Article
€50,000 Quote vs. Two Hours with Claude Code
One hour. Fifty-five minutes.
That is how long it took to build what a Czech software firm had quoted at over €50,000. I built it with Claude Code.
This is vibe coding applied to a real business problem: I described what I needed in plain language and the AI wrote the code.
Not a prototype. Not a proof of concept. A working tool, the one the company needed. By evening, it was running on staging.
What Is Claude Code?
Claude Code is Anthropic's AI coding agent. It runs in your terminal, reads and writes files across a whole project, runs commands, handles your Git workflow, and works through a task step by step from a plain-language description of what you want. It is not a chat window that hands you a snippet to paste somewhere. It is an agent that does the work inside your actual codebase.
That is the distinction most people outside engineering have not registered. When I tell them I built a production tool in under two hours, they picture me typing questions into a box. That is not what happened. I described the tool, the agent proposed a plan, and it began writing and testing real code. I reviewed it, corrected it, and shipped it.
The short answer to what Claude Code is comes from what it let me do: replace a month and a half of quoted agency work with a single afternoon. The full story is further down this page.
How to Use Claude Code
The mechanics are simple. You install it, point it at a project, and tell it what you want in ordinary language. The skill is not in the typing. It is in knowing precisely what you want built, and being able to tell when the agent is confidently wrong.
What I have learned using it on real, production projects:
- Bring a clear idea, not a specification. My only input on the €50,000 tool was a clear picture of what it had to do. No architecture diagram. That was enough.
- Give it memory and context. Most of the value is in the setup. I have written about training an AI agent that learns between sessions, so it does not start cold every time.
- Build the tool instead of compromising with off-the-shelf software. I built my own analytics system with it, shaped around what I need to see, not around what a ready-made product allows.
Claude Code vs ChatGPT, and Claude vs ChatGPT
This is the comparison most people are reaching for, and it quietly mixes two different things. ChatGPT and Claude, the chat assistants, are conversational. You ask, they answer. Claude Code is an agent built on Claude that acts on your machine. Comparing Claude Code with ChatGPT is comparing a builder with an adviser.
For coding specifically, the difference I feel day to day is autonomy. A chat assistant, whichever brand, hands you text. Claude Code does the task. I have also tested running local models on my own hardware, and written about the limitations of local AI models when you try to use them for work like this. For now, my real work happens with the cloud agent.
THE €50,000 QUOTE FOR A SIMPLE PILOT
The task came from our CEO. A tool that our customers would interact with directly on the website. The scope was specific enough to define. Not trivial, but also not an engineering mystery.
I took over the conversation with one Czech software firm. Colleagues had started it but did not have the capacity to continue.
I asked for a rough idea of price and timeline. Not a binding quote. A range. Enough to plan.
Their reply was a single sentence I will not forget:
"We cannot give you an estimate, not even a rough one."
For a rough estimate, they explained, they would first need to conduct a paid analysis. Two and a half thousand euros. To examine the problem properly.
I did not take the analysis.
ONE AND A HALF PERSON-MONTHS: FOR WHAT
Later in the conversation, without the paid analysis, the number slipped out anyway.
Roughly one and a half person-months of work. Somewhere above €50,000 to launch a simple pilot.
A simple pilot.
For over fifty thousand euros.
Five years ago, I would have taken the quote to our two CEOs. We would have weighed it against an off-the-shelf product and its limitations.
There is a third option now.
BUILDING IT WITH CLAUDE CODE IN UNDER TWO HOURS
I walked into the IT meeting with a short presentation. I had mapped the options: build with the external firm we had already been talking to, or build it ourselves with an AI coding agent.
That same afternoon, I built it.
The development did not need iteration cycles. The agent proposed a plan, started implementing, and within the first few minutes something was already working. The rest was refinement. Adding features. Tightening behaviour. Then handling the edge cases I noticed as I tested.
The only thing I brought to that build was a clear idea of what the tool was supposed to do.
Not a specification. Not an architecture diagram. Just a clear idea.
That turned out to be enough.
In Is AI Making Us Dumber? I described interview candidates who came in with polished AI outputs they could not defend, outputs built on problems they had not understood. They had delegated the thinking before they had done it. This project worked because I had done the opposite. I knew what the tool needed to do. The agent did the rest.
WHAT SOFTWARE AGENCIES ARE REALLY CHARGING FOR
I want to be careful here.
Join the Library
Full access to my findings, personal stories, and what I learn from the people I meet.
