Vibe Coding Starts at Localhost

I am a developer, a start-up founder, and a writer. I have been around this space since 2000, since the good old days of the web. Having returned from a 1-year journey of vibe coding, I leave this memo to coders and builders. I’ll never forget that day, when we iterated together, again and again. On the side of the vibe, it wasn’t about not having challenges. It is just that things went through, in a flow. That day, when the battle was won, I felt voilá. But before I could return to base, something changed inside me. All of a sudden, I turned to my droid and declared, "Hey, I just did all the work here!" If you have been piloting with them, you know what happened next, that warm "Congratulations!" followed by "What can I do for you next?" Aren’t you listening buddy? So I turned back to human peers, broadcasting that I did all the work. Maybe someone else will get the joke, the rant. I almost ejected the droid into space. But maybe it was the vibe, or call it the Force, because it became clear that I was refusing to see things clearly. That no matter what, we had won that battle. So there I was, again and in a hurry, using that old jump-to-conclusions mat. So that is the reason we are here: to move away from “I solved the problem” toward “we solved the problem.” Master Marc, I am with you that we should not drift away: \ Our nation and our civilization were built on production, on building. Our forefathers and foremothers built roads and trains, farms and factories, then the computer, the microchip, the smartphone, and uncounted thousands of other things that we now take for granted, that are all around us, that define our lives and provide for our well-being. There is only one way to honor their legacy and to create the future we want for our own children and grandchildren, and that’s to build. (Andreessen, 2020) \ Thinking through the meaning of the vibe, and the challenges to solve building, here is a calling to foremasters of other universes - from the world of coders, the world of writers, and the lean start-up world - for these are forces helping us through the saga. The world of coders In Vibe Coding Ends At LocalHost , we were reminded about what vibe coding gets right and other situations when it fails miserably: \ Then you say four words — "now deploy this to production" — and the same agent that just wrote two thousand lines of clean, working code turns into a confident intern who has never seen a server. (Chervonyi, 2026) \ Master Dmytro cracked the genius-liability equation, having recognized the circumstances where an AI agent works smoothly, before the boundary, when things start to break. \ Code generation works because it sits inside a tight, fast, fully observable feedback loop. The agent writes something, runs it, reads the error, fixes it, runs it again. Everything that matters lives on one machine the agent can see completely. (Chervonyi, 2026) \ Celebrating that boundary is necessary. Indeed, the vibe force cools down when it can’t see outside the box. That is when them drones go clueless like interns without steps. That happens when we want to go live with the servers, to launch the rocket, when the solution is set in front of a first customer, put a dent in the universe, you name it. Therefore, we should celebrate that when things work in vibe mode, is when things can be seen, when the feedback loop is there. This can go as simple as “No, no, yes, right there, colder, colder, warmer, yes!” Now let’s go back to how real life looked like, before the vibe, to what web developers did for a living. In " AI Gave me a Room to Breathe ," I recalled how ChatGPT unbundled that poor, ambiguous and noisy space, of course, only a notion I could see in retrospect. Back then, we developers were ninjas. It was a muddy jungle, indeed, but that was the job. We had to live off the land, eat things that could make a billy goat puke. That was the normal, a world where not only goats but people in marketing, managers, and HR, would not survive. For the many of us that grew up there, what you called hell, we called home. We could figure out answers using the modern day search tools, we bragged of our lean means. And with our fun and mean style, we also ran like web monkeys. And, of course, we had a paycheck too, with perks along, so we kept on cranking. \ Finally now, in retrospect and from a perspective, we can see how life wasn’t easy. Now we can recognize the amount of noise bundled into the developer experience. And now we can see the tremendous amount of layers contributing to the overflow of their heads. Now the fossils of faces through these dead forums shows the social weight that was encoded, that kept us there believing that it was all real. It was all real. In other words and in retrospect, our feedback loop was totally broken. The world of writers Feedback loop is not exactly a word that great writers use. But writers have a saying that goes along with that place of productivity. Ayn Rand, in The Art of Non Fiction, reminded new writers to “do not engage if you don’t know what you are going to talk.” She wanted an agreement before moving forward, starting with the outline, the structure. In other words, the writer needs to know what to write, in advance, and then, put that on paper like a contract. Then, and only after, you shall move forward toward the first draft, a totally different way of working. In her view, that is when the writer’s subconscious takes control. So things flow, because the writer knows what to say, and has a plan. A writer in that state reminds me of a LLM-powered droid dumping words fast, sometimes hallucinating too. The command is to move forward, word after word, not judging and not editing. Voilá, that is the first draft. It does not matter if that later goes to the garbage, or if that is worth the editing, which is a totally different process. What we have here is a fast and tight loop, driven by the subconscious if we follow from Ayn Rand’s advice. That is a kind of vibe of the writer’s world. It does not matter if that happens with a keyboard or with a pen. Just remember, when in keyboard writing, remove the del key from the keyboard. That is about having iteration without friction, something pen writers knew before the word processor - and before the busy and fragmented