He Co-Founded a $2.5B company

Welcome to another Founder Wisdom podcast. We have Charlie Feng. He's the CEO and co-founder at Runner. Runner is an intriguing agency company we're gonna talk about in a second. He's also previously the co-founder of a company called Agora and most notably Clearco. I love Clearco. I've studied that company extensively and arguably I'm trying to build eventually my own founder bank. And Hemingly as well. You've founded a bunch of things, Charlie. So welcome to the podcast. Tell us a bit more about yourself and what you've been up to.

No, I appreciate that. Thanks for having me, Charles. Tried out a lot of things, like probably many other founders and entrepreneurs out there. Not all successful, but a lot of learnings along the way. So happy to share and kind of chat through any of those. Excited that you're thinking about the founder bank. That's kind of What was one of the visions for Clearco as well? It's like, hey, there's a lot of underserved founders that don't have access to capital yet. They're kind of bounded by the structure of just like getting underwritten on their FICO scores, which really doesn't make sense when you're running a business. How do you solve for that? So yeah.

Yeah, let's start historically, right? So are you from Montreal? I'm from Montreal. I saw HypeJar. That was like your first job, right? Or are you from Toronto?

I'm from Ottawa actually, so I should know more French than I do. And I went to college in Montreal at McGill University, and Clearco, like more business-wise, was based out of Toronto.

Interesting. How did you start out in life? Were you like always this entrepreneurial-minded? Were you a geek, a nerd, like just coding things? How were you as a kid?

Yeah, I think I was always a, I guess in hindsight, a geek and a nerd. I'd liked, I remember one of my first forays into building things were just something called RPG Maker. I don't know if you heard of that, but back in the day, I loved the little RPG games, the Pokémon games. And there was a tool back in 2000s, early, mid-2000s maybe, where you could make your own RPG games with a bit of Ruby and this coding. So I did that. I remember early days with my cousin and that was just a ton of fun. I honestly never thought about what a startup was or the entrepreneurship or building your own company. I didn't even think that was a path. As someone who grew up both in a more traditional household as well as the The town, like Ottawa, where I grew up, as well as the school I was in, were very STEM-based, and everyone was very much like, you're either gonna be a doctor or you're going to be a lawyer or you're gonna be an engineer. And the idea of you starting your own business was never on the table. So I kind of fell accidentally into that somewhere during college.

Right. And why take the founder path? That's the uncommon path. Your parents probably didn't approve that.

They did not. They did not. That was definitely the uncommon path. And I think now I really appreciate where the ecosystem has gone now in 2026, where both it's encouraged there, it's the tools and the barrier to get started as a founder is pretty low. You kind of just honestly with Claude, right? Build something, ship something into the App Store and, or just on a website and started getting customers. And I think that this path was a lot less popular 10, 15 years ago. And yeah, my parents wanted me to be an engineer or become a lawyer or something like that. And for me, like I mentioned, it was kind of accidental, kind of started my career. My first job was HypeJar. I was kind of just doing some frontend development for them. And then even when we started our first startup, Hemingly, that was because we entered a school competition, was lucky enough to kind of get selected. We went through a whole summer bootcamp in Boston, and that was kind of my first taste of like, oh, this is kind of a— this career path was even possible. Like, this was very interesting. And I think fast forward to today, What gets me really excited about the more founder and entrepreneurial career path is that I think it's that saying of you pick your poison, you pick your hard. Everything in life is going to be hard, whether you are building a— whether you are like a baker at a bakery or whether you are an IC7 at Google or whether you're a founder. I think there's just different types of hard. I don't think the amount of hours you put in, in fact, Arguably, like someone who's working in the restaurants, they know they're working much more longer hours than many founders. But kind of like you pick your heart. And I think for me, there's a lot of things gruesome and terrible about being a founder, but I really appreciate that at the very least, I get to pick who I work with. And that makes all the difference for me.

