From Academy skill to your first paid AI gig

Most people who work through an AI course series get stuck exactly here. The skills are there, the confidence wobbles, and the question "how do I turn this into money now" feels bigger than everything you've learned so far. But it isn't. The path from "I can do this" to "someone pays me for it" isn't a talent thing, it's a sequence of small concrete steps. Those are exactly what we go through now.

A warning up front. This playbook doesn't sell you dreams. No "6 months to 10k a month". The goal is more modest and more honest, one single paid gig. The first is the hardest, and once it's through, something tips over in your head. After that it's repetition.

Step 1, An honest stocktaking

Sit down and write down three things you actually built in the Academy. Not "I learned something about memory", but "I set up an MCP server that does X". Of these three, cross out everything where you still had the tutorial open. What's left, the thing you could build again without instructions, is your starting capital.

Usually it's exactly one thing. That's enough. Nobody hires an "AI generalist", but people pay for someone who solves a concrete problem for them. Your one thing is more important than your whole breadth of topics.

Step 2, A benefit sentence instead of a title

Stop calling yourself "AI operator" or "AI consultant". That tells nobody what you do. Instead, write a sentence along this pattern: I help [whom] to solve [which problem] by [what you concretely do].

An example. Not "I'm an AI expert", but "I help small trade businesses answer their customer inquiries automatically by setting them up a chat assistant with their own data". The second sentence immediately conjures a person in your head who needs exactly that. The first gets nobody to call.

Step 3, The proof project in one weekend

Talking about skills convinces no one. A visible, working thing does. Take a weekend and build a real small thing for a real or made-up small business. A GEO audit for the website of the baker around the corner. A little MCP server that answers questions about a fictional restaurant. An n8n flow that sorts contact-form inquiries.

Keep it small. A thing that's explainable in two sentences and demonstrated in 60 seconds. If you need ideas, the playbooks your first MCP server in 90 minutes and GEO for small sites are ready-made templates for exactly such proof projects. Non-coders start with my first n8n flow for non-coders.

Step 4, Make it visible

A project no one can see doesn't exist for clients. Three things make it visible. A public GitHub repo with an understandable README, a two-minute screen recording where you use the thing once, and a before-and-after in one sentence ("before 20 minutes per inquiry, after 2").

The repo is more important than it feels. It proves you really built and didn't just talk. How you set up a clean public repo in five minutes is in Git for AI quickstart. For the non-coders among you, a tidy Claude Project with your best work does the same job, see Claude Projects for non-coders.

Step 5, The price reality check

On the first gig you don't do an hourly rate. An hourly rate punishes you for your slowness as a beginner and forces you into a discussion you can't win yet. Take a fixed price for a clearly outlined result.

Realistic for the first real gig, somewhere between 300 and 800 euros for a small self-contained project. Sounds low, but that's the point. With the first gig you're not buying revenue, you're buying proof and a reference. The second gig is twice as expensive, the fifth four times. Don't go below 200 euros, otherwise people don't take you seriously and neither do you.

If the low price feels uncomfortable, call it a pilot price. "That's my introductory price for the first three projects, after that it costs more." That's honest, gives the client the feeling of being early, and you don't have to justify yourself.

Step 6, The first five contacts

The first gig almost never comes from a cold mass email. It comes from warmth. Write to five people you know who have a small business or know someone who does. Not "do you have work for me", but "I'm building this kind of thing right now, do you know anyone it would help". People like to refer, they just need to know what for.

On top of that, two communities where your target group really hangs out. A local entrepreneurs' network, a Facebook group for freelancers in your region, a fitting subreddit. There you don't post "I'm looking for clients", you answer questions people ask for two weeks. When someone asks "how do I answer inquiries faster", you write a real, usable answer without selling. The gig comes from the trust that builds, not from the pitch.

Step 7, The first offer in five lines

When someone shows interest, don't overcomplicate the offer. Five lines are enough. The problem you heard in one sentence. What you concretely deliver. The fixed price. How long you need. A clear next step ("if that fits, send me X and I'll start on Monday").

Here's what that looks like for real: "You spend hours a week answering the same customer questions. I'll set you up a chat assistant that answers the 30 most common questions with your own texts. Fixed price 500 euros, done in a week. If that fits, send me your existing answer templates and I'll start Monday." The most common beginner mistake is a three-page PDF with methodology and phases. Nobody reads that and it comes across as insecure. A short, clear offer the client understands in 30 seconds beats any embellished concept.

Step 8, Deliver without burning yourself out

The gig is signed, now the part beginners underestimate. Put the scope in writing, in the same terse language as the offer. One revision round is included, further ones cost extra. There's a moment of sign-off, from which the project is done. Without these three sentences you work three weeks on a 400-euro gig and hate your new business.

As soon as you work with real customer data, the legal side gets real. Who processes whose data where, what the AI is allowed to see, what belongs in a data processing agreement. No drama, but don't ignore it either. The basics for the DACH region are in EU AI Act and GDPR for the DACH region.

Step 9, Collect the proof

After every gig you pick up two things. First a short quote from the client, three sentences are enough, what was the problem and what's better now. Second, you turn the project into a small case study for yourself, along this three-liner: starting situation ("baker answered every inquiry by hand, 20 minutes each"), what you did ("set up a chat assistant with the 30 most common questions"), result in one number ("response time from 20 minutes to under 1 minute").

This collection is your real capital. After three gigs you have three references and three case studies, and from there you no longer have to convince anyone that you can do it. The proof is on the table. Ask for the quote right at sign-off, while the joy over the finished result is fresh, not two weeks later by email.

Step 10, Once becomes twice

The most expensive mistake after the first gig is to start every next project from scratch. Look at what you just built and ask yourself, can I sell exactly this to someone else again. Usually yes. From "chat assistant for this one tradesperson" becomes "chat assistant setup for trade businesses", an offer you sell ten times instead of reinventing ten times.

From here it pays to make your profile permanently visible too, on a job board, in a freelance directory or via the Academy's operator funnel. That's one of several channels and never replaces the warm contacts from step 6, but once the proof stands, it works for you while you sleep.

What comes next

Take next weekend for step 3, the proof project, and send out the five messages from step 6 on Sunday evening. Nothing more. The rest follows from the answers that come back.

If while building you notice you're still missing craft, go back into the levels. Level 3 shows you automation, Level 4 memory and MCP, Level 5 the agent patterns. And if you want to set up the legal side cleanly for the first client, EU AI Act and GDPR for the DACH region is required reading before the first contract is signed.