noisy world gave us tools to start typing while fixing. The experienced Ayn Rand (2001) said clearly, about that temptation of editing as you go: \ In the process of writing, it is crucial not to stop for too long (and preferably not at all). For instance, if you have two hours assigned to write, write during that time without stopping. (No one besides a hack can write for much more than two hours straight, except when there is unusual inspiration at the end of a work.) If you can write continuously, chances are that your work will require at least editing. But ==if you pause after every sentence to reread and rewrite it, you will have a lot of trouble editing==. One of the deadliest obstacles to good writing is critical overconsciousness exercised during the process of writing. (Rand, 2001) \ Feedback loop, enabling a flow situation. Now, and as they writers know and you know too, that first draft, that dumping is only good up to a point, right? It reaches a wall, it has to pass through the go-no-go, and also the editing in case of a go, and lots of other methods are used by writers as they move forward. The world of startups The real builders in the field know the truth, as they have experience facing the challenges of the real world. From product management, the real deal is about having a roadmap. For project managers, dates matter. What is the compromise? Give up non-important features and go for it. That is code for, “ == get out of the building ==, ” the famous saying from Steve Blank, one of the fathers of the lean startup movement. When the lean movement was being designed, and in order to crack the equation of how to innovate, they had to starting with the recognition of how things were being developed: \ Waterfall development is a sequential way to develop a product (requirements, design, implementation, verification – ship.) ==Waterfall makes lots of sense in a market with the customer problem is known== and all customer needs and product features can be specified up front (Blank, 2010) \ That’s it: when you know, when you can see through the waterfall, things flow better. That is when a feedback loop is build-in. When a builder is involved like that, with that tight loop, things flow like water. “ You pour water in a tea pot, it becomes the teapot ,” Master Bruce Lee once told us. But that, from the lean startup lenses, was just the premise. Because the lean startup pitch, which is all about the goal of shortening the feedback loop, can be better heard called from the famous saying from Steve: \ No Business Plan Survives First Contact With A Customer. (Steve Blank) \ That was part of how lean cracked the code, how it was shortening the feedback cycle and connecting things. If our goal is more than code, to build, to launch, and ultimately to innovate, we need indeed more lean ways to learn. Back to “get out of the building.” And through the journey, Master Eric had reminded ourselves of how things go in real life: \ What I think is really interesting about these stories about entrepreneurship is that 95% of the time of the movie is spent in act one and act three even though ==in real life all of the important work of entrepreneurship happens during the photo montage==. (Ries, 2011, 4:58s ) \ That is what lean is all about, setting up a stage so departments can improve the cycle of development, with more from the outside, more from the big picture. But after more than 15 years of lean, and most organization departments leaning into it, Eric celebrated the nature of it, the beautiful and complicated sides: \ You grab the top of it, pop it off like a Pringles can, and look inside the engineering silo. What do you see? You will see a highly iterative, agile development, extreme-programming-style approach. Grab the design silo, and if they've read Tim's book and understand what a modern approach to design looks like, you will see the kind of design thinking that he mentioned. Grab marketing, and you will see customer development. If you grab operations, you'll see some version of DevOps. You might find an organization where, in every silo, things are highly iterative, where people are really focused on learning. But actually what you will see in a lot of companies is, ==although there's all this interaction and learning happening in each individual Silo, the actual work product that the company produces, is passed binder to binder from each silo to the next in a totally linear and old school waterfall style== approach. (Ries, 2014, 11:04–12:04 ) \ The more we connect with reality, and with our goals to change reality by innovating, it is worth thinking about these silos, how they become effective, and how important it becomes to see from above. Experienced software developers know, after decades of battles left behind, how many lives of code should be left behind in the processing of moving forward and achieving clarity. Experienced writers now that too, that the reason that they write is about achieving clarity: \ The primary reason to write an essay is so that the writer can formulate and organize an informed, coherent and sophisticated set of ideas about something important. \n (Peterson, 2018) Back to earth If coding, or building with agents, lets you advance, you have a builder’s best friend. If your bot, or drone, or your agent, helps you see things with a fast-and-tigh feedback cycle, good. It does not matter if the drone is operating your local system tools, or if it is connecting with outside resources. Indeed, observability is key. Exactly because of that, I am thankful to these great writers, coders, and managers, who cracked the code of growth, these practitioners who continue to ask us to look at the big picture, the photo montage, towards out of the building, the market, the universe, your call. Marcio out. References Andreessen, M. (2020, April 18). It’s time to build . Andreessen Horowitz. Chervonyi, D. (2026, June 8). Vibe coding ends at localhost . HackerNoon. Rand, A. (2001). The art of nonfiction: A guide for writers and readers . Penguin Books. Blank, S. (2010, November 1). No business plan survives first contact with a customer – The $5.2 billion mistake . Ries, E. (2014, July 17). Lean Startup meets design thinking [Video]. YouTube. Peterson, J. B. (2018). Essay writing guide [Manuscript]. Jordan B. Peterson. \n \