Right. Yeah, it depends with your upbringing. Let's say your father ran like a bread store, right? Like you have an advantage and maybe that reminds you of your childhood. Maybe you love doing it. But I mean, sort of biased on mine. I was also a video game kid and I grew up like listening to my dad running his business in the family office, right? And listening to him make some cold calls. I guess that's my quote unquote safe place. What I know subconsciously, right? That's easier. But I also think running digital business is so much more easier than pretty much everything there. And some people running, I don't know, D2C store might not even realize it, right? Because they've never experienced anything else, but just getting from something physical from point A to point B, right? Rather than just on a computer shipping from Charles to Charlie takes 1 second. I think that's incredibly inefficient. I'm not saying people shouldn't start these businesses and there's like a trend in the US to start like these first materials type of business and like more old school, but that's fake melancholia in my opinion. I mean, we need those, some of those, but yeah, how many of those, right? Sure, I'm not saying like should be a monopoly like Anduril and should have like everything built by itself. They need competition to make themselves better and not create like unfortunate circumstances for other competitors, but I think people— I don't think people should build physical shit. What do you think?

You don't think people should build physical? I think—

Except robotics. Except robotics. Like robotics, I mean, you can automate pretty much everything. I would start a robotics company, but it's also so exponentially hard, right? Especially like a humanoid robot in your In your home. But I mean, once you crack that, you crack like everything, right? Like you can have like construction companies with these workers building shit for you. And yeah, that's probably the only physical stuff I would get into.

Yeah, I don't know if I'm the right founder to build physical stuff, at least not right now, it feels like. But I am very appreciative of the people who are. in that space, and I'm pretty excited about all the robotics, all the materials. So my parents, maybe it didn't come as naturally for me because growing up my parents were both scientists, both chemists, and my dad would bring back devices to measure the quality of the air and things like that, but never was there a code call in the background or things that kind of rubbed off on me. But instead, kind of like a lot of the curiosity around both sciences as well as that physical atoms that you're touching, I think is at the end of the day, we are humans that lives in a physical world. So I think the atoms are very important. Now the question of depending on what stage you are through your founder journey or what your background is, what should you work on? For me, I totally agree with you. Software has been so much easier than when I've talked to my friends in the— you know what they say, hardware's hard, right? Because when I talk to my friends who are doing hardware, Just the cycle of iteration, I think it's incredibly challenging or it's a different type of learning loop. And I think one of the things that is really nice about being in software, at least in the last probably 5-year decade and probably going forward, especially with AI, is that the loop is just so quick in terms of how quickly you could iterate. And as a result, you could really compound your own learnings and you just get ahead, right? Yes.

Sold faster, and the loop is money, right? Like that. The reason I got started in business back then, I had like three jobs, and I just realized I could make fifty bucks an hour reselling shit I would buy on AliExpress instead of my ten dollars an hour would make at these three jobs, right? I was like, easy, easy patch, right? Like you need to close the serotonin loop, right? And my first company, by the way, I'm telling you this, Charles, because my first company was a nootropic, right? So. pill that you take for the brain and memory and all of that. And this is a passion project, so we had that fuel, right? Because we would take our own supplement and we would increase our own productivity. And I had other businesses so I could have an ROI even if I didn't sell supplements, right? But obviously that one got down to the ground because just building a supplement is what, like 6, 12 months of R&D, building the formula, having it approved by Health Canada and all of that. Doing research on it, and yeah, that fell to the ground. AI—I mean, you spent a lot of time. I mean, MIT, right? There there was a couple of AI godfathers in there. You spent time in Toronto. There was Jeff Hinton, Montreal, Joshua Bengio. So it seems like AI was always part of your life early. Unlike myself, you acted upon it, right? Like you started a company called Hemingway in 2014. Was that your first stint in? AI, how was— how did that one go? Your first one?

Uh, went, went horribly. Or wasn't, wasn't terrible, but it was— did not go anywhere. Uh, so, um, that was our first foray. And I think you're right that like it was always in the air in some ways, whether it was, you know, MIT Boston or Toronto or Montreal. Um, there were a lot of friends that were close friends of mine that much more successful than myself, who were kind of more deeper into the research back in the day, the ML side or kind of the reinforcement learning. And I think as someone who was more on the business side and less technical than a lot of them, I was always very fascinated and intrigued and trying to learn a lot of those areas. And I think Hemingway was a good example where we tried to We thought we were doing AI back in the day of 2012, but the technology was definitely not mature by any means. And we were trying to do something actually now just these last couple weeks, right? We had the ChatGPT with Dot, Facebook with Muse, right? And Google's probably launching their own version of these personal assistants. And we were trying to do basically some version of that. And it was supposed to be a an AI agent or an agent, an assistant that could help you do things. But well, it became much more of human routing and kind of fancy Python scripts than it was true, truly anything LLM-based. So the technology was just frankly not there, but always had a fascination for like, how can you actually automate these work that I think nobody is very excited to do? Nobody's excited to wake up to say, I want to do these rote manual work of sending emails and coordinating scheduling, but it has to get done. And it's what makes up bulk of the, I'd say, the white-collar job load these days. It's the admin work.

