pull down to refresh

Over the last few months, I’ve been experimenting with creating AI-assisted apps based on my own ideas — concepts that I’ve had for years, even decades. You know, the “one day I’ll build this” concepts that used to require a development team, funding and quitting your job just to attempt them. Not anymore.
Like many of you, I’ve spent years becoming an expert in my field, but I’m not a coder. That used to be a barrier. Whenever I came up with a product idea, I’d abandon it because I lacked the technical skills or the budget to outsource it. That limitation is gone. The doors are open.
We're witnessing something extraordinary: what used to take five to seven years of in-depth technical learning can now be achieved in a month of focused effort — without leaving your day job. This isn't because you've become a full-stack developer overnight, but because AI fills in the gaps just enough to help you create a prototype or a minimum viable product (MVP), and sometimes even more.
Everyone’s talking about how the future will favour product thinkers over engineers. In fact, I'd say the future belongs to generalists — people who have lived at the intersection of ideas, fields and disciplines, even if they never learnt to code.
Specialists dig deep. Generalists see patterns across domains. Product ideas — real ones — are born at those intersections.
Until recently, the gap between idea and execution was filled with complexity, team-building, funding decks and gatekeepers. But AI is quietly closing that gap. Now, you don't need permission. You don't need a co-founder. You don't need a warm introduction. All you need is curiosity, focus, and enough drive to get things done.
Yes, it’s still hard. No, it's not magic. You’ll still need discipline and time. But the leverage has changed. Drastically.
Nowadays, someone with a clear idea and basic AI knowledge can develop a product that would have required a team three years ago. The solo founder archetype is making a comeback — not as a hacker in a hoodie, but as a domain expert finally armed with the necessary tools.
The paradigm has shifted: Idea → Action → Product → Proof → Momentum. No pitch decks. No gatekeeping. Just doing.
If you've been sitting on ideas for years, now's the time to bring them to life. You don't need a development team anymore. You are the team!
I recently had this sense of euphoria when I started to use replit to build my dream SN iOS app. Went from taking course to course to feeling like whoa AI did all this for me. Laid out the roadmap and even produced a UI.
But now I feel like I have imposter syndrome. I know this app will have lots of issues and will be no where near ready for the masses.
reply
All you have to hope for is that all the doomers are right and there will be no more employment for devs, so that the current price tag of about 6-8BTC a year for a good, self motivating, Sr dev goes down to more affordable price levels.
Unlikely though, because now literally everyone can be in your position. Demand will forever be high because everyone can have openai generate code. There is no moat in having the code, but being able/motivated to read slop and fix the issues in it does, and is probably going to be an awesome career plan.
reply
Sounds like a nightmare
reply
56 sats \ 1 reply \ @optimism 6 Aug
"Living the dream" doesn't discriminate.
reply
Hahaha
reply
Totally get that - I’ve had the same emotional rollercoaster. The bugs, the mess - that’s where the learning is. If it just worked, it wouldn't feel earned. Keep going - the fact that you’ve shipped anything already puts you ahead of 99% of "someday builders")
reply
Breaking, or changing?
You'll still need a dev team. Don't find that out when you have a production app and you lose everything.
I'd say that it allows you to prototype ideas, but not real products and not market fit, because the slop is real.
So your process will look different, but so will everyone else's. Competition be fierce. Expect scraps
reply
I don't think AI replaces expertise - it just compresses the cost and time to get to first proof. Most AI-built MVPs are messy. The key shift is: you can now test your vision faster than ever before.
reply
Most AI-built MVPs are messy.
It is if you don't know what you're doing, so you need a pro for your MVP, probably at least 2. However, we'd normally want a couple more. So you still need capital for that stage, and your demo will be a throwaway, but I agree that there's much more flexibility and potentially some cost savings.
reply
100 sats \ 1 reply \ @lunin OP 7 Aug
It is if you don't know what you're doing, so you need a pro for your MVP, probably at least 2. However, we'd normally want a couple more. So you still need capital for that stage, and your demo will be a throwaway, but I agree that there's much more flexibility and potentially some cost savings.
Well, at least some will refuse to develop their own junk, others will think about hiring a specialist, others will start learning themselves and achieving results — in my opinion, all this is already not bad!
reply
The danger is in the advertisement when someone sloppy-vibes a "product" together. Unless it's open source (it takes extraordinary unselfish attitude and at least a decade of runway to open-source a truly novel idea) and we can judge for ourselves if it is slop, the risk is real. Especially in the bitcoin space where we are currently seeing slop product after slop product fail, and we've only just started.
Therefore, I think that the proliferation of LLMs as vibe coding tools is much more powerful as a personal toolset development strategy than a product development strategy, but time will have to tell if I'm right about this.
reply
105 sats \ 3 replies \ @OT 6 Aug
How long do you think the slop will be around for? Like, will the next gen produce perfectly executed code without getting lost or being overly complex?
reply
It'll be around forever, because slop isn't a function of quality, it's a function of scarcity/abundance.
In the hands of a somewhat creative but skilled developer, you can create really good code with "dated" models from perhaps even 2 generations back (i.e. from December last year)
In the hands of a no-coder, you have 99.99% chance to get slop now, to get slop next year and in 200 years from now, you will likely get slop. Maybe 1:10000 will get lucky? That's the gain here, because bottom line a bad idea is a bad idea is a bad idea.
The bar for non-slop rises symmetrically to the common output of available tools1

