[MUSIC] >> Welcome to Category Visionaries, the show dedicated to exploring exciting visions for the future from the founders or in the front lines building it. In each episode, we'll speak with a visionary founder who's building a new category or reimagining an existing one. We'll learn about the problem they solve, how their technology works, and unpack their vision for the future. I'm your host, Brett Stapper, CEO of Frontlines Media. Now, let's dive right into today's episode. [MUSIC] >> Hey everyone, and welcome back to Category Visionaries. Today, we're speaking with Dion Nicholas, CEO and co-founder of For Thought, a customer support platform that's raised $92 million in funding. Dion, welcome to the show. >> Brett, thank you so much for having me. >> Not a problem, super excited, and let's go ahead and just jump right in and talk to us about what you're building today. >> Absolutely. So, we at For Thought are building the most advanced, generative AI agent for customer support. Basically, it's fine-tuned on a customer's, or on a business's real customer data, and it can then resolve issues, discover generative insights, and then assist human agents as well. >> From my conversations with other AI builders and AI founders, what they've told me is that the company is kind of divided into two distinct chapters. There's post chat GPT and pre-chat GPT. Talk to us about pre-chat GPT for you. >> Yeah, definitely. So, it's been interesting when you think about pre-chat GPT to post chat GPT. I would say the biggest thing that's changed is it's been the market awareness, the market demand, and the market education. We started in 2017, officially launched in 2018. And so, right around the time we launched, we were always thinking about this modern state-of-the-art AI. At the time, the technology was based on neural networks and transformer models, like the T in GPT, had really just been released. And so, we were always trying to be on the forefront of AI for customer service. But it was interesting because people didn't yet realize that they needed AI, right? Because we were going up against clunky chatbots that were decision tree-based, rules-based, and stuff like that. And everyone was like, isn't that already AI? Like, what are you doing? Like, what is this AI thing that you're doing that's different than what's out there already? Like, why would we need other AI? And so, we had to spend a lot of our time banging on people's doors to explain to them why our technology was different. And then once they would see it, they were like, "Oh, okay, this is cool." Post chat GPT, it's completely flipped. We spend a lot of time, people banging on our door saying, "Hey, we need AI. We now know, you know, what is possible with this stuff?" And we want the best-in-class AI for customer service. And so, the demand curve has really shifted and the education of the markets really shifted. And take us back to those early days, 2017, the founding of the company. What was your big aha to say, "I want to go out and build something like this?" Yeah, so, let's actually go way, way back. So, for context on my background, I grew up in Canada, inner-city Toronto. I had been interested in technology and coding for a very long time since I was a kid. And in high school, I had the good fortune to intern at an AI lab. It was called "Ani" or the Alberta Machine Intelligence Institute in Canada. And that was where I learned about AI for the first time. And I became obsessed, pretty fascinated with the technology, particularly a branch of AI called question answering. So, can AI ingest information and then answer questions for people? As a student, I was always fascinated with, "Okay, can AI help me with my study questions, right?" Like history questions or whatever. But I had also previously worked in customer service, stocking shelves at shoppers' drug bar, which is like the Canadian CVS, so to speak, answering customer questions and things like that. And so, this concept of, "Hey, can AI help people answer customer questions?" Or really, any kind of question was also just something that became a thread. And so, throughout my life, I kept seeing these threads of AI that can answer questions. Or really just being fascinated with, "Is that even possible?" Fast forward many years, so this was 2017. I was working as a software engineer at the time at Pure Storage in Silicon Valley. And I started to see a similar pattern, right? So, with Pure, they're a big tech software and hardware company. And they were priding themselves on customer service. They would even offer guarantees to some of their customers based on their customer service. So, this wasn't just something that was a curiosity. It was actually a competitive advantage for some of these businesses. And so, being in Silicon Valley, I started to get the startup bug and started thinking about, "Hey, would I go start a company?" And somebody told me, "You know, if you're going to start a company, pick a problem that hasn't left you for years. Pick a problem that you just can't put down." And for me, it was really this AI question answering problem. And so, it became very clear to me that I was going to go tackle this market somehow. So, spent a few months figuring out the models. Is it possible to build something in this space? And then, once we got the conviction, it was really off to the races. What were those first six months like as you started to bring this to market? Yeah. So, a few things we had to do. So, I would say, you know, even though this was all kind of one condensed phase, there were kind of three phases. So, the first was technology risk, like figuring out, can the tech work? And I'd already been doing some research in this space. I had an as an early advisor, Chris Manning, who's the professor at Stanford, who basically all the books that I was reading to figure it all out. So, had some early folks in some research in the area, I'm figuring out whether the technology was possible. The next step was raising a little bit of money. So, just trying to figure out if we could even raise a pre-seed round. And luckily, we were able to raise about five or 600k from K9 Ventures, which is a really great pre-seed fund. And then, from there, it was all about finding our first customers, right? And