The GitHub looks great. The bank account has a different opinion.
There was a guy in Chiang Mai. I'll call him Thomas because that was his name and he's probably back in Munich now so it doesn't matter.
Thomas was a genuinely good engineer. The kind who could look at a problem and see three solutions before most people had finished describing the problem. He had references. He had a GitHub that made me feel bad about my own career choices. He had six months of runway, a room in Nimman, and a plan that was entirely technical and not at all a plan.
I met him at the coworking space on a Thursday. He'd been in Chiang Mai for eleven weeks. He was building a portfolio project. He said "portfolio project" the way people say it when the portfolio project is the third portfolio project and the first client has still not appeared.
I asked him how the client search was going.
He said he was almost ready to start looking.
He went home eight weeks later. Not because he ran out of money, though he was getting close. Because he couldn't reconcile the version of himself he thought he was coming here to be, with the version of himself the life actually required.
He thought he was a developer.
He was a salesman who needed to learn to code on the side.
That's the thing nobody puts in the plan.
Every year, a new wave of people arrive in Southeast Asia with a legitimate technical skill and the assumption that the skill is the product. It was the product when someone else was selling it for you. The employer was the sales function. The job listing, the interview process, the HR department, the team structure, the meetings where someone explained what to build next. All of it was the sales function, and you just showed up to write the code.
When you go independent, you don't get rid of the sales function. You absorb it.
This is not optional. It is not a phase you get through. It is the job. Permanently. Finding the client is the job. The code is evidence that you can do the job you found.
Most people hear this and nod because they think they already know it. Most people do not already know it. The gap between knowing it intellectually and knowing it at 9am when you should be sending a cold email but you're refactoring something instead, that gap is where independent careers go to die quietly and without announcement.
The refactoring is comfortable. Writing code is the thing you know you're good at. Sending a message to someone who has never heard of you, explaining why they should pay you money, that is the thing you have no evidence about yet. The natural move is to stay in the comfortable thing longer than necessary and call it preparation.
It is not preparation. It is avoidance dressed in productive clothing.
I have done it. The language was different. The work was different. But the mechanism is identical across every independent skill. You work on the thing instead of selling the thing. The work improves. The income does not.
Here is where 2025 enters the conversation, because it would be dishonest not to.
The engineering market has changed in ways that are real and not temporary. AI writes code now. Not perfectly, not autonomously, not without a human who understands what the code is supposed to do. But well enough that the entry-level work, the small component, the boilerplate, the first draft of the function, that work is cheaper to produce than it was three years ago. The clients know this. Some of them are using it themselves without fully understanding it. Some of them are using it very well.
What this means practically is that the technical skill alone is worth less than it was, and the ability to find, convince, and retain a client is worth more. The people who can do both are more valuable than they have ever been. The people who can only do the technical half are in a more competitive position than they have ever been.
This is not an argument against learning the technical side. It is an argument for understanding which half of the job is now the distinguishing variable.
The distinguishing variable is the client relationship. It was always the distinguishing variable. AI has just made this more obvious, faster, and less forgiving of the people who were ignoring it.
Thomas was ignoring it. Not because he was lazy. Because nobody had told him clearly enough that the job he was doing in Munich, the technical part, that was not the job he was applying for when he bought the ticket to Thailand. He was applying for a different job that used the same skill set. The new job had a different primary function. He never quite accepted that.
I watched him spend nine weeks making himself better at the thing that was not the bottleneck.
The bottleneck was always the conversation he wasn't having with anyone.
Finding clients is a skill. It is not a personality trait. You do not have to be extroverted, American, or relentlessly optimistic. You have to be clear about what you do, consistent about telling people, and willing to have the conversation before you feel fully ready.
The feeling of being fully ready does not arrive before the conversation. It arrives, sometimes, about three conversations in. You do not get to skip the first two.
The channels matter less than the consistency. Email works. LinkedIn works despite being insufferable. Referrals work better than both. The nomad community in a city like Da Nang or Chiang Mai or Tbilisi is smaller than it looks, and the person who helped someone solve a problem last month gets mentioned in the next conversation. Not always. Enough.
What does not work is the portfolio project that nobody knows exists. The GitHub repository with no human attached to it. The website written for an imaginary client who found you through a search that never happened.
The work has to leave the laptop. It has to touch another human being who has a budget and a problem. Everything else is a rehearsal.
I have a rule. It is not sophisticated. Every week, something I have made or something I have said has to reach someone new. Not a thousand people. One person. Someone who didn't know I existed last Monday.
Some weeks that's a post. Some weeks it's a reply to an email thread I was copied on. Some weeks it's sitting at a different table at the coffee shop and saying something useful to someone who asked a question out loud.
The scale is not the point. The habit is the point. The habit of moving the work outward instead of inward. The habit of treating the relationship as the product and the code as the thing you use to justify the relationship.
Thomas moved the work inward for eleven weeks. He got better at something that didn't need to get better. He came here to freelance. He freelanced at having a GitHub.
None of this is cruel. Thomas is probably fine. Munich has a reasonable economy and his skills were genuinely strong. He'll find a job. He'll maybe try this again in a few years with different information.
But the information was available. It just required accepting something uncomfortable: that leaving employment to go independent is not an escape from the commercial part of work. It is a promotion into the commercial part of work. Full time. No HR buffer. No sales team. No senior engineer whose reputation is doing half the selling for you.
You are the product. You are the sales function. The code is the receipt the client keeps for their records.
The sooner that becomes obvious, the longer you get to stay in the country where the coffee costs less than you paid for the taxi to the airport.
The Thomas timeline is eleven weeks, six months of runway, and a flight home before anything was actually tried.
You do not have that much time to spend on the comfortable thing.
Andrew - No Refunds •••