Dharmesh Shah: evaluation

Transcript context

…Yep. It's not like just the entire mesh project anymore. Yep. So I don't know exactly where to go with this, but just why is AI so top of mind? And where do you think people should be spending time? Where do you think things are heading? What comes to mind around AI? Yeah. Obviously, I'm not alone in my excitement, but I will say this. So I'm about to hit the 30-year anniversary of my working in commercial software, right? So I've been at this for a long, long time, and that's all I've ever known, that's all I've ever done. And the last time I was this excited was when the web first came along in the mid to late-'90s, and I'm like, "Oh, this literally changed everything," right? It's like relevant to everything. It's like literally everything is going to be impacted by this thing called the internet. And I was in my early twenties, right? I've been here for mobile, I've been here for social, I've seen lots of things that people got really excited about, but nothing to me has hit the Richter scale to the degree that AI does, and the reason is, okay, so if you think about early computers, PCs, that kind of gave us compute at scale, which was awesome. The internet effectively gave us distribution at scale. It allowed us to take information, products, whatever, and get them out at scale because everything was now connected, where before it was not. What AI gives us now is cognition at scale, the ability to literally amplify the human brain, right? That's never been done before. Look, not that we haven't been trying, but now we have demonstrable evidence and products that actually work. We've used ChatGPT and products like it. And so it's life-changing, right? And so for the product people out there, which is a lot of you, AI... By the way, the internet unlocked a massive opportunity because, especially for startups, it's like, "Oh, we can just reimagine the world with this new thing called the internet that was not possible a year ago and now is. What can we now unlock?" and lots of great companies were created. I think something similar will happen here. And there's a couple of things just tactically that I think are possible vectors that can satisfy my spreadsheet of if I were going to choose ideas, and these apply across industries, is that, so as software companies and product builders, most of us have always called our products like, "Oh, we build intuitive products." We say that, right? As it turns out, it's just not true because as a user of X product, whatever it happens to be, it's like, okay, I've got this mental model, the thing I'm trying to accomplish, and then I have to translate that into a series of clicks, drags, touches, and swipes in order to accomplish the thing that I want in the software product that I'm using, and what varies is the degree to which that translation needs to occur. But there is this impedance mismatch, right? There's a translation that has to happen from what's in my head, like, "I want to remove this background in Photoshop," "I want to create this report in HubSpot." re is this impedance mismatch, right? There's a translation that has to happen from what's in my head, like, "I want to remove this background in Photoshop," "I want to create this report in HubSpot." And so the opportunity now is that we're going from what was an imperative model, this is what engineers would call like a step-by-step, you give instructions to the computer like, "Do this, do this, do this, do this," and then hopefully you get the output you're looking for, to what engineers would call a declarative model. A declarative model is you describe the outcome you want, not the steps to get there, and the closest analogy that most people are familiar with is SQL, the querying language, right? So before SQL, if you wanted to get data out of a system, you wrote code that says, "Okay, loop over all my customer records, then filter out the ones that have less than a million dollars in revenue, and then give me the result," and SQL said, "No, just put a where clause on it. Describe the data you want, and what columns you want, and how you want to order it, and what filters you want to apply," and it figures out if the database figures out how to get you the thing that you're looking for. Now we have the ability to do that for interfaces for software, right? So instead of saying, "Click here, drag here," it's like, "I want you to generate a report that's all my customers in Europe that we signed up last quarter that were more than 100K in ARR." And you just literally take the thing that's in your head and express it in whatever medium you choose, text, voice, doesn't matter, and the software now can actually... So now the technology exists to pull that off. Did not exist before, and now it exists. I know it didn't exist before because six and a half years ago, I built this product called GrowthBot. That was a chatbot for HubSpot software and business software generally for marketing and salespeople. It's essentially what ChatSpot is now. Amazing product, GrowthBot, and it had just one teeny tiny small problem, which is it didn't work. Really well-conceived, really brilliant idea. It just didn't work and the technology wasn't there. You just couldn't do it with any degree of fidelity because AI hadn't... And now you can. And that's what caused me to go prove out my theory, is like I had a chip on my shoulder, particularly with that idea, right? But I think, and I'm not suggesting that web and mobile interfaces are going to go away, but there are a class of use cases across pretty much all software where we can make things easier for the human by not forcing that translation layer and just make it truly intuitive. Yeah. So thank you for coming to my TED Talk or my [inaudible 01:37:01] podcast.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.