just figuring out, hey, okay, the technology works. There's something here, but is this something that people want? And that became the most critical focus in those kind of early 2018 days. How long did it take to find your first paying customer and have revenue start coming in? Good question. So, it's interesting because when I just started this company, I love to read. I read a lot. I try to learn from the experts. I try to figure out what I don't know. I don't know in order to kind of short circuit any mistakes. And so, I would watch a lot of the YC startup school videos. I would read a lot of the Lean Startup methodology, like Eric Reese's Lean Startup, Steve Blank's books as well. And one of the recurring themes that I started to see was that, more important than your technology or your product, is your ability to solve a problem for customers, right? And I think that's actually a common pitfall that technical founders often have, is like, we think that, hey, if we build it, they will come. But actually spending as much time trying to find your early customer base is as important as building the early technology. So, I was lucky enough to kind of internalize that lesson early on. And so, every conversation we had, whether it was an investor who said no, or, you know, an advisor or a friend or whatever, I would ask for an introduction, hey, do you know somebody in the customer support world or in operations who would be interested in giving us feedback? And then that became that loop. And I would do it every single time. And so, within a few months, I think we launched our first paid pilot, probably May of 2018. And yeah, and that was like, you know, we had one paying customer. It was really important for us to make our pilots paid, you know, so that we could prove there was a problem worth paying for, so to speak. And then after that, we had actually luckily got into TechCrunch Disrupt. They have this battlefield competition, which is probably the most prestigious, largest pitch competition on the planet. And so, we ended up launching at TechCrunch Disrupt in September of 2018. And so, between that time when we got our first paid pilot to launching and Disrupt was another three months, and our whole goal was just to get at least five paying customers or paid pilots before we launched on stage. So, that was like this crazy period of focus, building the technology, and then getting ready to launch and making sure we had at least some early customer traction. This show is brought to you by Frontlines Media, a podcast production studio that helps B2B founders launch, manage, and grow their own podcast. Now, if you're a founder, you may be thinking, I don't have time to host a podcast. I've got a company to build. Well, that's exactly what we built our service to do. You show up and host, and we handle literally everything else. To set up a call to discuss launching your own podcast, visit frontlines.io/podcast. Now, back to today's episode. What was it like in the build up to that moment? You're getting ready to go on stage. What was that like? Oh, man, it was so, so fun. I mean, the month leading up, I think that period, that three month period, was one of the most fun periods in startup life. Again, because once we found out we were getting into tech launch disrupts, it was cool, but I really wanted to come back to first principles. I was like, "Pin, it doesn't matter." At this point, it was a couple of us and maybe one engineer. It doesn't matter if we're going to go on stage. We know there's going to be a lot of hype. We know there's going to be a lot of vanity. What really matters is that customer obsession. I almost joked. I was like, "We're not getting up on stage unless we have five customer logos." It was that focus. We were like, "Okay, we're going to find five customers, five paid pilots. Before we go on stage." It was intense and so much fun. But then, yeah, the day came. We had practiced our pitch and it was this vision of what is now considered gen AI applied to customer service and really any knowledge work. Yeah, and I had practiced. I'd said that pitch like 30 plus times in the mirror. My co-founder and I had asked each other every single question that we think the judges would ask us about the business. What's your defensibility? What's your go-to-market strategy? Everything under the sun, and we were just locked in. We were dialed in and my favorite slide was that logo slide. I think six actually got to six paying customers by the time we launched. We were able to tell this story. We were able to answer the FAQ questions and it was just an amazing, amazing launch. Talked us about the marketing strategy today and what the marketing strategy looks like. Yeah, so it's evolved over the years. In the early days, it was super scrappy, I would say. It was very much like we would literally cold call customer support leaders and just be like, "Hey, we have this thing. Do you want to give it a try? Do you want to book a meeting?" That was literally our marketing strategy. It wasn't very sophisticated, but it works. If you call 100 people and five or spawn and one becomes a customer, that's a win. That was that in the very early days, but over time, we've gotten more sophisticated. We now have a strong content strategy. We're in a very, I would say, noisy market, right? Even pre-GPT, the understanding of how does AI even work for customer service? Why is that different than decision tree-based chatbots? What do you need to do in order to get the most bang for your buck? The point I'm trying to make is that we're in a very crowded market. We're also in a market that requires some re-education because there's so much noise. We actually spend a lot of our time trying to meet our customers and trying to take AI, which is in general this really complicated, convoluted concept, and really bring it to the point where it's actionable. That's enabled us to get this really strong brand where customers trust us to kind of bring them the future now. I mean, it's in the name for thought. But bring the future to now and make it in a way that's actionable for people. So our marketing has been really around that kind of customer education, which has really paid dividends because now I think we're seen as a really, really