Domain expert (independent retail) turned saas dev here 🙌🏻 this article nailed my experience. There was such a blind spot in creating inventory solutions for retailers that source from many vendors and are brick and mortar yet every time I met with devs the math didn’t math as you say. Everything was about DTC and Ecomm. I am so proud to say we have very happy beta users and are taking it to the broader market now!! It really is a fantastic time to have a lived problem you could never find a solution for.
thanks for the post! cheaper software creation means more of it, not less, so this isn't devs-get-replaced, it's the market expanding to people who were always priced out.
I think we're conflating two very different things:
Building software.
Building a successful software company.
The first has become dramatically easier. The second has not.
If you're building for yourself, fantastic. AI and no-code tools are unlocking possibilities that didn't exist a few years ago. More people should absolutely build tools to solve their own problems.
But if you're building for customers, coding was never the hardest part. Never.
Much of the business building skills that have been and continue to be needed have not changed, but the expensive, time-consuming, and human-capital intensive nature of designing software has changed. I think this leads to a commoditization of software skill which allows more people who were scared off by the prohibitive cost of making software to try their hand at creating a SaaS company. I don't think that the inherit skill of building said successful SaaS company is any more prevalent in society, but with more people trying it I think the pattern of a non-developer SaaS founder will indeed increase! I think more of the people who CAN build for customers can now come out of the wood works and do so.
I don’t disagree that more people will build SaaS companies. My point is that AI is solving the production problem, not the business problem.
Software is becoming abundant. Customers are not.
As building gets easier, distribution, trust, differentiation, and customer adoption become the real constraints. I don’t even want to go into difficulties of integrating niche SaaS into enterprises. Nobody wants to deal with fluctuation, additional risks, additional security. That’s why I think building software and building a successful software company are becoming even more distinct skills, not less.
yeah i agree. My point was moreso that with more SaaS companies emerging, more winners and loser will increase in the same proportions more or less, but the defunct companies are largely forgotten while the winners remain. So overall more successful companies. I agree that the skills for building a successful software enterprise are still so very rare
This resonates — and there's a wrinkle worth adding from the infrastructure side. The translation layer you describe (domain expert → developer) collapsing is also happening one layer down: between developers and infrastructure. For years the "expensive specialist" problem wasn't just writing app code — it was everything underneath it. Provisioning, auth, data, security, the plumbing that turns a prototype into something real.
What I'm watching: as these mom-and-pop builders ship things that actually get traction, the next wall they hit isn't building, it's operating — and in regulated spaces, compliance. Here's the encouraging part, though: the same way AI let a recruiter build recruiting software without learning to code, it's starting to let people deploy, scale, and secure that software without becoming infrastructure engineers. That's what we're building at Pvotal — not consulting or advice, but the actual guardrailed platform your product stands on, so that deploying, scaling, and staying secure is the default rather than a project you have to staff for. Which is really Jevons applied up the stack: cheaper software creation means more software to run, not less. More builders → more infrastructure demand, not less.
The "economic participation, not developer productivity" framing is the right lens. Great piece.
The compliance wall is real, and the money agrees with you: incumbents spent roughly $70B in a year buying up AI security companies while 175 startups raised $8.5B trying to build the same thing. When buyers outspend builders eight to one, that's a bet the plumbing gets harder before it gets easier. The recruiter who shipped an app still has to answer for who its agents are authorized to act as.
Oh. My. Goodness. Now this is how you explain the real value of vibe-coding. It’s always been in the democratization of capabilities to SMBs not as a replacement for the proprietary platform builds that require an entirely different level of software engineering prowess.
Let's get building ladies! The #SheBuilds event in March this year launched me into a world of possibly and reignited my creativity ✨️ I haven't quite figured out how to monetize but I know I'll get there. Elena, I remember you saying that day that 'we belong in this space so take up space' 🔥 Thank you for encouraging us - we can do hard things!!
I'm really excited to see what AI can help unlock for local businesses and solopreneurs. here in Latvia, we have an abundance of passionate people (and mom-and-pop duets), and with AI, they can a) boost their visibility by getting the fundamentals right (e.g. the go-to website option for these are an instagram or facebook page at the moment) and b) help them with all the admin of being business owners, so they can pour more energy into what they're creating.
Spot on. Previously, just getting a demo to discuss with potential users was difficult and expensive and that's before iterating and tweaking. Now its a prompt. In the hands of domain experts, things are going to get sporty.
Optimistic take, and I follow the storytelling — but two points push back on it.
First, you're underestimating how hard process description actually is. Most people lack the metacognition to articulate what they really do, so they can't reliably describe their own workflows in the first place.
Second, the moment you want to build something more complex than porting a set of Excel files, you need some understanding of how software is designed. That barrier doesn't go away.
Vibe coding will quickly become a commodity. Design will still make the difference.
This: “A soccer coach can create soccer software.” I am a track and field coach attempting to build track and field software. I feel a big sense of urgency to move fast on all my ideas but after every start, I end up hitting some constraint so I pause and move onto the next idea prematurely. Any tips on getting over this hump? Do we stick to one idea or try them all and see what sticks?
I think AI is the era lots of SaaS or software products. You are not tied up with a few browsers or social media giants. You can already see that with the disruption of Google search. I no longer use Google search. I type anything into my favorite browser and the AI engine associated with it takes care of the search and bringing the output. Again we have multiple llm models we can subscribe not just one. The age of software monopoly should be over if AI is properly regulated. It took only 30 minutes for me to design and launch a portfolio website for free which five years ago would have taken weeks or months and costs money.
