How I Built a €3,200 Per Month AI Automation Agency Starting With $89 in Tools: Full Process Documented

A one-person AI automation agency serving local businesses in Europe. Month 1 revenue: €840. Month 3 revenue: €3,200. Total tool spend to reach month 3: $89 per month ($267 total over three months). This is the complete guide including which services to offer, how to price them in euros and dollars, how to find clients who will pay, the three services that generated all the revenue, and the automation setups with actual code.

🔧 Tools mentioned in this article

Make.com

Visual automation platform used to build client automations — Core plan at $9/month handles most small business automation needs

www.make.com

Visit

ChatGPT API

API used to add AI intelligence to Make.com automations — billed per token, approximately $20-40/month at client volume

platform.openai.com

Visit

Claude

Used for writing client proposals, automation documentation, and troubleshooting complex automation logic

claude.ai

Visit

Airtable

Database layer used in most client automations — free tier sufficient for most small business clients

airtable.com

Visit

Tool Costs Per Month

• Make.com Core plan: $9/month (€8.30 / £7.10) — handles up to 10,000 operations per month
• ChatGPT API: $35-45/month (€32-41 / £27-36) — varies by client volume, billed per token
• Claude Pro: $20/month (€18.40 / £15.80) — for proposal writing and troubleshooting
• Airtable: $0 — free tier sufficient for small business client databases
• Notion: $0 — free tier for documentation and client portal
• Total monthly: $64-74/month (€59-68 / £50-59)
• Note: clients were billed for their own tool subscriptions separately — these costs are for the agency tools only

The Three Services That Generated All Revenue

The original plan was to offer five different automation services. After month 1 it became clear that three specific services were sellable to small businesses and two others required too much trust-building and sales time to be worth the effort at this stage. The three that worked were lead response automation, invoice and follow-up automation, and social media scheduling automation. All three solved a problem every small business has and none of them required the client to understand how AI worked.

Service 1: Lead Response Automation — €650 Setup + €120/Month Retainer

# Service 1: AI Lead Response Automation
# What it does: responds to new leads within 2 minutes with personalized email
# Tools: Make.com + ChatGPT API + client's email/CRM

## How it works (explain to clients in simple terms):
'When a new contact form submission arrives on your website,
this system reads their message, generates a personalized reply
using AI, and sends it from your email address within 2 minutes.
The reply sounds like you wrote it and references their specific inquiry.'

## Make.com scenario structure:
Trigger: New form submission (Typeform/Google Forms/website form)
Step 1: Format lead data into structured input
Step 2: ChatGPT API call to generate personalized reply
Step 3: Send email from client's Gmail/Outlook
Step 4: Add lead to client's CRM or Airtable with status 'Responded'
Step 5: Notify client via Slack/WhatsApp that response was sent

## ChatGPT API prompt used in the automation:
---
You are a customer service representative for [Business Name],
a [business type] based in [location].

A potential customer has submitted this inquiry:
{lead_message}

Write a professional, warm reply that:
- Addresses their specific question directly
- Mentions 1-2 relevant services from: [list client's main services]
- Invites them to book a call or visit using: [client's booking link]
- Is signed by: [client's name], [title]

Tone: [friendly/professional/casual — match client voice]
Length: 150-200 words
Do NOT use AI-sounding phrases or generic templates.
---

## Pricing:
Setup fee: €650 ($710 / £609) — one-time
Monthly retainer: €120 ($131 / £112) — for monitoring and updates
Time to build and deliver: 6-8 hours
Hourly effective rate: approximately €81/hour on setup

## Client results that make renewal automatic:
- Client 1 (property management firm): 34% faster lead response time,
  reported closing 2 additional deals in month 1 they would have lost
  to faster-responding competitors
- Client 2 (dental clinic): reduced receptionist time on email by 40 minutes/day

Service 2: Invoice and Payment Follow-Up Automation — €480 Setup

# Service 2: Invoice Follow-Up Automation
# What it does: automatically sends payment reminders at defined intervals
# Tools: Make.com + ChatGPT API + client's invoicing tool

## Sequence that runs automatically:
Day 0: Invoice sent by client (triggers automation)
Day 3 before due: Friendly reminder email sent
Day 0 (due date, if unpaid): Payment due notification
Day 3 after due: Overdue notice with late fee information if applicable
Day 7 after due: Final notice, client notified to handle personally

## Make.com structure:
Trigger: New invoice created in Xero/QuickBooks/FreshBooks
Scheduled checks: daily check for invoices at each day threshold
ChatGPT API: generates tone-appropriate reminder (friendly to firm progression)
Email send: from client's email, appears to be manually written
CRM update: marks invoice status at each stage

## Result for clients:
Average reduction in days to payment: 11 days faster
Client (freelance designer): reduced overdue invoices from 8 per month to 2

## Pricing:
Setup fee: €480 ($524 / £450) — one-time
No retainer — system runs without maintenance once set up
Clients often upgrade to retainer for feature additions after 2 months

How Clients Were Found Without Cold Calling

• Primary channel: LinkedIn posting 3x per week about automation case studies and time-saving examples — 4 of 7 clients came from LinkedIn
• Secondary channel: local business networking event in Berlin — met 2 clients in person, closed both within 2 weeks
• Referral: one client referred their accountant who became Client 7 on the strength of the invoice automation results
• Cold email: 47 sent, 3 replies, 0 closes — abandoned after month 1 as too low conversion for time spent
• The LinkedIn content that generated the most inbound: posts showing before/after automation workflows with specific time savings numbers, never general AI claims

Mistakes That Cost Time and Revenue

• Mistake 1: Building an automation in week 1 without scoping the client's existing tools first — spent 8 hours building a Zapier integration for a client whose email was on a platform Zapier did not support. Rebuilt in Make.com, 6 hours lost.
• Mistake 2: Charging too little in month 1 — first client paid €350 setup for work that took 12 hours. Raised prices after discovering competitors charged €600-900 for equivalent setups.
• Mistake 3: Not documenting automations for clients — when Client 3 needed a change and contacted the agency, 2 hours were spent reverse-engineering work done 6 weeks earlier. Documentation template now created for every project.
• Mistake 4: Over-promising AI capability to a client — told a restaurant owner the AI would handle complex custom orders. It could not. Spent 4 extra hours scoping limitations and rebuilding expectation.

Final Thoughts

The path from $89 per month in tools to €3,200 per month in revenue took three months and came from solving real, specific problems that small businesses face every day. The AI is not the product. The time saving is the product. Every successful client conversation started with the same question: what task do your staff do that is repetitive and time-consuming? The AI automation is the answer to that question and the price is justified by the time it saves rather than by the technology it uses.