strong brand and one of the leaders, if not the leader in our space. You kind of touched on it there, but I think anyone who's building right now knows there's a huge amount of buzz with AI. Everyone is talking about AI. How do you separate yourself from all of that noise? How do you make it clear that your AI is better than everyone else's that's out there? Yeah, I think there are two things. So going back to that point about just customer education, one of the things I spent a lot of time doing is really just getting to the nuts and bolts of what makes one AI better than another. And the best articulation I have for our space, AI for customer support, is that 99% of competitors are doing one of two things. They're either building this outdated decision tree, clunky rules-based bots that we've seen for decades and they don't work and we all know that. Or nowadays, we're seeing a bunch of competitors take GPT and wrap it. And they're just basically building GPT wrappers on top of your knowledge base. And they're doing a technique called rag or retrieval augmented generation. And what's funny is that everyone thinks this is all that AI can do. So it's like really hard to explain like, no, that's only the first step. But what we've been able to explain is, well, let me use an example. Let's say you go and you're in support and you ask a support person or a bot, the question, hey, can you help me reset my password? Simple thing we've all had to do. With one of these rag bots, what they'll do is they'll take GPT, they'll look at the knowledge base and then they'll return, hey, here are the five steps resetting your password. Good luck. And that usually can resolve or even deflect 10 or 20% of issues. But with four thought, we actually on average see a 60 to 70% deflection rate with our customers or resolution rate by the bot or by the agent. And the reason for that is when somebody asks us, hey, I need to reset my password, our AI will do the following. It will say, hey, I first need to verify it to you by your email address. What's your email address? Great. I've sent off the password verification or password reset email. Did you receive it? Yes, I did. Cool. All right. Great. Your password has been reset. Thank you. Carry on. That's a very big difference. You have AI that can actually take action, make plans, follow up and resolve issues end to end rather than just giving you FAQs, and that's the difference between 20% and 70% deflection. What about your market category? How do you think about the market category you're in or maybe a better way to ask that? You know, what's the line item that customers are buying today? Yeah. So ultimately, four thought is generative AI agent, and it can do three things. One, it can resolve customer issues. As I mentioned earlier, it can detect generative insights. So tell you things about your support or grieving your product that you didn't know. And then it can also assist human agents. And between all of these, starting with the first example, which is resolving issues, the whole goal there is two things. One, reduce customer support costs, and two, improve the customer experience. And so those translate directly into metrics like your margins or your cost savings, as well as retention. And so, for example, if you can have an AI that can resolve 50, 60, 70% of your issues without ever having to go to a human agent, then think about the total cost reduction, but also for your customers, how much better that is if they're getting their answers immediately. And then that frees up your human agents to do more interesting things like figuring out, hey, where can I drive upsells? Where can I drive more value to become a strategic partner to my customers? Where can I bring insights back to the product team and things like that? So ultimately, this is making the entire business more efficient, and that's really the line item where it hits. This show is brought to you by the Global Talent Co. A marketing leader's best friend in these times of budget cuts and efficient growth. We help marketing leaders find, hire, vet, and manage amazing marketing talent for 50 to 70% less than their US and European counterparts. To book a free consultation, visit globaltalent.co. When we think about the future, what does the future of customer support look like? So not forethought and what the company's going to look like, but just customer support in general. Five years from now, let's imagine from the consumer's eyes, what does it look like? Yeah, I think in so many ways, it is pretty intertwined with AI. When you look around at Gen AI, Gen AI for customer support is going to be the single largest software category on the planet. And so this is really the beachhead market where Gen AI is going to make the most radical impact. And so in five or 10 years, I think a lot is going to change. I think there's going to be a lot more AI and automation. But again, in a good way, not the chatbots that piss everybody off. It's generative AI that is human, you know, is empathetic and help resolve issues and then escalate to agents at the right time. So one, you're just going to see a lot more AI and automation. And it's actually going to be better for everyone involved, not worse. The second is the role of that customer service agent is going to evolve. You're going to see a lot more customer service agents who are product specialists, who are adjacent to the sales team working on upsells, cross sells, or who are adjacent to the product team working on product insights. And so you're just going to get a lot more intelligence out of your customer support because it really is the beachhead market or it's like, you know, the tip of the spear when it comes to knowing what your product is actually doing in the wild. So I think you're going to see all that. And then my last big prediction is that you're going to see all of this start to meld together. There's not really going to be a concept of human agents and AI agents. There's not really going to be a concept of a help desk with a ticketing system and your contact center or phone. It's all going to meld together because AI is going to be in so many ways glue that brings it all together. So when you're talking to support, you might start off talking to an AI and then