Everyone’s saying Lean is dead: too slow for AI, too analog.
It's a very thought provoking post and I've been exploring this on my own for a while. I arrived at a different conclusion, Ill try to break it down below:
The main argument behind "Lean is dead" is that if AI can simulate users, generate code, why test, iterate, or validate?
Here's what I think: at its core, Lean is a mindset: build less until you know more +
+ test your riskiest assumptions early + let users (not egos) shape the product.
That doesn’t go away with AI.
If anything, AI makes true Lean faster.
Yes, the cost of building is nearly zero, but the cost of building the wrong thing is similar, and the mindset is still relevant too.
And “nearly zero” applies more to prototypes than production-grade, scalable, secure, ethical products.
Great reply. I also see Lean Startup as a mindset more than a method. When I started participating in Lean Startup after years of Waterfall, it was a complete mindset shift.
My stack would be…
Systems thinking…to create better foundational paradigms and models. This also is more of a mindset shift.
Lean Startup to gauge initial interest in an idea.
Agile for when you decide to formally build and scale.
> As a result, significant emphasis was placed on validating, defining, and refining product requirements before writing any code. Organizations grew entire functions dedicated to researching, designing, and validating concepts prior to development.
What? This is the exact opposite of the lean startup approach, which is about validating ideas quickly by building prototypes. It seems like you’re describing waterfall.
When code was expensive to write, it was often faster/cheaper to try to invalidate an idea without writing any code. So via customer research or mockups, companies could figure out what not to build without having to write code. That was considered a lean best practice. My point is that it no longer makes sense to do that, you can just go straight to a working prototype.
So is the lean startup dead? Arguably not, but the title “The old way we did the lean startup is dead and now there’s a new way”… just isn’t as catchy.
I agree with you, JT. One thing that happened to agile was that it was grossly misunderstood as just ‘discovery first’. It’s more radical than that, and mid- to big organizations seldom reached that level. Perhaps OP never got to see it close.
Great post, and thought provoking. I asked Chat GPT what it thought :)
“The article rightly notes that AI accelerates software development and shifts bottlenecks to adoption, but misframes this as the death of Lean Startup. Faster prototyping and iteration still align with Lean principles. It overlooks that product judgment, complex scaling, and quality assurance remain critical and aren't easily automated.”
As a marketer I am still giving you 10 /10 for the “The Lean Startup is Dead” headline!
Disagree. The generative code typical of tools like Cursor is not production-quality. It’s great for prototyping and proof of concept work but I wouldn’t use it for shippable products at scale. It may get there in a couple of years but, as of today, I wouldn’t trade in a small team of seasoned engineers for what is essentially text-based Dreamweaver on steroids.
I do agree that product designers will become more merged with their FE engineering counterparts. Not sure about the PMs being traded for PMMs—maybe, sure. More likely that the PM and product designer roles will merge.
I think finding PMF and TAM is as hard as it ever was. Access to tools like this will definitely clog the market and you’ll have a lot of small startups that will die as quickly as they launch. That said, I think it’s still pretty much the same old game.
Also, it’s “knock-on” effects, not “knock-off” effects. A knock-off is an imitation, as in “she bought a shitty Chanel knock-off on Canal Street.” Knock-on effects refer to a series of causal events, not dissimilar to a domino effect—as in, a chain of events, one “knocking onto” (usually causing) the next one.
PMF and TAM continue to be as challenging as ever. 💯 right. Perhaps AI and vibe coding give the impression that it can be done cheaply, which places pressure on everyone.
The attitude described in Lean Startup? It’s the same.
Some commenter picked up on that passage: “significant emphasis was placed on validating, defining, and refining product requirements before writing any code.“
They suggest that it’s the opposite of what Lean Startup is.
I’d agree only partially. On “defining and refining,” yup, I’m with other commenters. It’s not what Lean Startup is.
It is all about validation, though.
And that part is no different in dot-ai era than it was in dot-com one (and anytime in between).
What has changed is that we can, indeed, build prototypes faster. And all the smart Lean Startup folks would totally capitalize on these new capabilities. They’d insist on using LLMs during the ideation (pre-coding) phase to improve on the ideas.
A simple fact that it is cheap(er) to build a feature or a prototype does not mean that one should build it. That part of the Lean Startup body of knowledge hasn’t changed a bit.
Sure, these days, having the same funds/time/resources, you can flail around more. But flailing around was never a recipe for a successful product.
