正在加载视频...
视频加载失败
It is BS that vibe coding will replace software companies... "People underestimate how hard it is to maintain software over time. It's very easy to create the first increment but to change it, adapt it over time, it takes a lot of effort to do that. Businesses have a... show more
33 条评论
This is NOT the right conversation. The right conversation to be had is, "A flood of competitors will emerge that can offer better software at a vastly lower price because of their lower cost base. Something that incumbents can't compete with because of their legacy and now bloated cost structure. How do you plan to compete with that?" Internal vibe-coding argument gives all incumbents an easy out.
I run a dev agency. vibe coding is incredible for prototyping and MVPs but the moment you need auth, payments, scaling, security, and a team that can debug at 3am it falls apart. the first 80% is easy. the last 20% is where companies actually pay.
This is the nuance most people miss. V0 is easy. V10 is where the real work lives. The maintenance, the edge cases, the integrations. That's why "I could build that in a weekend" always ages poorly.
I think both things can be true. Vibe coding lowers the cost of creation. Maintenance, reliability and long-term evolution are where real companies earn their keep. Software isn’t just built, it’s operated and operations only scale when the economics work over time.
Maintenance is hard when humans wrote it. When AI writes it, you just throw it away and regenerate.
The coding was never the hard part. The hard part is knowing what to build in the first place. Vibe coding might collapse build time from weeks to hours. But the discovery loop — talking to customers, iterating on feedback, handling edge cases nobody warned you about — that still takes years. Software companies that survive aren't just code. They're accumulated domain knowledge.
The magic word is "adapt".
SaaS multiples of 20, 50 or 100 PE are no longer valid. It’s the speed at which the AI is improving, and it’s almost a certainty that in five years, the AI can do everything that you say it can’t today.
Vibe coding doesn’t replace companies. It lowers the minimum viable team size. The next $50M ARR company won’t have 200 engineers.
Also, what a reductive view that product = company. Just as there is broad recognition feature ≠ product, its worth clarifying product ≠ company. True cost of development is somewhat converging to some asymptote close to zero over time, but CACs continue to rise.
Software companies that win won't be displaced by vibe coding they'll be the ones who recognize it as their customers' new prototyping layer and build deeper moats around maintenance, scalability, compliance, and integration that actually justify the enterprise premium.
Why wouldn’t agents evolve to adapt and maintain software over time?
Reminds me of the no-code hype. Great for prototyping, terrible at scale. The complexity ceiling always hits eventually and then you need real engineers.
Won’t the ai eventually be able to maintain it tho? Especially Monday. It’s a simple project management software.
Its BS to conflate two different points here. Yes, software companies wont be replaced but there will be much lower-cost alternatives, which is the point.
100%. Will it get better and better, sure. Will we still require people that understand what the hell is happening, oh yeah.
disrupt? definitely. destroy? probably not. SaaS as a whole will have to adapt. simply being a CRM isn’t good enough anymore.
Yes 100% when we are talking about how we operate today, but thats because we are still better then AI at coding, that will not be the case in a year from now. Deploying, architecting, maintaining and debugging are the hard parts, but when agents are built to maintain, to deploy, to upgrade themselves by reading new documentation, then who ever runs the department simply just asks it for what they need: the sales person knows exactly what they need to improve sales, the product person, the GTM person. We only have 0.3% of people even paying for a AI sub right now, so of course this isn't happening this week, but it just won't be hard to solve these problems pretty soon.
I feel like his idea of vibe coding outdated by at least 12 months. Which in the AI world is like 6 cycles behind
Uhhh, if it's very easy to build the first version, why not build it from scratch every time as well as build a series of micro apps?
you can literally have AI maintain the code. software will be worthless and companies like this are totally cooked.
Maintenance is real, that's the right argument. But it doesn't follow that software companies are safe, it means the surviving ones need fewer people to do that maintenance work.
My thought exactly as someone who’s tried and failed to build anything useful with scalability. Noted I’m also probably lacking some depth and understanding. It’s like people building unbelievable excel models that break when people leave and take the IP
Once a month you have these legacy CEOs on your channel whining around about how the market is treating them but really they have not shown any valuable agents/AI featuers on their platform that would have incremental willingness-to-pay
Makes good and bad arguments the truth is no one really knows right now. If the curves start to flatten he is more right than wrong if the continue up and to the right is just wrong.
Maintaining software over time is hard if your software is serving hundreds or thousands of customers with various needs. It is not hard if it's only serving your own company.
At one point, people were saying the same about writing code.
Agree 100%.
exactly. vibe coding is a fast ramp, not a replacement. the first 20% of a product is always the easy part. it's the next 3 years of bug fixes, edge cases, and legacy debt that separate software companies from weekend demos.
Enterprises won’t rely on vibe coding. But founders already do. And when the cost of building drops, more experiments happen. Some of them become real companies.
So will soon handle that
Exactly. The allure of vibe coding is real, rapid prototyping, low-code wins, but long-term software is a different beast. It’s not just about building features; it’s about scalability, maintainability, security, integrations, and evolving with the business. A no-code app might solve a short-term need, but without proper engineering, technical debt piles up fast. Software companies don’t just write code, they build systems that survive years of change. Vibe coding complements them; it doesn’t replace them. It’s the difference between “can you make it work today?” and “will this still work when your company grows 10x?”
at $1.3B ARR and a 3x multiple tells you everything. the market is pricing in that half their product gets replaced by custom tooling within 5 years. vibe coding won't kill SaaS. but it'll kill the margin that justified SaaS pricing.
