Furqan Rydhan

Build an AI 'signup agent' that researches every lead and sends a tailored email

Furqan Rydhan describes Third Web's signup agent: for every new signup it decides if the person is interesting, researches their website and company, matches them to relevant products, then writes and sends a personalized email. They've deployed 8-10 such agents across the company.

“And so we built an agent to do this and every signup that comes in, it looks at it, it determines is it an interesting person or not? It will research their website, it'll research them, it'll then use the knowledge it has of Third Web products and try to figure out what it actually like, what products they might need, how they would use it, and then it would send them an email or an upsell or something like that. And we've deployed probably 8 or 10 of these throughout Third Web to do a lot of different functions.”

Steal thisPoint an AI agent at your signup feed: have it research each new lead, match them to a product, and draft a tailored outreach email automatically.

The 'clear directive + digital task' test for what AI agents can do today

Furqan's rule for spotting agent-ready work: if a task has a clear directive (go do X, with defined steps) and is fully digital, it can be done by an AI agent now. Anything fuzzy or physical is not yet a fit.

“And what I think about clear directive is like, let's go do X. Like sign up comes in, go look at the person, go research them, go figure out what's their title, what's their company, what products do they have. So that's like kind of clear what the job to do is. And if it's a digital task and a clear directive, All of it can be done with agents now. The technology is here, it's ready and it's working.”

Steal thisAudit your workflows for 'clear directive + fully digital' tasks; those are the ones to hand to an AI agent first.

Build a fully autonomous AI 'company': a CEO that can market and make money

Asked what he'd build as a 21-year-old today, Furqan skips the obvious AI influencer and pitches the digital equivalent of a company: an AI agent with its own Twitter distribution and a bank account that can market itself and generate revenue, like an AI dropshipper or Amazon FBA seller.

“You know, AI influencer is like the first obvious thought it goes to. I think, uh, if I played in this space, I would be creating the digital equivalent of a company, which is a CEO of a thing, the ability for it to market and the ability for it to make money.”

Steal thisGive an AI agent a Twitter account for distribution and a wallet for transactions, then point it at a clear money-making directive like dropshipping.

The one-person billion-dollar company is coming as headcount drops to 1-3

Furqan predicts AI will drive another 10x reduction in how many people a company needs, echoing Sam Altman's one-person billion-dollar company idea, and argues team sizes are heading toward 1-3 people, fundamentally changing how companies are built.

“I think Sam Altman said this, which is like the one-person billion-dollar company. Like this is happening. Like we're already experiencing another 10x decrease in how many people you need and the abilities that they have. And I think it goes down to probably 1 or 2 or 3 and You know, like that is a, you know, total shift to everything, right?”

Gorilla Tag, a VR social game, has done $200M in revenue

Furqan cites the VR social multiplayer game Gorilla Tag as proof VR is a 'sleeping giant': it did tens of millions on App Lab and has reached roughly $200 million in revenue.

“I think Gorilla Tag has done $200 million of revenue, and this is where I'm like, VR is a sleeping giant.”

Emerging tech has one rule: survive until the industry arrives

Furqan's thesis for betting on emerging tech (VR, crypto, AI): the only job is to survive. If you're a small share of an industry that 100x's, you 100x with it, and you've already built the expertise nobody else has.

“In these emerging tech things. So everything I found out that we do is emerging tech. I think the theme of all of them is survive. If you make it to when the industry happens, you will grow with it. If you were a small percentage of the industry and the industry grows by like 100x, you grew by 100x or more. You've already been there.”

Steal thisIn an emerging field, optimize for staying in the game cheaply rather than winning fast; survive until the wave arrives and you grow with it.

Watch the builders, not the narrators: out-of-fashion + shipping code = buy signal

Furqan's edge as a technical investor: he lives in GitHub and with builders, not narrative-shapers. When everyone publicly says a space is dead but the builders keep shipping code, that contradiction is the signal to lean in.

“If I see that somewhere where it's like, huh, all the people talking are like against it. These people, like nobody told them yet that like it's dumb. Like, oh, this is dead, right? Like they're all saying it. And then these guys are just shipping more code. I'm like, hey, did anybody tell you it's over? Like, what are you talking about? I don't care. Like, and I just think that's the perfect environment.”

Steal thisFind spaces the narrative calls dead where builders are still actively shipping; that gap between sentiment and activity is where the opportunity hides.

What A+ execution looks like: think, decide, act with no delay

Furqan on AppLovin co-founder Adam Foroughi: in an 8-person team, decisions were delivered to the team within minutes, whether starting something, changing roles, or killing an idea. The delay between decide and act is where most teams lose; Adam removed it.