And I’m saying that from a perspective of accompanying 100+ startups on their product development journey.
Those that failed (the vast majority, duh!) haven’t done so because development was their bottleneck. I’m not sure if more efficient would save any single one of them, in fact.
If they studied what happened under the product development/Lean Startup umbrella over the last decade or so, some would still be alive. Some more would never start. And many would cut their losses early.
Where is step 0? Where do the features come from? Also, there’s more to it than adding features. Sometimes you have to step back and look at the whole system. I’m not saying AI will never be able to do that, but I don’t think we can see that yet.
While I appreciate the excitement around generative AI and the changes it brings, I don't think the Lean Startup approach is "dead"—it’s just misunderstood if you look at it in isolation. The Lean Startup is not a magic bullet for understanding markets or uncovering customer jobs. It belongs squarely in the solution phase, after you’ve already done the hard work of validating your problem and customer understanding.
In fact, if you dig a bit deeper into the origins of Lean Startup, you’ll see that it has its roots in Steve Blank’s concept of Customer Development, which is all about building a solution based on real customer insights. You can check out Steve's work https://www.startuplessonslearned.com/search?q=%22customer+development%22, where you’ll find his focus on understanding the problem before even thinking about a solution. This goes even further to the first thoughts of Peter Drucker and co.
Whether the code (or solution) is written by AI, a human, or a hybrid, you can’t bypass the critical work of defining the job-to-be-done and deeply understanding the user’s pain points. If you dive straight into a solution (even with AI’s help), you’re not building a business—you’re just building a product, and that's a very different thing.
The Lean Startup is about optimizing the solution phase (and have an alternative to waterfall development) once you've nailed the problem. It helps you move faster, validate assumptions quickly, and iterate based on feedback. But without the rigorous problem discovery process that informs your customer’s needs, you’ll just be another startup throwing out random features into an already saturated market.
So, yes, Lean Startup is far from dead—it’s a powerful tool, but only once you’re in the right phase of building your business.
In lean I never heard of a ‘solutions phase’ as the core concept is about hypothesis testing. Agile is, alas, profoundly misunderstood and seldom practiced with simplicity. Too many people are waiting on the wings for their ‘phase’.
Hi CJ, thanks for the valuable comment. Sorry, this one takes a bit longer to explain my thoughts:
Eric Ries refers to Lean Startup as a replacement for traditional waterfall methods (and an replacement/extension of Steve Blanks Customer development without waterfall practice), as far as he is describing in the book and in his videos. It provides a “scientific” approach to the building, measuring, and learning loop (and pivot) — essentially forming the foundation for a hypothesis driven “solution phase” instead of “waterfalling”. Steve Blank, in his blogs and videos, frequently explains that Lean Startup sits at the intersection of intersection of Customer Development, Agile Development and if available, open platforms and open source tools. (https://www.startuplessonslearned.com/search?q=%22customer+development%22).
He says lately in addition, Alexander Osterwalder’s Business Model Canvas offers a visual way to describe and iterate on an entire business model. It complements Steve Blank’s method by filling in the missing pieces, particularly around the broader business perspective.
From my humble understanding, whichever method you choose for agile business model innovation — which is what it's really all about — you’ll often find yourself combining different frameworks and tools. No single method/framework is truly holistic or practical on its own. For example: how do you conduct user research in Customer Development or Lean Startup? How do you form hypotheses about your first user cohort and validate them in real life? “Get out of the building”, “there are no facts inside your building” is a bit shallow and has not enough substance. So in other words Customer Development and Lean Startup doesn’t fully address these questions. Or how do you assess different business opportunities (remember: to have a good idea you need the many) and assess them before starting any user research activity?
That’s why you're often forced to integrate other methods and tools — like Opportunity Mapping, Jobs-to-be-Done theory. Lean UX Canvas — to add more depth and make things actionable. So yes, you’re absolutely right when you say, “In Lean I never heard of a ‘solutions phase’”. In the end, it’s about applying methods in real life — and that requires more than just theory. You need to adapt and combine approaches to make them truly work in practice.
I also agree with your comment about “Agile”, but ultimately, it’s about finding and validating a new business model — and that’s no easy task. It takes real substance and a structured approach, otherwise you risk ending up like the >90% of new ventures that die in the valley of death before discovering a viable business model. Your opinion?
It’s an interesting perspective, Fletcher, full of tech optimism.