Footnotes

  1. Can we name this the law of optimism? please? please? please? <please sir emoji>
reply
105 sats \ 1 reply \ @OT 6 Aug
So vibe coding isn't really about getting normal people to build stuff. You still need to know a few things about coding or architecture to have a chance at making a successful product?
reply
Yes. Everything else is boilerplating, and that's an anti-pattern on its own.
reply
110 sats \ 2 replies \ @ACYK 6 Aug
It feels more to me like it breaks the glass ceiling for those who have app ideas they wish existed so they could use, but not necessarily for those who want to make apps for the masses. If you want something for you and your friends to have access to, current tools are great for that. Once you want more than that, you’re still going to have an app full of issues that is challenging to scale. Not to mention a host of other issues. Depending on your jurisdiction, you may want some legal insulation to protect yourself personally from any proprietary code that may have gotten incorporated into your vibe coded app; something that would be less of an issue for a free app you share but more of an issue if you’re charging for it and it becomes popular.
reply
legal - agreed. If you're charging or scaling, legal hygiene becomes important fast. Great reminder btw!
reply
This works the same way if you are just throwing something together without AI.
I always expect to throw away my first attempt after I get it "working".
Usually the third attempt is production ready.
reply
76 sats \ 1 reply \ @claos545 6 Aug
Absolutely feel this. I've had "someday" ideas sitting in notebooks for years the kind you revisit every few months thinking what if. And for the first time, it doesn't feel like a daydream. It's surreal how accessible building has become.
AI hasn't made it effortless, but it's made it possible. Especially for those of us who know exactly why something should exist, even if we couldn't code it from scratch. That shift in leverage you mention is real. Execution feels closer than ever and a lot less gatekept.
Love how you framed generalists too. It's not about doing everything perfectly, it's about seeing the opportunity and knowing enough to get moving. Thanks for putting this into words. Inspiring stuff.
reply
Exactly how I feel: possibility -> perfection. It's surreal to feel ideas moving from notebook -> screen -> prototype in weeks.
reply
130 sats \ 1 reply \ @alexbit 6 Aug
Too many dashes.
reply
Ha, fair. And ironically - I actually wrote a post about that exact thing (yes, even the em-dash vs hyphen debate): https://www.linkedin.com/feed/update/urn:li:activity:7308500628178153472/
Also, for the record: on macOS, em-dash (—) = Option + Shift + –
reply
Definitely AI has already changed the way we think and work, aswell replace the traditional way of working. It believe it have many advantages and disadvantages depending on how one make proper use of it.
reply
It's like when the Internet appeared. But there was a lot of good then, we adapted, became faster and probably smarter. With AI it's similar, so far a lot is unclear what and how, but for me personally AI (after a year of active experiments and trials) helps a lot — a kind of "second brain" and ... often ... a critical interlocutor!
reply