“Like, if we were talking about something, we made a choice within, like, it felt like within minutes that was like delivered to this team. And look, when you're like an 8-person team, it's really easy to do it. People decide something, go slow, maybe next week. Oh, we'll do it later. We'll make these role changes later. We'll tell everybody later. No, it was like immediate.”

Steal thisShrink the gap between deciding and acting to near-zero; the delay on 'act', not the decision itself, is what kills most teams.

An AI teddy bear prototyped for $50-100K instead of $500K-$1M

Furqan says Magical Toys shipped 60-70 units of its AI teddy bear over 4-6 months. In the old hardware world that prototyping run would have cost half a million to a million dollars; with Raspberry Pis, 3D printers, and cloud AI they did it for $50,000-$100,000.

“And like, I think he shipped like 60 to 70 units across 4 to 6 months, which in software world just feels like, wow, ancient, like 60 users. In hardware, this might have cost like half a million to $1 million. I think we did that for like $50,000 to $100,000.”

Sell automation tools to small service businesses, don't try to replace them

Furqan's lesson from Lucid Drones (a building power-washing drone): the winning model isn't replacing the workers, it's selling the tool to existing small operators like 'Dave's Powerwashing'. They serve more buildings more efficiently, and their distribution is already built in.

“most people think you're going to build this product and then you're going to take the people that like manually wash a building and you're going to like get rid of them. What actually happens is there's like a team that's like a little, like, small business somewhere. They have like 5 people at their company.”

Steal thisSell your automation as a force multiplier to existing small-business operators instead of trying to disrupt them; their customer base is your built-in distribution.

AppLovin's failed $1.4B sale became the best thing that ever happened to it

Furqan Rydhan recounts how AppLovin's agreed acquisition by a Chinese acquirer collapsed after the 2016 election blocked foreign approval. The deal converted into a non-voting investment, employees still got over $1 billion in liquidity, and the company kept compounding all the way to a ~$30B IPO.

“And so it looked like the deal was not going to happen. The company had multiplied in value during that time. And so now there's all this leverage on the company side to be like, well, we don't actually need this sale, right? We're a lot bigger than what that value was. And I think they turned it into an investment instead. With non-voting or some kind of distinction there. And we ended up getting some liquidity then, which was fantastic. It was over $1 billion, felt great.”

$4-5M of founder money and 9 months on a dead Pinterest clone before the pivot

Furqan recounts AppLovin's three founders putting in ~$4-5M of their own money, which bought years of runway. Their first project, Style Page, was a Pinterest-style fashion site they killed after 9 months because, as he put it, they were five guys in Palo Alto with no fashion sense.

“But them three put in the initial funding. I think it was like around $4 or $5 million, if I remember correctly. But it was a lot in the sense that it gave us many years of runway to work on whatever We started with one project called Style Page. We did that for like 9 months. That was kind of like Pinterest. We really quickly realized we're like, you know, 5 dudes in like Palo Alto with no fashion sense trying to build a fashion website. This is not going to work very well.”

3-4 flow states a week is all a builder needs to produce a lot

Furqan argues creative and engineering work doesn't reward every-minute efficiency; it comes in bursts of flow. Hitting just 3-4 real flow states in a week is enough to be highly productive, which is why he optimizes for impact rather than hours logged.

“But I know that creative work is like, it happens in kind of these moments of flow. If you get 2 or 3 good flow states a week, I feel like you produce a lot. And I think most developers or creative people or builders, they'll feel this. Not every hour is super efficient, but there's these sometimes where maybe it's because you did all this stuff to build up to it or you got the right conditions. But if you hit like 3 or 4 of those a week, man, you're really humming, you're producing a lot and a lot of impact's happening.”

Steal thisStop measuring builder output by hours; engineer the conditions for 3-4 deep flow states a week and let the rest of the schedule be loose.

Commodity hardware + ML lets software devs build hardware companies

Furqan's core thesis: cheap off-the-shelf boards like a ~$30 Raspberry Pi mean a developer no longer has to design custom computer boards for a device. Pairing commodity hardware with machine learning collapses what used to be half a hardware company's work.

“Yeah, it's like a $30 computer. It's like a little board. And it's important because now a developer previously, they would have had to design their own little small computer board to go build a small— like, if you wanted to build a Roomba, you'd have to go design a little computer to put in the Roomba. Then designed the Roomba. Now you can kind of go pick up this like computer that's just a little board and you can put it into anything. So like, you know, it took half your company away of what you needed to do to make this happen.”

