Ezinne Udezue: evaluation

Transcript context

…I love so much of the advice you two are sharing. There's a post I'm going to link to in the show notes that shares a bunch of examples of what people are byte coding, building with AI. That seems to be the biggest challenge for a lot of people is just like, "What should I actually do?" And something that I realized as I was putting that together is people think about like, "Oh, can you build real product and businesses on Bolt, Lovable Replit?" But interestingly, most people just want to build a thing that solves a specific problem for themselves. And sometimes that turns into a thing, but there's almost a bigger opportunity just to build those personal softwares similar to the example you shared Ezinne. There's a site, someone built this like, "How many layers should I wear today.com." Or whatever. I don't think that's the actual domain and it's just three. And that's sometimes all you need to know. Today's episode is brought to you by Coda. I personally use Coda every single day to manage my podcast and also to manage my community. It's where I put the questions that I plan to ask every guest that's coming on the podcast. It's where I put my community resources, it's how I manage my workflows. Here's how Coda can help you. Imagine starting a project at work and your vision is clear. You know exactly who's doing what and where to find the data that you need to do your part. In fact, you don't have to waste time searching for anything because everything your team needs from project trackers and OKRs, the documents and spreadsheets lives in one tab all in Coda. With Coda's collaborative all-in-one workspace, you get the flexibility of docs, the structure of spreadsheets, the power of applications, and the intelligence of AI all in one easy-to-organize tab. Like I mentioned earlier, I use Coda every single day. And more than 50,000 teams trust Coda to keep them more aligned and focused. If you're a startup team looking to increase alignment and agility, Coda can help you move from planning to execution in record time. To try it for yourself, go to coda.io/Lenny today and get six months free of the team plan for startups. That's coda.io/Lenny to get started for free and get six months of the team plan, coda.io/Lenny. So let me take this opportunity to zoom out a little bit. We've been talking a lot about personal skills, what you think people need to get better at, what it takes to be successful. I want to talk about at the company level. You two work with a bunch of different companies, you've both worked at a bunch of different companies. When you look at the companies that seem to be doing well in this new AI world, adopting the best practices, seeing productivity gains versus the companies that are just struggling and not getting anywhere and just having a hard time, what do you find is most common across the companies that are succeeding, shifting their culture and the way they work in this world of AI? And I'll throw this to Ezinne first. The main thing I will say is companies that recognize that AI is a core set of multiple capabilities that they've been gifted. I think you need to understand that. It is not this magic thing that you're going to slather on to your product and it will do perfectly. The problems are still the problems. The customer is still the customer and their pain or sharp problems still exist as they have. The idea now is how do you take AI and not have it be something we call it, you put at the edge alone, but you actually re-transform the way you solve the customer's problem? So we have this phrase called AI at the core and AI at the edge. And the best way to think about it is that with AI at the edge, you're probably still using software as you always have. The bits and bytes that you wrote before still are there. You're instead inserting LLMs or AI at different intersections or connection points. Okay? AI at the core means fundamentally looking at the problem space and the workflows and using AI to solve the problem, not just sprinkling it at the intersections of GUIs or user interfaces that are out there or using it to just accelerate things. So the way we talk about it is that companies where their code base remains and they are attaching AI to it often aren't the ones who are going to revolutionize their industry. Companies for whom their code base perhaps shrinks and the LLM becomes a core part of what it is that they used to solve the problem are companies that actually are doing it right. Another characteristic is specificity. We've seen a lot of companies want to do so many things and expect LLMs to be as broad as the problem sets they want to solve for. And we in our experience have seen that to do that you actually will need to have specialized first and then create a layer of a connective tissue that can offer intelligence. So we've seen companies instead just stay on top, try to build this massive solution set or LLM that can do it all, but the successful ones that we've consulted with, partnered with or talking to are those who instead have seen, "This team that I have on here can actually build this very specific solution set, this other team can do this. And then my job is to tie it all together with a broader multi model solution and create an offering for the customer." I think those are two core things. I'm sure Oji can remember a few others. No, I was humble listening to you. No. So maybe I'll embellish the point. We are seeing companies that are starting from a blank slate, taking an old problem, a sharp problem, and saying, "If we built it today with LLM as the core capability as a core blob of code, how would it look?" And often these people are finding a lot of acceleration and they're actually taking on adjacencies. At Formless Typeform, a new product built by David Okuniev and me and a small group of people, we built a new Typeform and we ended up building more than it because it also became a sales lead agent. So it did pre-sales, which ordinarily Typeform gets you into qualification before you talk to a salesperson. But the product itself did pre-sales. So that capability was so strong. We're also seeing the shrinking thing. It's funny, in the last week I've seen companies that have been around for five years who, once LLMs came, they just re-imagined their product. Clay.ai is a six-year-old company, super successful. So in the first category is things like Lovable, but the second category are people who are like, "Okay, we're going to rethink this and we're going to do it." Now, the problem for the rest of the industry is this... [inaudible 00:44:53] is dilemma. They have lots of revenue tied to old code bases. And so their job is to... We tell PM leaders all the time is to navigate how you use LLMs in the old thing, but navigate to the new thing before someone eats your lunch. And it's complicated, it's hard, but you have to do it. Now, let me make maybe a couple of smaller things. The people who are succeeding are taking chance on user experiences. We don't believe that chat interface is the final boss for AI user experience. We think that is a reason viewers exist and the command line didn't work. Otherwise, DOS will be the biggest operating system. And so people have to do different things and be dynamic. One of the big things is dynamic user experience is that just personalized to the customer. So we see a lot of people succeeding there. And one of the things that we don't see people doing actually that we want people to do is something Ezinne mentioned, which is ethics. We think of AI and its core capabilities as ordnance level, meaning... At some point we'll think of this as more powerful than the fusion bomb. And we have a bunch of software people who have no consideration of ethics play with this stuff all the time. And so, one thing we'd like to see the best companies do is lead on that. Think about the responsibility we have to the human race when we build new digital products, I will give them superpowers.…

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