After breaking down Lovable’s strategy last week, I’m sharing a good bit of that optimism. Still, I’m missing the whole discovery piece - strategy, interviewing target users, sensing & learning to address unmet customer needs.
And I don’t see where you addressed the main part of Lean Startup - rapidly testing with target users, learning & iterating before building anything, then just building enough (MVP) and iterating from there.
Fully-coding bad features faster with AI won’t make them better features.
And you can only A/B test so much trying to squeeze another .0003% conversion rate out of a different color blue on a “Buy” button.
The smallest thing to validate an idea will always survive.
The core issue; you nailed it; becomes super compressed and some of the ideas in the lean startup can become fully functional ideas sooner, but failure inherently comes sooner too.
Therefore the idea of the pivot becomes such a sensitive button to press.
I am not sure why people look for teams with a functional divide. That’s a big mistake and lowers collaboration and therefore increases cycle to innovative concepts.
This is a fascinating and thought-provoking take. I usually share such posts with my team and peers to stir deeper thinking — and I genuinely appreciate contrarian perspectives like this.
That said, my one recurring concern with such opinions is this: they often only scratch the surface. We hear bold claims like, “Marketing Management is dead,” or “Peter Drucker is outdated,” usually backed by a couple of recent examples. But very few of us actually take the time to do even a semi-detailed breakdown of how much of it is dead, and what should replace it. If we’re to abandon Drucker, then give us something new to wrestle with — not just rejection without replacement.
I admire frameworks like Lean Startup or work from Jim Collins and Eric Ries — not because they were always right, but because they were backed by years or even decades of deliberate research and pattern-finding.
So yes, let’s challenge dogma. But let’s do it with some depth, even if messy. More of that, please.
I don’t agree with that take, speaking as both a software developer and an indie hacker.
Yes, AI is accelerating development and making it easier to ship features and products. But that doesn't kill the lean startup, it just shifts the game.
As the technical barrier lowers, more products will flood the market and many will become commodities. That pushes the goalpost of what people are willing to pay for toward more complex, higher value solutions.
Nobody would pay for a basic word processor today, not because it's not useful, but because I could build one in an hour. The bar for value is rising and lean thinking is still crucial for figuring out what’s actually worth building in that new landscape.
Individual People will iterate faster, but I think it’ll be hard for already existing companies to change their product “culture” overnight - you see this with the droves of people saying things like “vibe coding is irresponsible, so I am NOT using cursor!”
I think companies and teams that empower their builders to move faster, with caution, will be the ones that truly benefit from this.
> if the primary bottleneck isn’t writing code anymore, then what is it?
Find problems worth solving. Specifically, searching for People in Pain who could become users and ultimately paying customers. This requires identification and quantification of special conditions, including empowered, accessible People with frequent, urgent, expensive, excruciating Pain. Once a founder identifies those People in Pain, I agree that it's easier than ever to provide products to solve their problems. At that point, user attention will be a bottleneck, but that's downstream -- a secondary problem. The PRIMARY bottleneck remains evaluating ideas for startups.
Another outcome of Ai is that it's easier for founders to throw spaghetti against the wall and see what sticks. Specifically, founders can use Ai to quickly build products and try to get feedback. But this guess-and-check approach is an unfortunate spin on your idea that "There will be proliferation of products in every category". Nope, guess and check will lead to a proliferation of "CATEGORIES", most of which will be for problems that nobody cares about.
A humble thought: Lean Startup is not dead, but it needs to be seen in context. It’s an agile method for building a solution within a startup. And in a startup, your product is not the product — your business model is.
Often, the scarcest resource and biggest bottleneck is the entrepreneur’s ability to turn an idea into a solid, validated business model that gains initial traction. In the past, this was mostly about software startups coding — but that’s becoming less relevant these days as you were mentioning.
If you see Lean Startup without Customer Development, the methodology from Steve Blank, you’re only looking at the solution phase.
That’s basically the founder’s main job. Maybe it’s more about “learn before you lean.” An article on this will be coming soon — follow me if you want to read it.
Everyone’s saying Lean is dead: too slow for AI, too analog.
It's a very thought provoking post and I've been exploring this on my own for a while. I arrived at a different conclusion, Ill try to break it down below:
The main argument behind "Lean is dead" is that if AI can simulate users, generate code, why test, iterate, or validate?
Here's what I think: at its core, Lean is a mindset: build less until you know more +
+ test your riskiest assumptions early + let users (not egos) shape the product.