Right. Seems Claude cracked that part, right? Claude Code specifically.

That's right.

Are you like, sometimes, shit, I wish I would have stayed there? And why did they crack it and you didn't? They just had like a bunch of Charlies in their team? What would you actually— or were they at the right place at the right time, right? Is it just like a talent moat and all these engineers cracked the problem with just like energy and time while you were solo? Like, why did they crack it and you didn't?

Oh, I don't think we would have even been close. I don't think we were even in the right direction. So like, even if I stayed in AI, maybe I would have joined, I don't know, one of the early labs or something. I don't think I was technical enough to really put a dent in any of the— so I think I see myself from an entrepreneur or like a founder perspective as— I think as maybe taking a step back as a founder, it's important to know where are your strengths and how do you— where do you want to play the game? Kind of like in sports as well, right? There's different people who are maybe a baseline player or maybe you're serving volley. I'm thinking about tennis. I am not a hyper-technical— I like to nerd out on these things, but I am no PhD in ML, and that is not my forte. I'm more probably on the application side and more how do you— now that you have this technology, how do you apply it? What are the areas that could really leverage this technology? And I think that— I think the technology is really just the first part. And then I think now we're kind of entering this phase of work continues to improve the LLMs and technology, which credits to all the labs. And then I think there's going to be a whole swath of companies around the application layer of, well, how do you actually commercialize and properly put this technology into actually increasing the GDP or actually increasing the productivity of all these industries and companies? around the world, right? And I think right now, frankly, we haven't really seen that with AI. I think the technology is very incredible. Everyone can— you don't need to squint anymore. Everyone could just flat out see that like, oh my God, the Claude, or whether it's GPT, it's amazing. It could do a lot. But I think one thing that CFOs that I'm talking— I've been in talks with, For many companies are last year, it was very much of an era of spend as much money as you want, try all the tools. And now I think this year people are starting to kind of pay attention to, okay, we spent all this money, what's the actual ROI and margin expansion or the results we're getting to the bottom line? We're using all these tokens, is it actually increasing the productivity? And I think we're seeing a mixed bag. There's definitely productivity gains within the mix, but there's also a lot of areas where we're just kind of like wasting billions of tokens getting the latest Opus model and Fable model to read our emails, which is not really needed, right? So I think that is the area that I'm personally very excited about. And yeah, wish I was more of an AI researcher, but I'm not.

But you had the gigantic home run after that, which is Clearbank, right? I mean, that was literally your second company, dude. That's like a very good home run rate. How did you find out about Michel and Andrew, both crazy good entrepreneurs? And I guess you're a great entrepreneur yourself, right? Like, how did you find out about these guys? And tell us about that crazy story from Masa Son investing $215 million to the $2.5 billion valuation and the sort of crash and burn, right? Like, tell us a bit more about that.

