In this short interview, Jason M. Lemkin and two other speakers discuss how the AI era is compressing software development cycles to 18‑24 months and what separates winning B2B tech companies from those that will fall behind. Lemkin argues that, before AI, a reliable way to gauge a startup’s health was to track story‑point output relative to headcount each quarter—companies that consistently increased story points faster than they added people pulled ahead of competitors. He notes that founders or CTOs who dismiss story‑point tracking tend to see stagnation or decay. Speaker C provides a concise explanation of story points, user stories, and the common Fibonacci‑style scaling used to estimate effort. The conversation frames these metrics as a pre‑AI “signal” that can still inform judgments about which firms are positioned to thrive as AI‑driven acceleration reshapes the market.
Chapters
The rate of change is so fast, you can see that happening in 18 to 24 months, right? The software is just invisibly created behind us by agents.
Can you talk a little bit about the attributes that are gonna differentiate the winners and the losers in this next era? So like, if you had to handicap folks, like, what are some things that you're looking for in terms of like, these companies are gonna make it, these other companies maybe not so much?
For startups, you mean? For like— Or any tech company?
Yeah, yeah, B2B companies, the folks that you see now. I mean, you can smell it.
You can smell the decay.
Yeah. Tell us more about that. How do you, how do you sense that?
Look, pre-AI, you know, as silly as it is, I really think you could measure startups and scale-ups by story points, whatever version of story points you used. Okay. I don't care how you measure story points. I don't care how you calculate it. Come up with some consistent way to measure your output by story points and then each quarter drive it up. And I would go to board meetings and startups and I would see ones where the CTO would say, I don't believe in story. I don't believe. We're just, I believe in best efforts on my team. They would fail. Okay. And then you would see folks that would, that would do it, but you wouldn't see the growth. You would want to see whatever metric you want. You would want to see story points increasing faster than headcount. Okay. Then there was a linear correlation before AI, but you wanted to see however you measured that it would go up each quarter. And if you had a really good team and your story points increased, you would pull away from the competition. Like that was pre-AI, right?
Let's explain to people, because I think a lot of us understand maybe what a story point is, but a story point is like a user story. And then a developer typically assigns points points to this user story, and that's like how long they think it will take to kind of accomplish that task. Yeah. In development. And so in software development, when you're scoping out work, which means like, how long is it gonna take us to build this thing? You have these user stories, and that's what defines the features. And then those points get assigned. And so like what you're talking about is the ability to stack those points basically is showing, hey, our development cycle speed, or we're getting more user stories, we're getting more features shipped.
Yeah. You might, for a feature that took a day, it might have a certain amount of points for a feature that took, You know, the better part of a quarter would have a certain amount of points, and sometimes you'd even do them in a Fibonacci-esque, uh, scale so people could intuitively understand 1, 2, 4, 8, like how much energy went into it.
Yeah.
Mm-hmm. And it, it didn't matter. It was as long as you could roughly, as long as the, your smartest engineers could agree this should take a day, that was your scale. And it worked pretty well, right?
