Lessons from a Failed AI Startup Founder

Key Takeaways

On this show, we talk a lot about the companies that are really thriving. But the reality of startups, of course, is that they are really hard. And not every company pans out the way founders initially expected they would. And so today, we're bringing on one founder who had to make the difficult decision to wind down operations at his company. Aaron Dignon co-founded a company called Plum. It was an AI company that he's been working on for 5 years. And as you can imagine, Aaron has a ton of reflections on why things may not have worked out the way he initially expected. Aaron, welcome to TITV. It's great to have you. >> Hey, thanks. It's good to be here. >> So, you've had a a big week. I mean, it's it's been a big week of decisions for you, it looks like. >> Yeah. Or a half decade, depending on how you count. >> There you go. Yeah. Well, that's true. It was all it's all part of the big story. Um, look, before we talk about the week that you just had and the decision, what was the original vision for POP? >> Yeah, the the core vision here is you see tools like NADN, the open AI agent builder that just came out, uh, you know, Gum Loop and others automating processes and tasks and and work with AI. And our thesis was while there are many people in the world that will take the time to learn how to do that, it's quite technical, no matter how no code it is. And it would be amazing if there were kind of a substack for AI workflows where you could subscribe to someone else's creation that would then automate and customize something in your business life. And so that's kind of the the mission that we ran against. >> Got it. And so you you had that mission. Why did you decide ultimately to wind things down this week? >> I think we have decided that we're either early or wrong and probably early. Uh but the the reality is the the number of people who can actually automate like full agentic workflows in work that actually work right because there's a lot of hype and then there's the reality is a small number and of those a lot of those people benefit greatly by charging a lot of money to customize and consult and and kind of go inside organizations and fix that. So we really had an inability to build a creator economy side of this equation. Lots of people want to subscribe to solutions, but there weren't enough people that were willing to create something that they share at a at a low enough price point to create a whole marketplace uh effect right now. So, I think we're just uh still in the early innings of what agents and agentic workflows are going to do. And we think that that marketplace may may come to fruition later down the road. And so from from your end as you reflect on this, was this more of a of a technical challenge that that really dominated or was it a business model challenge? >> I think it's a business model challenge. We we were able to invent the technology. I mean things like metad schema that allow you to actually share a workflow with thousands of other people. But the but the challenge was more, you know, is the market there? Is the business model right? How do you kind of optimize those those lines? And and while we felt like we were very close, it you know, from a from a like line of sight to venture scale perspective, we just couldn't quite see it. And so we're we're still really committed to the space. We're still really interested in AI automation as as a valuable place to play. We just felt like we needed to like wipe the slate clean and think about it fresh. >> How many employees did you did you have at the company? >> Seven people. Really like a lean kind of 0ero to1 team. Yep. >> Okay. and and what's what's in store for the team now? Are you guys going to try to start a a new startup? I mean, have you started talking about who's going to hire these people? What's the response been? >> It's been a very weird week. So, I I tweeted about our lessons learned and about the shutdown uh you know, day before yesterday and we've had over a hundred inbound AI companies looking after the team. >> Um so, like basically everybody but Sam has DM'd me and been like, "Hey, what's next?" So, we're filtering and sifting through all that and and we we hope to to find a good place to keep this band together because it's an incredible team. Um, so that's that's job one. And then, you know, job two might be to to hang a shingle and and try again. >> And so, are you going to take your hand again at an AI star? Are you going to start something new? >> I think we will. Yeah. The question is whether we do that right away or if we if we find a place to kind of uh rest and exercise some of the skills that we learn before we do that. But I'm I'm a multigenerational entrepreneur and this is my fourth or fifth thing. So I'm definitely going to get the itch again soon. >> Right. I do want to talk to you broadly. I mean you you've you've been in this ecosystem for a while here. You know you have made the difficult decision to wind things down and maybe start fresh. This is something that we don't hear a lot about in Silicon Valley because not because it doesn't happen but because when it does happen few people want to talk about it and it's an emotional decision. >> Yeah. You know, I I I wonder we have all of these AI startups. Not all of them are going to work out. >> Most um most Exactly. And and we're not going to hear about the times when they don't work out because they'll sort of quietly fizzle out. You know, >> I guess what I'm trying to get is have you seen a paradigm shift at all? You know, is it becoming more sort of accepted to sort of talk openly about the times when things don't work out? um you know, how does that bode for the next opportunity people look for at at big tech companies? I mean, has the stigma kind of gone away if there was one? >> I think that the market is ahead of people's mental models on this one. So, you you know, you would think shutting down, you know, struggling to succeed in what you set out to do would come with some judgment. I have heard nothing but support from the broader community. I mean, it's, you know, hundreds of comments and they're all supportive. So, that's unusual on X just in general. Um, and and I feel like a lot of entrepreneurs, especially if they're younger and this is their first or second swing, they they feel like somehow there's something, you know, wrong and that they need to hide from it. And I would just encourage people like this is a really supportive community as as as hard as it can be in competition, it's supportive otherwise. Every single competitor of ours has DM'd me and with supportive words. And so I would say like the the value of the learning far exceeds the the problem of the stigma and and is worth getting out there. >> And last question for you, you know, we talked about how most of these AI startups won't pan out the way people hope. >> What are you concerned about as you sort of watch these startups proliferate? We have these big funding rounds that are coming in. Let's put valuations aside for a And I mean, you know, we'll put the sheer amount of money on the side for a second, but you know, you've sort of you've obviously taken apart your strategy that you've been working on the last 5 years. What concerns you um or what's a bit of a red flag for you in the way that some of these AI startups are going about building their technology or building their business model even? >> I think there are two big concerns that we have and one of them we spent a lot of time on which is just the overall reliability and security of these platforms. There's a lot of excitement in the category about things like MCP that represent real security gaps we we believe. So I think there's going to be some stories of both technology failure like just not living up to the hype, but also like pretty critical gaps in in you know getting information or or uh even money moving around that shouldn't be moving around. So we'll see how that plays out. I think we'll see some some negative you know stuff there. And then the other one is just the underlying economy of AI right now runs on kind of token arbitrage. And and there are a lot a lot a lot of startups that are selling a dollar for 90s right now. And sometimes that pans out if you can achieve enough scale and you can be the biggest planet in the solar system. But for a lot of folks, the math eventually just isn't going to pencil. And I think we'll even see that play out in the public markets as well. >> Great. Well, Erin, thank you so much for coming on the show. We really appreciate it. We appreciate you being candid with us here. It's obvious it's a difficult decision and uh I'm I'm sorry that things didn't work out the way you wanted, but I I I really I'm I'm excited to hear that you have um a lot of enthusiasm for what's coming next. So, thank you for coming on the show. That's Aaron Dignon, the co-founder and the CEO of Plum here on TITV.