Yeah, yeah. And honestly, a lot of credits to my co-founders, Andrew, Michelle, Tanay, and Ivan there. And I think it's one of those stories where, and I think this is kind of something you almost forget as you start new companies, and then I have to like, you know, pinch myself a little bit to remind myself that The early days of product market fit is quite humbling, no matter how much experience you have, and you have to kind of just talk to the customers, really understand where are the pain points. But ClearBank, now it's called Clearco, but as the previous name, you can probably trace back, it's actually called ClearBank, where the vision by, well, mostly credit to my co-founder Andrew, who's now building Bordi. He— it was very much, can we build kind of a bank for founders and entrepreneurs out there? Because there was no dedicated bank for folks. I think there was Silicon Valley Bank back then, but that was a very different type of capital for folks who's raised venture, specifically venture. And then there was a whole swath of companies where we started to specialize in around e-commerce, for example, where these D2C founders are building incredible businesses, but nobody wants to fund them. And we were also kind of entering the stage of asset-light businesses, AKA these digital businesses, right? Because before, back in the day, you would have like, you would have a storefront with a warehouse, you would have chairs and ovens, and then all the type of lending that the previous generation of financial infrastructure was built around is basically asset-backed lending where you look at the asset, you take a percentage of that and you say, hey, I'm, you know, if all goes to shits, at the very least I'm going to liquidate your ovens and make some money and recoup some money back. Which was in hindsight, I mean, in some ways kind of silly because while it makes sense on paper, are you really going to liquidate the baker's ovens, right? That kind of thing. And I think we were entering also an era of a lot of software businesses and what we kind of called or coined kind of the asset-light businesses where they don't have that much asset, but they have revenue cash flow and they have what really matters, which is a strong customer list of recurring demand for whatever product that they're trying to sell. And that was overlooked in our eyes. So we basically built a whole new underwriting system around Their Shopify data, their Stripe data, their Facebook ads data, all the data that kind of is more that A, you could get it from APIs. So it's very easy to just ingest. And B, they're actually very granular pieces of information that you could get a lot more depth into the business than simply just a 12-month financial, right? Where there's good data there. But when you kind of get into the transaction data, you start to build like customer cohorts. You can build a lot of interesting things there. So we basically build our whole underwriting system on that as a premise. And yeah, I think probably everyone saw kind of like the heights of it in 2021. I think 2021, 2022 with kind of the interest rate, as most, if you're in the lending business, it's really tough when your fundamental cost of goods, right? Because if you think about it, our cost of goods effectively interest rates. We borrow money from other capital providers and then we do a set of underwriting and we give that money out. It's basically, we're not selling clothing t-shirts, we're selling capital. And well, the underlying cost of goods kind of spiked in 2021, 2022. And there's a whole other story maybe we'll save for another day there. But the short version of it is that it tripped up a bunch of kind of things and covenants that we, we in hindsight should have been more conservative around. But that's, that's, yeah, that's how the story went.

And what a crazy story. Yeah. How did, how did it end up for you? And how did you get the idea for what was next, which was Agora?

Yeah. It was kind of separate. I wouldn't say that they're very well correlated. I think that it was going through our own, my own founder journey of what do I want to do next? After our SoftBank round at Clearco with Masa and kind of the height of everything, I chose to walk away kind of at the height. It was more of a personal decision than it was anything kind of on the business at that point. I think I did— well, that was what, 2021? So I wish I was smart enough to say, hey, I saw the ZIRP era and all that thing crashing down, but I didn't have that foresight. So for me at that time, it was more so I really enjoyed the early stages of building a business and Clearco was more on the path towards IPO and kind of trying to figure out what the what that longer path looked like. And I was more interested in, well, what are some of the earlier stage problems that I could be tackling? So took some time off, took almost a year off actually, and then started Agora. That's eventually pivoted into what we're building today, which is Runner.

And how did Agora go? I believe that Governance was very trendy and important during these years? Did it explode? Why did you pivot to Runner?

Yeah, yeah. We were in the— so Agora, for those who don't know, we were building basically crypto compliance. We're building kind of voting compliance software for kind of on-chain protocols and communities. And the idea was that there was a lot of boom of, I mean, still is very much of the, whether it's Ethereum, Bitcoin, kind of like crypto protocols. And our view is that the only reason that these protocols can be durable and stand the test of time is with strong governance. And because traditionally you have companies that would actually— that traditional companies have governance, right? You have a board of directors, you have shareholders, you have shareholder voting and all of that. And all these things does not exist in kind of this on-chain world that was starting to get built out by the Ethereum and crypto community. And we thought, well, someone needs to build the picks and shovels or the tools to allow things like a shareholder vote or allow things like, who actually gets to control these protocols and how do you actually set the rules? So that's what we set out to build. So it was actually quite a great ride. We built a lot of very kind of strong primitives and foundational kind of software around that. And I think what we missed was a little bit of the timing. I think in hindsight, did we misread the timing of the market? And I would say that The governance side of protocols or crypto is just not quite— not that it's not needed, but it's not as high prioritized as other areas of the market. Like right now, if you're seeing kind of the crypto market, it's very big into stablecoins, very big into kind of like trading, and that's kind of where most of the value is being accrued. And yeah, we kind of Built the business, it was a great kind of ride up, and then we kind of tapped out on what the market could even look like. So hence we're like, okay, what's next?

And Runner was next. How would you describe Runner, right? And how do you use it yourself to get more revenue from your clients?