Domain expert (independent retail) turned saas dev here 🙌🏻 this article nailed my experience. There was such a blind spot in creating inventory solutions for retailers that source from many vendors and are brick and mortar yet every time I met with devs the math didn’t math as you say. Everything was about DTC and Ecomm. I am so proud to say we have very happy beta users and are taking it to the broader market now!! It really is a fantastic time to have a lived problem you could never find a solution for.
thanks for the post! cheaper software creation means more of it, not less, so this isn't devs-get-replaced, it's the market expanding to people who were always priced out.
I think we're conflating two very different things:
Building software.
Building a successful software company.
The first has become dramatically easier. The second has not.
If you're building for yourself, fantastic. AI and no-code tools are unlocking possibilities that didn't exist a few years ago. More people should absolutely build tools to solve their own problems.
But if you're building for customers, coding was never the hardest part. Never.
Much of the business building skills that have been and continue to be needed have not changed, but the expensive, time-consuming, and human-capital intensive nature of designing software has changed. I think this leads to a commoditization of software skill which allows more people who were scared off by the prohibitive cost of making software to try their hand at creating a SaaS company. I don't think that the inherit skill of building said successful SaaS company is any more prevalent in society, but with more people trying it I think the pattern of a non-developer SaaS founder will indeed increase! I think more of the people who CAN build for customers can now come out of the wood works and do so.
I don’t disagree that more people will build SaaS companies. My point is that AI is solving the production problem, not the business problem.
Software is becoming abundant. Customers are not.
As building gets easier, distribution, trust, differentiation, and customer adoption become the real constraints. I don’t even want to go into difficulties of integrating niche SaaS into enterprises. Nobody wants to deal with fluctuation, additional risks, additional security. That’s why I think building software and building a successful software company are becoming even more distinct skills, not less.
yeah i agree. My point was moreso that with more SaaS companies emerging, more winners and loser will increase in the same proportions more or less, but the defunct companies are largely forgotten while the winners remain. So overall more successful companies. I agree that the skills for building a successful software enterprise are still so very rare
1000% agree. Been seeing this crop up all over.
This resonates — and there's a wrinkle worth adding from the infrastructure side. The translation layer you describe (domain expert → developer) collapsing is also happening one layer down: between developers and infrastructure. For years the "expensive specialist" problem wasn't just writing app code — it was everything underneath it. Provisioning, auth, data, security, the plumbing that turns a prototype into something real.
What I'm watching: as these mom-and-pop builders ship things that actually get traction, the next wall they hit isn't building, it's operating — and in regulated spaces, compliance. Here's the encouraging part, though: the same way AI let a recruiter build recruiting software without learning to code, it's starting to let people deploy, scale, and secure that software without becoming infrastructure engineers. That's what we're building at Pvotal — not consulting or advice, but the actual guardrailed platform your product stands on, so that deploying, scaling, and staying secure is the default rather than a project you have to staff for. Which is really Jevons applied up the stack: cheaper software creation means more software to run, not less. More builders → more infrastructure demand, not less.
The "economic participation, not developer productivity" framing is the right lens. Great piece.
The compliance wall is real, and the money agrees with you: incumbents spent roughly $70B in a year buying up AI security companies while 175 startups raised $8.5B trying to build the same thing. When buyers outspend builders eight to one, that's a bet the plumbing gets harder before it gets easier. The recruiter who shipped an app still has to answer for who its agents are authorized to act as.
Oh. My. Goodness. Now this is how you explain the real value of vibe-coding. It’s always been in the democratization of capabilities to SMBs not as a replacement for the proprietary platform builds that require an entirely different level of software engineering prowess.
Let's get building ladies! The #SheBuilds event in March this year launched me into a world of possibly and reignited my creativity ✨️ I haven't quite figured out how to monetize but I know I'll get there. Elena, I remember you saying that day that 'we belong in this space so take up space' 🔥 Thank you for encouraging us - we can do hard things!!
another great read, thanks Elena! :)
I'm really excited to see what AI can help unlock for local businesses and solopreneurs. here in Latvia, we have an abundance of passionate people (and mom-and-pop duets), and with AI, they can a) boost their visibility by getting the fundamentals right (e.g. the go-to website option for these are an instagram or facebook page at the moment) and b) help them with all the admin of being business owners, so they can pour more energy into what they're creating.
Spot on. Previously, just getting a demo to discuss with potential users was difficult and expensive and that's before iterating and tweaking. Now its a prompt. In the hands of domain experts, things are going to get sporty.
Why? Where we will see more customers willing to pay for fluctuated software?
Optimistic take, and I follow the storytelling — but two points push back on it.
First, you're underestimating how hard process description actually is. Most people lack the metacognition to articulate what they really do, so they can't reliably describe their own workflows in the first place.
Second, the moment you want to build something more complex than porting a set of Excel files, you need some understanding of how software is designed. That barrier doesn't go away.
Vibe coding will quickly become a commodity. Design will still make the difference.
Exactly! Integration with the ecosystem, security, data…. It’s a very long list.
This: “A soccer coach can create soccer software.” I am a track and field coach attempting to build track and field software. I feel a big sense of urgency to move fast on all my ideas but after every start, I end up hitting some constraint so I pause and move onto the next idea prematurely. Any tips on getting over this hump? Do we stick to one idea or try them all and see what sticks?
I think AI is the era lots of SaaS or software products. You are not tied up with a few browsers or social media giants. You can already see that with the disruption of Google search. I no longer use Google search. I type anything into my favorite browser and the AI engine associated with it takes care of the search and bringing the output. Again we have multiple llm models we can subscribe not just one. The age of software monopoly should be over if AI is properly regulated. It took only 30 minutes for me to design and launch a portfolio website for free which five years ago would have taken weeks or months and costs money.