Join the Library · €29.99 per year Fable 5 vs Opus: I Ran the Same Audit… I Ran Object Detection on My Laptop, and Saw… Bot or Human? What My… Dependent on AI: Are We… Which Work Will AI Not Replace? Not Yet… Dead internet theory is becoming reality. Will we return…Get the full article by email and feel free to reply if you want to discuss it further.
Summary
Common questions on this article's topic
How much does a software agency charge for a custom web tool?
Can you build a customer-facing web tool with Claude Code?
What can AI coding agents replace, and what can they not?
Are software agencies losing business to AI agents?
What is Claude Code and who is it for?
Can Claude Code run locally?
Can Claude Code read PDFs?
Can Claude Code run autonomously?
Does Claude Code use Git?
If you have any thoughts, questions, or feedback, feel free to drop me a message at mail@richardgolian.com.
LinkedIn
Related articles
When you hear that artificial intelligence still does not deliver good enough output, that it is unusable for certain tasks, it is worth realising that this may no longer be true today. And even if it were, the most intelligent minds on this planet are working on making AI more useful. Most work activities will be automated. To understand where this is heading, some questions have to be asked.
One of my family members described a firm that has its data in several different tools, and its employees still join it by hand, in spreadsheets. That is the ordinary state of firms. When somebody tells such people not to do it by hand and to use AI, they will not understand. Most firms do not even have the basic connectors between the tools they use. These firms need help with AI transformation.
Employees have the best AI models available at work, the connectors are hooked up, and colleagues already use them. And they still do not. AI adoption fails even when the tools are already there. It is not about time or AI skills. It is the inertia of old thinking.
More articles
The same task, two models. Fable 5 against Opus 4.8. On paper Fable is the better model, with a larger context and stronger specs. And still it lost. Opus handled the task with a single round of checking for 721,000 tokens, while Fable needed nine rounds and burnt through 2.78 million tokens. The difference was not in the model, but in how I set the task. And I know it, because I measured it.
A few weeks ago I installed a small local AI model on my laptop that watches a live camera feed. I turned the webcam on in the dark, and in near total darkness it recognised me and the objects in the room. That such things exist, I have known for a long time. What opened my eyes was the accessibility. I installed it in one prompt, free, and it runs entirely on my machine, sending data nowhere.
I once wrote about building my own privacy-friendly analytics tool. It had bot detection from the first version, yet it was not enough. Direct visits took a strangely high share of my traffic. When someone claims that 20% of their visits are bots and 80% are humans, I used to think the same. Today I would say the opposite ratio is closer to the truth. This is how I got there.
I have Heidegger and my notebook beside me. I am asking where all of this is heading, where artificial intelligence is taking us.
Seventy per cent. That is where the first AI output begins, even when you give it the full company context and the best examples from the past. We are talking about the kind of output that cannot be defined programmatically. It is more complex. Often it is creative work. On one repeated type of output I reached eighty per cent within a week. Every further percentage point is harder than the one before.
For a long time we treated the internet as the main road. The place where work and relationships happen. Yet most of what we see on it today is, or soon will be, AI-generated: text, images, profiles and comments. The internet is turning into an online game full of bots, where you cannot be sure that a human is on the other side of anything. So I ask: was the online world the main road, or only a temporary detour that part of us will return from, back offline?
A few days ago I interviewed a senior marketer. An experienced man, years of practice. I asked him about AI. He said he barely uses it. He had one bad experience with the output and decided he was too senior for it to add value when it is not perfect. I know the other side too: professionals who automate everything that can be automated.
Europe does not have the capacity to face a full-scale, mass drone war of the kind we see in Ukraine. Three dependencies weaken it: China supplies the physical material for defence systems, the United States supplies capabilities Europe does not have, and twenty-seven states cannot agree how fast, or who pays. Rearmament plans exist, but they are being carried out slowly.
AI produces the graphic, the newsletter and the product page faster than a person. What is left for the one who used to do it is the judgement, knowing whether the output is good. But most people have worse judgement than AI. And whoever cannot judge quality cannot delegate either. How do you tell whether yours is the judgement a company relies on, or the kind it can replace?
In April, in the first part of this series, I wrote about an AI prediction system I had started building on my own machine. At the time the software was a few hours old and the prediction record was empty. The record since then has shown one thing: the system does not yet understand the market it is being asked to forecast. It can pull macro context, book value, earnings. But it cannot put those together into something that helps it understand the price.
Four days in Catalonia. No computer, no AI, almost no social media. I bought this notebook so that I could write down what I would think about, and what I would come across and learn on the trip.