Yeah, Runner is very much of, I would say, a combination of A lot of the experiences I've had myself, both at Clearco, Hemingly, as well as Agora. My role was always a little bit of the operator, the operations of these businesses. I built a lot of internal tools in the early days, whether it be using other tools like Retool or whether it be using Google Apps Script to stitch things together. And then when you mentioned Claude Code, Claude, we were actually pretty early onto using Claude Code. And just using Claude Code specifically not for engineering tasks but for non-engineering tasks. And that was a little bit of the holy shit or unlock moment for us where it felt like, oh wow, you could just—

When did you get that? Like a couple months ago?

For the Claude Code?

The holy shit moment.

Oh no, this was like a year and a half ago. So like we were pretty early. Like we were, yeah, yeah. So These are still kind of things like early day Claude code. People are still using WinServe and Cursor back in that era. And we too were using these tools like everyone else. And sometime kind of like mid, early last year, as we were using these tools, we started— I'm a big Obsidian user, for example. So I was using some of these tools for my own triaging my emails or organizing my my writing and files. And I was like, oh wait, this is— you can start to— the things that you could do to coding, you could do to knowledge work. And we got very excited around this as a premise. And we still rely, Runner, on all the model companies. So we use a variety of different open source models as well as closed source models, frontier models. What we are very excited about is how can you solve enterprise workflows or use cases with agents? And we see that agents are able to take care of a lot of the mundane tasks, I would say, that then allows the humans to kind of focus on what really matters. So one example of this is that like we will spin up agents for, let's say, patients or maybe supply side for a marketplace. And the type of things we would be able to do is help with coordination, right? Things that would traditionally require kind of like an admin or require someone on maybe like a customer ops type team to do these kind of things. It's a lot of human time. It's a lot of emailing back and forth and double-checking numbers in the database. And it turns out AI is much better than humans at doing that. They're both faster, more reliable, And then I think a lot of teams that we— or the enterprises we work with, the human ends up spending their time talking to the customer, right? They spend the time on the phone actually working through the real problems rather than trying to schedule or coordinate a meeting. So those are the areas of application that we're very excited about when it comes to R.

Right. And last time I think you told me about like a taxi app, like a driver app, right? Like, where do you see The strongest applications, because you could see that in organizations with like a lot of customers, right? And you can think of so many businesses that neglect their customers, right? And you can think of them as sitting on gold mines right now. They're not like monetizing that. They're just neglecting that. And having you on day one could unlock tens of thousands, if not hundreds of thousands of dollars and eventually millions, right? Per day. Which type of organizations would be that? And I know you need to run Let's just finish off on that one.

Yeah, no, not at all. I think organizations where I do really believe the phase we're entering is a lot more of a higher degree of personalization, right? Before you would have a human follow a script, you kind of do some degree of segmentation. You kind of bucket people into different categories to say, okay, Charlie, you fall into this 30 to 40, whatever age group you are in, you're male, and then you would try to tailor their experience accordingly based on demographic or their user persona or things like that. I think now we're getting to the age where you could start to have one-to-one tailoring, what the type of experience you would get if you were treated as a true VIP of any given customer. And you don't do that today just because it's a lot of context, it's a lot of work. There's no way for a human to to hold thousands of Rolodex or contacts themselves without flipping through a whole CRM and doing all that, but very easy for agents to do so, right? You could give an agent a set of files and that's their one thing that they need to really focus on is making sure that customer is treated like gold, treated like a VIP. And some examples of this are like, I'm actually very excited about the space of like both financial services as well as healthcare. So examples in healthcare, is that you have patients that will show up at hospitals where they haven't got their blood works done and they're wasting everyone's time. They're wasting their own time because now they've made a trip and paid money for the Uber to get to the hospital. The doctor shows up and they're like, hey, I can't really do much for you. You got to go back and get these things that I prescribed to you last time. Wouldn't it be nice if you had an assistant that could actually— or an agent in this case— that could actually make sure that before they came in, the blood work is done, check with the, the, the EHR database and make sure that these things are all done and then do any rescheduling if needed. And then financial services. I think the whole financial service side is very similar where these are both cases where there's a lot of complexity in the system and therefore traditionally it's kind of hard for people to get very— it's kind of confusing for everyone. It's confusing for the company and the customer support or whomever who's doing this. It's confusing for the patient or the clients. And everyone kind of gets a slightly subpar experience. And I hope that really evolves as we kind of get more into this personalized agent world.

I can see a vision and I can see money as well. Where can people find out more about you, Charlie?

Yeah, I'm on Twitter and LinkedIn. It's Charlie C. Fang. And then for Runner, the website's runner.now.