AI Agent Business — How I Got From $0 to $10K in 4 Months

Month 1: $0. Month 2: $1,400. Month 3: $5,800. Month 4: $10,200.

No coding experience. No existing clients. No course purchased. Just a specific model, a lot of outreach, and a clear understanding of what businesses actually pay for.

Here is the full story.

What an AI agent business actually is

An AI agent is software that can perform multi-step tasks autonomously using a combination of language models (like ChatGPT), automation tools (like Make.com), and external services (like email, calendars, or databases).

In plain language: instead of a workflow that just moves data between apps, an AI agent can read a document, understand what it says, make a decision based on that understanding, and take the appropriate next action.

Examples of what AI agents actually do in client businesses:

• Read every new lead email, extract the relevant details, score the lead based on criteria you define, draft a personalised response, and book a call — all within 60 seconds of the email arriving
• Monitor a company's support inbox, categorise each ticket by type, pull relevant answers from a knowledge base, and draft responses for the support team to review and send
• Scan a property listing database every morning, identify properties matching a client's specific criteria, compile a briefing document, and email it to them automatically

None of these require a developer. They require someone who understands business processes and knows how to connect tools.

Month 1: Learning and demos (revenue: $0)

I spent the first month almost entirely on learning and building demonstration projects. No client outreach until I had something to show.

Tools I learned:

• Make.com — four hours of YouTube tutorials per week for two weeks. By week three I could build multi-step workflows with conditional logic.
• OpenAI API — How to send a piece of text to ChatGPT within a workflow and receive a structured response back. This is the AI agent layer.
• Airtable — Basic database for storing and reading structured information within workflows.
• Botpress — For building conversational interfaces (chatbots) that clients and their customers interact with.

Demo builds:

1. A property enquiry agent for a fictional estate agency: reads incoming Rightmove lead emails, extracts property address and buyer criteria, checks availability against a simple database, and sends a personalised follow-up
2. A dental practice intake agent: reads a new patient enquiry form, asks follow-up questions via automated email, classifies appointment type, and books into available slots

Both were functional systems built without a single line of code. I recorded 3-minute screen recordings showing them in action.

Month 2: First clients ($1,400)

I targeted one industry to start: small property agencies. My reasoning: they receive a lot of repetitive enquiries, they have limited admin staff, and the time they spend on manual lead handling is obvious and measurable.

My outreach message to 25 agencies:

"Most property agencies are losing leads because they respond too slowly. The industry average is 4 hours. I've built a system that responds within 60 seconds, any time of day, and books viewings automatically. I have a working demo I can show you in 10 minutes — want to see it?"

Responses: 9 (36%). Calls booked: 5. Clients: 2.

Client 1: Wanted the lead response and booking system. Paid £600 for the build, £150/month retainer.

Client 2: Wanted a leads database that their agent could query by typing natural language questions ("show me all two-bed leads in Manchester under £250k"). Paid £800 flat fee, declined retainer.

Month 2 revenue: £1,400.

Month 3: Referrals and a bigger project ($5,800)

Client 1 referred me to another property agency in their network. That agency wanted a more complex system: a full lead qualification pipeline plus an automated marketing follow-up sequence.

I quoted £3,200 for the build based on the scope and complexity. They accepted.

At the same time, an inbound enquiry came through my LinkedIn. A local accountancy firm had seen me post about the property work and wanted something similar for their client onboarding process.

I quoted £2,400 for that build.

Two projects running simultaneously. Both deliverable within three weeks total.

Month 3 revenue: £5,800 (£3,200 + £2,400 + £150 ongoing retainer from month 2).

Month 4: The system working properly ($10,200)

By month four I had a case study, two happy clients willing to give written referrals, and a clearer picture of which industries and problem types I was best positioned to serve.

I focused outreach on accountancy firms (following the inbound I had received), dental practices (using my original demo), and recruitment agencies — a type of business I had identified as having enormous amounts of repetitive email communication.

Month 4 revenue:

• 3 new builds: £2,200 + £1,800 + £1,600 = £5,600
• 4 ongoing retainers (months 2 and 3 clients plus 2 new): £600
• 1 workshop for a digital marketing agency (teaching their team how to use Make.com + AI APIs): £4,000

Total: £10,200.

The things that made the difference

Picking one industry first. Every conversation where I said "I've built this specifically for property agencies and here's an example" was 10x more effective than a generic "I build AI automations for businesses."

Charging for the workshop. I had not planned this. A digital marketing agency saw my LinkedIn posts and asked if I would teach their team the tools rather than build for them directly. The workshop format at £4,000 for a day became a new service line I had not anticipated.

Building retainers from day one. Every build proposal I sent included a monthly maintenance option. Not all clients took it. But the ones who did created predictable recurring income that freed me from always needing new projects.

Posting on LinkedIn consistently. I posted every week about what I was building and learning — not sales content, just genuinely interesting process notes. The accountancy inbound and the workshop enquiry both came from people who had seen these posts.

What the business looks like now

Month 7 monthly revenue: £13,500–£15,000.

Breakdown:

• New builds (2–3 per month): £6,000–£9,000
• Retainer clients (currently 8): £3,200/month
• Occasional workshops: £0–£4,000/month depending on bookings

Working hours: roughly 30–40 hours per week. Not passive income. But considerably better than the salary I left.

The one thing most people get wrong about this business

They learn the tools and then try to find clients.

The tools are the easy part. Make.com tutorials exist in abundance. The hard part is understanding a specific business's specific problem well enough to have a credible conversation about solving it.

If you spend three months learning every Make.com feature without talking to a single potential client, you will be technically knowledgeable and commercially invisible.

Build a demo for one industry. Book calls with five businesses in that industry before you feel ready. The first client will teach you more about what to build next than any tutorial ever will.

The $0 to $10K happened in four months. The timeline was set by how quickly I moved from learning to talking to real people with real problems.

That clock starts when you book the first call.