That doesn’t go away with AI.
If anything, AI makes true Lean faster.
Yes, the cost of building is nearly zero, but the cost of building the wrong thing is similar, and the mindset is still relevant too.
And “nearly zero” applies more to prototypes than production-grade, scalable, secure, ethical products.
On the mark! You should write the article,
‘AI makes true Lean faster.’
Thank you Som! That's a good idea hehe:)
Great reply. I also see Lean Startup as a mindset more than a method. When I started participating in Lean Startup after years of Waterfall, it was a complete mindset shift.
My stack would be…
Systems thinking…to create better foundational paradigms and models. This also is more of a mindset shift.
Lean Startup to gauge initial interest in an idea.
Agile for when you decide to formally build and scale.
> As a result, significant emphasis was placed on validating, defining, and refining product requirements before writing any code. Organizations grew entire functions dedicated to researching, designing, and validating concepts prior to development.
What? This is the exact opposite of the lean startup approach, which is about validating ideas quickly by building prototypes. It seems like you’re describing waterfall.
Seems like an intensification of the lean startup philosophy if anything
When code was expensive to write, it was often faster/cheaper to try to invalidate an idea without writing any code. So via customer research or mockups, companies could figure out what not to build without having to write code. That was considered a lean best practice. My point is that it no longer makes sense to do that, you can just go straight to a working prototype.
So is the lean startup dead? Arguably not, but the title “The old way we did the lean startup is dead and now there’s a new way”… just isn’t as catchy.
Not catchy but isn’t misleading. Your title is! Click baits are not a good way to improve your reputation…
I agree with you, JT. One thing that happened to agile was that it was grossly misunderstood as just ‘discovery first’. It’s more radical than that, and mid- to big organizations seldom reached that level. Perhaps OP never got to see it close.
Great post, and thought provoking. I asked Chat GPT what it thought :)
“The article rightly notes that AI accelerates software development and shifts bottlenecks to adoption, but misframes this as the death of Lean Startup. Faster prototyping and iteration still align with Lean principles. It overlooks that product judgment, complex scaling, and quality assurance remain critical and aren't easily automated.”
As a marketer I am still giving you 10 /10 for the “The Lean Startup is Dead” headline!
Disagree. The generative code typical of tools like Cursor is not production-quality. It’s great for prototyping and proof of concept work but I wouldn’t use it for shippable products at scale. It may get there in a couple of years but, as of today, I wouldn’t trade in a small team of seasoned engineers for what is essentially text-based Dreamweaver on steroids.
I do agree that product designers will become more merged with their FE engineering counterparts. Not sure about the PMs being traded for PMMs—maybe, sure. More likely that the PM and product designer roles will merge.
I think finding PMF and TAM is as hard as it ever was. Access to tools like this will definitely clog the market and you’ll have a lot of small startups that will die as quickly as they launch. That said, I think it’s still pretty much the same old game.
Also, it’s “knock-on” effects, not “knock-off” effects. A knock-off is an imitation, as in “she bought a shitty Chanel knock-off on Canal Street.” Knock-on effects refer to a series of causal events, not dissimilar to a domino effect—as in, a chain of events, one “knocking onto” (usually causing) the next one.
PMF and TAM continue to be as challenging as ever. 💯 right. Perhaps AI and vibe coding give the impression that it can be done cheaply, which places pressure on everyone.
The tools we use are changing, indeed.
The attitude described in Lean Startup? It’s the same.
Some commenter picked up on that passage: “significant emphasis was placed on validating, defining, and refining product requirements before writing any code.“
They suggest that it’s the opposite of what Lean Startup is.
I’d agree only partially. On “defining and refining,” yup, I’m with other commenters. It’s not what Lean Startup is.
It is all about validation, though.
And that part is no different in dot-ai era than it was in dot-com one (and anytime in between).
What has changed is that we can, indeed, build prototypes faster. And all the smart Lean Startup folks would totally capitalize on these new capabilities. They’d insist on using LLMs during the ideation (pre-coding) phase to improve on the ideas.
A simple fact that it is cheap(er) to build a feature or a prototype does not mean that one should build it. That part of the Lean Startup body of knowledge hasn’t changed a bit.
Sure, these days, having the same funds/time/resources, you can flail around more. But flailing around was never a recipe for a successful product.
And I’m saying that from a perspective of accompanying 100+ startups on their product development journey.