seamlessly that will switch in and become a human agent. And that'll happen in multimodal, whether it's chat, email, phone, over video. All of these things are going to be super possible. So I just think this industry is going to evolve rapidly and it's going to be really, really impactful for many people. What are you most excited about that's coming in the, let's say next six to 12 months? What are you building that you're just really excited about? Yeah. So one of the things we're building is a fully agentic framework. So I mentioned earlier that a lot of AI customer support folks are really focused on reg. So just like search, retrieval augmented generation, answering FAQs and QA. We're building out an engine that we call auto flows that enables the AI to actually operate like a human agent. It gives it the ability to reason, gives it the ability to take actions and hit APIs without even having to be explicitly taught how to do that and gives it the ability to ask follow-up questions and things like that. For example, if you're like, hey, I need to update my insurance policy. Maybe the answer or the way to do that is different depending on what state you're in. And a dumb chat bot will just give you a stupid answer. But an AI agent should be able to say, hmm, I need to know what state you're in. So I'm going to ask you, I'm going to follow up. I'm going to look up based on the state, okay, this is the right way to do it and then help you update your insurance policy. And heck, I'll even look into the possible policies. So this auto flows framework is something we're super excited about because it's almost like this general purpose computer, so to speak, or an AI that can really take on actions and do tasks just like any human would. And I think that's really the future of support. As I mentioned there in the intro, you've raised 92 million to date. What have you learned about fundraising throughout this journey? Yeah, at least in the early stage. And I think this holds true at every stage of the journey. Fundraising is storytelling. And it's very easy to get lost in the trap of the Guy Kawasaki 10 slides. Like I remember the early days, I googled how to write a funding pitch deck. And you always get this link of here are the 10 slides you need. You need your problem solution, this, that, and look, every pitch will have those elements. But if you're starting from building a deck, you're actually doing it wrong. The real way to start a fundraise is to really get clear on what is the story. And a story is a sequence of interactions and engagements and a vision you're trying to pitch and paint that is received by another human. It has an emotional effect, both an intellectual and an emotional effect on the recipient. In this case, the investor. You know your pitch is right when your investor has this rising energy and they're asking questions and you answer the question by the next plot point in your story. And by the end of it, they're like, this is amazing. I need to invest now. Like that is when you know you can feel it in your bones when you've told the right story. And so I think there's an art to it. It's very similar to comedy, right? Like a comedian knows exactly which line will get the most laughs. And they've said it so many times. It sounds natural. It feels conversational. But they know the audience down to the T and that's their set, right? And I think that the same thing has to happen with investment fundraisers. You need to know that, hey, I'm going to say this line and they're next. They're going to think about traction. That's the question they're going to ask me. And by the way, my next line is all about traction and this and that. And so anyway, I think there's actually a real art to storytelling. And there's a real art to VC fundraising. Final question for you, let's imagine that you're having lunch or having a coffee with a AI builder, an AI founder, based on everything that you've learned since founding the company in 2017, what's your number one piece of go-to-market advice for them? Dang, go-to-market advice. I would say the phrase that comes to mind is nail it, then scale it. And so what I mean by that is you should be spending, again, as much time as humanly possible figuring out both your product, your offering, but also your go-to-market motion. Everything we just talked about from, hey, cold calling or selling or how to do marketing or whether you have a partnerships motion. Many founders, especially technical founders, if you started with a product vision and you've never sold before, you think, oh, I need to outsource this. I need somebody else to figure this out. I need to go hire somebody to do this. It turns out that the founders are often the very best people to figure this out because not only are they passionate about the product, but they know the story. They know what people need and they're going to iterate on that. And so I think the first thing is build it yourself, do it yourself, sell the first couple deals, sell the first million dollars of ARR if you have to, nail it, build the playbook, write it all down. And that's when you think about scaling it. That's when you think about bringing, for example, in other sales reps or salesmen or VP of sales, things like that. I think that'll work really well. Amazing. I love it. All right, man, we're up on time. So we're going to wrap here. Before we do, if there's any founders that are listening in and they just want to fall along with your journey, where should they go? Yeah. So if you are interested in trying out for thought, just go to our website, forthought.ai. If you want to follow me, follow me on LinkedIn. I post pretty regularly always thinking about really cool things, whether it's AI or startups. And so yeah, I'm Dion Nicholas on LinkedIn. Amazing. Thanks so much. I really appreciate it. All right. Thanks, Brad. Cheers. This episode of Category Visionaries is brought to you by Frontlines Media, Silicon Valley's leading podcast production studio. If you're a B2B founder looking for help launching and growing your own podcast, visit frontlines.io/podcast. And for the latest episodes, search for Category Visionaries on your podcast platform of choice. Thanks for listening, and we'll catch you on the next episode. [MUSIC]
