date. It is not free work — free work teaches the client your work is worth nothing — and it is not a big contract either, because no sensible business signs a big contract with an unproven agency. The pilot is the bridge: small enough to say yes to, real enough to prove the workflow.
This lesson prices the pilot from your costs: hours to build, tool costs for the month, and a margin that keeps you in business, with the scope written down so the pilot cannot quietly become an open-ended project. You will learn what the pilot must prove — that the workflow works on the client's data — and what happens after it works: the retainer conversation, in which the client pays monthly for the workflow to keep running.
Exercise: Price your pilot: build hours, tool cost for the month, margin, and the scope boundary.
Check yourself — two questions on this lesson
1. What is a paid pilot?
A small, fixed-scope build at a real price, with a clear result and end date.
2. Why not do a free pilot?
Free work teaches the client your work is worth nothing, and it cannot become a retainer.
Lesson 6: Finding the first client — outreach and the free audit
The first client comes from the same niche you picked, through outreach that offers something before it asks for anything. The free audit is the offer: you look at a business's public process — its contact form, its inbox, its follow-up — and write down what is slow, what breaks, and what an automated workflow would do instead. It is proof of your thinking before they pay, and it only works when the audit is honest, even when the answer is that the business does not need you.
This lesson teaches the outreach sequence: who to contact, how to open with the audit instead of a pitch, how to run the audit on a real prospect's public data, and what to do when the answer is no — learn from it, move to the next prospect, and keep the audit as a sample. Warm network first, audited prospects second, cold outreach last — the order that gets the first client without burning the niche.
Exercise: Run a free audit on one real local business: three slow points, and what a workflow would do for each.
Check yourself — two questions on this lesson
1. What is the free audit?
A written review of a prospect's public process, showing what is slow and what automation would do.
2. Who do you contact first?
Your warm network, then audited prospects, then cold outreach.
Lesson 7: Running costs and wrapper risk — subscriptions, API metering, what happens when the model adds your feature
An automation agency has running costs that a freelancer does not: the automation tool's subscription, the AI model's usage charges, and the time spent keeping workflows alive when an app changes. Those costs sit inside your price, and if you forget them, a profitable pilot becomes a monthly loss. The maths is simple — tool cost plus model usage plus maintenance hours, per client, per month — and it must be done before you quote, not after.
This lesson covers the second risk as well: the wrapper risk, which is that the model you built on adds your feature, and your product stops being needed. The defence is owning something the model does not have — a workflow, your client's data, or the distribution to their customers — not the feature alone. It ends with the capstone: a demo workflow and a one-page paid-pilot offer, with the running costs shown.
Exercise: Capstone: finish your demo workflow and a one-page paid-pilot offer, with monthly running costs listed.
Check yourself — two questions on this lesson
1. What costs sit inside an agency price?
The tool subscription, model usage charges, and maintenance hours per client per month.
2. What defends against wrapper risk?
Owning a workflow, client data or distribution — something the model does not have.
Try the tools
These free tools let you practise what the lessons describe — everything runs in your browser, and nothing you type is sent anywhere.
- Startup Cost Calculator — total the cost of starting the agency
- ROI Calculator — judge a pilot by its return
- Freelance Project Pricing Calculator — price the pilot with a buffer
Check what you learned
7 questions, all based on the lessons above. Answer them all, then check your score — each answer explains itself, so a wrong answer teaches as much as a right one. The quiz is practice, not a test: there is no pass mark and no certificate.
Answer all 7 questions, then check your score.
Download your notes
Download a plain-text summary of this path: the lesson list, which lessons you marked complete, and your quiz score. The file is generated on your device — nothing is uploaded.
Key terms in this path
The terms this path uses are defined in the Glossary, each with an example. If a lesson uses a word you do not know, look it up there first.
Frequently asked questions
Do I need to code to start an AI automation agency?
No. The main tools — n8n, Make and Zapier — are visual and no-code. What you need is the ability to map a small business process, build it as a workflow, and explain the result in plain language.
Which automation tool should I learn first?
It depends on your niche and budget. This path compares n8n, Make and Zapier on cost and limits, with current figures — Make Core, for example, is $12 a month on annual billing or $16 monthly for 10,000 credits, last checked 15 November 2026.
What does an AI automation agency charge for?
A pilot project first — a small paid build that proves the workflow works — then a monthly retainer to run and maintain it. The path teaches pilot pricing in its own lesson.
What is wrapper risk?
The risk that the AI model you built on adds your feature, and your product stops being needed. The defence is owning a workflow, data or distribution the model does not have — not the feature alone.
Can I start an agency with no clients?
Yes — that is what the demo workflow is for. You build one real workflow, show it to businesses with the same problem, and sell the pilot. Nobody buys an agency; they buy a workflow that works.