Those that failed (the vast majority, duh!) haven’t done so because development was their bottleneck. I’m not sure if more efficient would save any single one of them, in fact.
If they studied what happened under the product development/Lean Startup umbrella over the last decade or so, some would still be alive. Some more would never start. And many would cut their losses early.
"significant emphasis was placed on validating, defining, and refining product requirements before writing any code"
Sorry, but this is the exact *opposite* of Lean and Agile.
The "new product development cycle" that you amazingly invented IS what Lean and Agile are all about.
🙄
Where is step 0? Where do the features come from? Also, there’s more to it than adding features. Sometimes you have to step back and look at the whole system. I’m not saying AI will never be able to do that, but I don’t think we can see that yet.
Literally just wrote about shifts in GTM emphasizing the need for creativity. I’m a bit scared of (and I’m already feeling) the battle for attention.
Eventually it’ll become non-effective so either algos or in-person is going to dominate GTM. We shall see!
While I appreciate the excitement around generative AI and the changes it brings, I don't think the Lean Startup approach is "dead"—it’s just misunderstood if you look at it in isolation. The Lean Startup is not a magic bullet for understanding markets or uncovering customer jobs. It belongs squarely in the solution phase, after you’ve already done the hard work of validating your problem and customer understanding.
In fact, if you dig a bit deeper into the origins of Lean Startup, you’ll see that it has its roots in Steve Blank’s concept of Customer Development, which is all about building a solution based on real customer insights. You can check out Steve's work https://www.startuplessonslearned.com/search?q=%22customer+development%22, where you’ll find his focus on understanding the problem before even thinking about a solution. This goes even further to the first thoughts of Peter Drucker and co.
Whether the code (or solution) is written by AI, a human, or a hybrid, you can’t bypass the critical work of defining the job-to-be-done and deeply understanding the user’s pain points. If you dive straight into a solution (even with AI’s help), you’re not building a business—you’re just building a product, and that's a very different thing.
The Lean Startup is about optimizing the solution phase (and have an alternative to waterfall development) once you've nailed the problem. It helps you move faster, validate assumptions quickly, and iterate based on feedback. But without the rigorous problem discovery process that informs your customer’s needs, you’ll just be another startup throwing out random features into an already saturated market.
So, yes, Lean Startup is far from dead—it’s a powerful tool, but only once you’re in the right phase of building your business.
In lean I never heard of a ‘solutions phase’ as the core concept is about hypothesis testing. Agile is, alas, profoundly misunderstood and seldom practiced with simplicity. Too many people are waiting on the wings for their ‘phase’.
Hi CJ, thanks for the valuable comment. Sorry, this one takes a bit longer to explain my thoughts:
Eric Ries refers to Lean Startup as a replacement for traditional waterfall methods (and an replacement/extension of Steve Blanks Customer development without waterfall practice), as far as he is describing in the book and in his videos. It provides a “scientific” approach to the building, measuring, and learning loop (and pivot) — essentially forming the foundation for a hypothesis driven “solution phase” instead of “waterfalling”. Steve Blank, in his blogs and videos, frequently explains that Lean Startup sits at the intersection of intersection of Customer Development, Agile Development and if available, open platforms and open source tools. (https://www.startuplessonslearned.com/search?q=%22customer+development%22).
He says lately in addition, Alexander Osterwalder’s Business Model Canvas offers a visual way to describe and iterate on an entire business model. It complements Steve Blank’s method by filling in the missing pieces, particularly around the broader business perspective.
From my humble understanding, whichever method you choose for agile business model innovation — which is what it's really all about — you’ll often find yourself combining different frameworks and tools. No single method/framework is truly holistic or practical on its own. For example: how do you conduct user research in Customer Development or Lean Startup? How do you form hypotheses about your first user cohort and validate them in real life? “Get out of the building”, “there are no facts inside your building” is a bit shallow and has not enough substance. So in other words Customer Development and Lean Startup doesn’t fully address these questions. Or how do you assess different business opportunities (remember: to have a good idea you need the many) and assess them before starting any user research activity?
That’s why you're often forced to integrate other methods and tools — like Opportunity Mapping, Jobs-to-be-Done theory. Lean UX Canvas — to add more depth and make things actionable. So yes, you’re absolutely right when you say, “In Lean I never heard of a ‘solutions phase’”. In the end, it’s about applying methods in real life — and that requires more than just theory. You need to adapt and combine approaches to make them truly work in practice.
