As you’ll have noted from our coverage, as far as startup land is concerned, AI is hot, hot, hot.
Meanwhile, the lean methodology — think of a hypothesis, test it, iterate on it — has been canon for entrepreneurs and founders the world over for the past decade. But AI will most likely play a role in building startups faster, cheaper and more efficiently. So I asked the man who invented the concept of the lean startup, Steve Blank, to see what he thinks.
AI might not have started with ChatGPT, but the ability for the general public to interact with generative AI on a wide scale did. But even still, Blank says we’re collectively underestimating the potential of generative AI.
“I’m surprised no one has done a parody of actually reviewing a three-month-old baby,” Blank said, “and saying all it does is poop in its pants, and it can’t even finish complete sentences, rather than going, ‘Holy shit! This is like the first week of this thing’s life!’ and we really don’t realize it.”
And, just like a three-month-old, nobody can fully predict what ChatGPT has in store for us next. “Copilot has changed the life of every programmer, period, who writes software,” Blank said. “It’s probably increased productivity by 50%, and that’s if you’re using it poorly. That’s a mind-blower.”
Blank highlights how AI-assisted research has made leaps of progress in the form of AlphaFold, a project that is trying to translate proteins into their three-dimensional structures, which can help us understand processes in the body, including aging. AI, obviously, goes far beyond, and we’ve only just started to see the tremendous evolution that will span all sciences. Blank might be on to something: Just last week, researchers at University College, London and Moorfields Eye Hospital in the U.K. identified markers for Parkinson’s disease in eye scans using AI.
That kind of diagnosis might’ve taken years without the help of AI.
Blank is wide-eyed about where it can yet go, considering how much money is being thrown at AI projects without much VC scrutiny. And to him, the concept of the “lean startup” is about to change forever. “I just think it’s going to be an amazing ride,” he said. “Part of the amazing ride, let’s go back to the VC part, is that the dumb money is just poured into it. Think about it: There’s probably 10,000 experiments being run with capital. It’s not 10,000 experiments being run in your basement; there’s a ton of stuff that’s been thrown against the wall to see if any bit sticks.”
But what of the potential that could be unleashed by attaching AI to the lean methodology?
At its core, the lean methodology is about devising a hypothesis, testing it and iterating on it to produce the optimal outcome. In an e-commerce setting right now, without AI, it’s possible to purchase fairly complete datasets that you can use to make reasonable tests and suggestions for potential products and services that could be viable business propositions. But what if you were to throw AI at this problem?
“It dawned on me that we’re going to take this whole lean startup pipeline, and if not by the end of this year, certainly in the foreseeable future, that machines will be doing this a lot better than human beings,” Blank said.
Introducing AI to the lean pipeline starts with taking those already available datasets and using them to fabricate synthetic humans. You can then ask this city’s-worth or country’s-worth of manufactured “people” what products appeal to them within a given bracket (e.g., toiletries or household goods). This can very quickly present you with reams of potential products to develop — or to avoid — that have a chance of making you money. It can certainly spit out more ideas with greater speed than an entire team of humans working from the same dataset could manage. But what if you could take that a step further down this e-commerce lean pipeline?
Blank explained that you could use AI to generate a slew of minimum viable products and with them images and websites. With these images and websites, you’d be able to run A/B tests 100 times, maybe even 1,000 times, better than real humans throwing up test websites with imaginary products and photos. By sheer weight of numbers, you’re able to find new openings and possibilities, but without the cost of human input.
“I could imagine four people and their dog having the [same] reach in commerce as a company that took hundreds of people not more than 18 months ago,” Blank said.
Blank relates the potential of AI to the evolution of websites, growing from individual elements that needed to be created or completed using specialist knowledge (e.g., HTML) and self-maintained servicing (e.g., data centers) to all-in-one website creation.
“I think we’ll go through that same evolution,” Blank said. “I think this will create a whole new wave of new enterprise software companies selling to both startups and large corporations who want to automate this. And I think it’s going to be an arms race when the dust settles in this area of innovation [and] entrepreneurship.”
When you take a step back from the hustle and the noise of AI, it has the potential to create enormously exciting opportunities. It’s just a case of being able to see them.