I also agree with your comment about “Agile”, but ultimately, it’s about finding and validating a new business model — and that’s no easy task. It takes real substance and a structured approach, otherwise you risk ending up like the >90% of new ventures that die in the valley of death before discovering a viable business model. Your opinion?
It’s an interesting perspective, Fletcher, full of tech optimism.
After breaking down Lovable’s strategy last week, I’m sharing a good bit of that optimism. Still, I’m missing the whole discovery piece - strategy, interviewing target users, sensing & learning to address unmet customer needs.
And I don’t see where you addressed the main part of Lean Startup - rapidly testing with target users, learning & iterating before building anything, then just building enough (MVP) and iterating from there.
Fully-coding bad features faster with AI won’t make them better features.
And you can only A/B test so much trying to squeeze another .0003% conversion rate out of a different color blue on a “Buy” button.
The smallest thing to validate an idea will always survive.
The core issue; you nailed it; becomes super compressed and some of the ideas in the lean startup can become fully functional ideas sooner, but failure inherently comes sooner too.
Therefore the idea of the pivot becomes such a sensitive button to press.
I am not sure why people look for teams with a functional divide. That’s a big mistake and lowers collaboration and therefore increases cycle to innovative concepts.
This is a fascinating and thought-provoking take. I usually share such posts with my team and peers to stir deeper thinking — and I genuinely appreciate contrarian perspectives like this.
That said, my one recurring concern with such opinions is this: they often only scratch the surface. We hear bold claims like, “Marketing Management is dead,” or “Peter Drucker is outdated,” usually backed by a couple of recent examples. But very few of us actually take the time to do even a semi-detailed breakdown of how much of it is dead, and what should replace it. If we’re to abandon Drucker, then give us something new to wrestle with — not just rejection without replacement.
I admire frameworks like Lean Startup or work from Jim Collins and Eric Ries — not because they were always right, but because they were backed by years or even decades of deliberate research and pattern-finding.
So yes, let’s challenge dogma. But let’s do it with some depth, even if messy. More of that, please.
I don’t agree with that take, speaking as both a software developer and an indie hacker.
Yes, AI is accelerating development and making it easier to ship features and products. But that doesn't kill the lean startup, it just shifts the game.
As the technical barrier lowers, more products will flood the market and many will become commodities. That pushes the goalpost of what people are willing to pay for toward more complex, higher value solutions.
Nobody would pay for a basic word processor today, not because it's not useful, but because I could build one in an hour. The bar for value is rising and lean thinking is still crucial for figuring out what’s actually worth building in that new landscape.
Individual People will iterate faster, but I think it’ll be hard for already existing companies to change their product “culture” overnight - you see this with the droves of people saying things like “vibe coding is irresponsible, so I am NOT using cursor!”
I think companies and teams that empower their builders to move faster, with caution, will be the ones that truly benefit from this.
More Rigor, less Rigidity.
> if the primary bottleneck isn’t writing code anymore, then what is it?
Find problems worth solving. Specifically, searching for People in Pain who could become users and ultimately paying customers. This requires identification and quantification of special conditions, including empowered, accessible People with frequent, urgent, expensive, excruciating Pain. Once a founder identifies those People in Pain, I agree that it's easier than ever to provide products to solve their problems. At that point, user attention will be a bottleneck, but that's downstream -- a secondary problem. The PRIMARY bottleneck remains evaluating ideas for startups.
Another outcome of Ai is that it's easier for founders to throw spaghetti against the wall and see what sticks. Specifically, founders can use Ai to quickly build products and try to get feedback. But this guess-and-check approach is an unfortunate spin on your idea that "There will be proliferation of products in every category". Nope, guess and check will lead to a proliferation of "CATEGORIES", most of which will be for problems that nobody cares about.
A humble thought: Lean Startup is not dead, but it needs to be seen in context. It’s an agile method for building a solution within a startup. And in a startup, your product is not the product — your business model is.
Often, the scarcest resource and biggest bottleneck is the entrepreneur’s ability to turn an idea into a solid, validated business model that gains initial traction. In the past, this was mostly about software startups coding — but that’s becoming less relevant these days as you were mentioning.
If you see Lean Startup without Customer Development, the methodology from Steve Blank, you’re only looking at the solution phase.
That’s basically the founder’s main job. Maybe it’s more about “learn before you lean.” An article on this will be coming soon — follow me if you want to read it.