Steal thisBuild hardware products on commodity boards plus ML instead of custom electronics; the off-the-shelf board removes half the work and lets a software team ship hardware.

FitCam: a $30 box that turns your TV into a Mirror competitor

Furqan pitches plugging a cheap GPU board with a webcam into the TV everyone already owns, using pose estimation to turn it into a fitness game. He'd call it FitCam, undercutting Mirror's ~$2,000 wall-mounted hardware by using a screen people already have.

“My idea, if I wanted to do this as an idea, I would do like the best fitness product would be you buy this little device, it plugs into your TV because everybody has a TV on their wall. Now your TV is this awesome fitness game and you could just do like everything. I think I wanted to call it like FitCam or something like that.”

Steal thisInstead of selling expensive smart-mirror hardware, ship a cheap camera dongle that turns the TV people already own into the fitness screen.

The hype-cycle dip is when the real building (and value) begins

Furqan's lens on every emerging tech: a massive hype wave, then a crash, and the post-hype stopping point is where real, valuable products actually get built. He applies it to VR, crypto, hardware, and ML, and says DeFi is at that build-the-real-thing stage now.

“And my entire view of everything is there is these massive hype cycles, then these hype cycles go away, which is obviously what happens, right? When kind of this massive hype happens, then there's a stopping point there where the real building begins and the real value would be created. And all of these kind of industries flow like that. VR, crypto, these kind of hardware spaces, machine learning back in the day.”

Steal thisWait for the hype to die before building; the post-crash trough is when real products win and the noise has cleared.

Stablecoin liquidity pools paid 8-15% (and early Uniswap pools topped 100%)

Furqan explains why he provides liquidity in aligned pairs to avoid impermanent loss, then cites the yields: a USDC/DAI stablecoin pool earning 8-15% for over a year, with early Uniswap pools on good assets exceeding 100%.

“So like two stablecoin pools, a USDC and DAI pool that does anywhere from like 8% to 15%, and that's been like that on the low end of the spectrum for the last year plus.”

Furqan's hardware lab: ~$500K/year just for the building, ~$1M with people

Furqan is building a hardware and robotics lab (Founders Inc.) at Fort Mason in San Francisco. He estimates roughly half a million dollars a year just to keep the base building, likely double that once he funds the people and projects inside.

“It's going to be a lot. It'll probably cost about half a million bucks a year just to even have the base building in place. And my guess is it'll be double that just from the people that I want to bring in and kind of investments I want to do and what I want to support.”

Build the biggest farm in America by going city to city, Uber-style

Furqan reframes vertical farming as a direct-to-consumer land grab: a vertical farm is a machine that produces fruit and lettuce, so instead of buying farmland you put a small farm near each city and expand city by city like Uber to become the biggest farm in America.

“Well, if you can produce the fruit yourself, like you just have a machine that can produce fruit, that's the vertical farm. Well, why don't you become the biggest farm in America really quickly by kind of going city to city and kind of doing that? And you've seen Uber do that. You've seen kind of all of these other, right? You could literally invent, kind of build the biggest farm in America now, right?”

Steal thisTreat a vertical farm as a deployable machine and roll it out city by city, building a national D2C produce brand without owning farmland.

Companies die in the market, not the tech — even for a technologist

Furqan's most-repeated advice to young builders: most companies die in the market, not the technology. Even as a deep technologist he tells founders not to over-invest in tech, because someone else will take your market with the same tech if you're not aggressive on growth and customers.

“And if you're, you know, really most of these companies will kind of die in the market, not the tech. And there's some challenges sometimes, but even as a technology person, I would not spend too much time on technology.”

Steal thisSpend your energy winning the market and understanding customers, not perfecting the tech; treat technology as a weapon for growth, not the product itself.

Shopify for NFTs: pay the gas, two clicks to mint and sell

Furqan notes the dollar value at the top of NFTs is huge but only ~5,000 users actually mint because it's too hard. His NFT Labs project removes that friction: pay users' gas, deploy their contract for them, and give a landing page so minting is as easy as opening a Shopify store.

“And I believe that's one of the biggest opportunities in NFTs is just like, make it really easy for people to do it, almost as easy as creating a Shopify store. So that's like project idea one. We're going to basically pay for your gas. We're going to make it so you can deploy your own contract without ever thinking about it. You're going to just click a couple of buttons and get a landing page that you can send to people and they can buy stuff.”

Steal thisFind a category with huge dollars at the top but tiny user counts because of friction, then build the Shopify-easy on-ramp that abstracts away the hard parts.

He took the lower salary, moved in with his parents, and chose the messier startup

Furqan had a polished job offer with an early Google infra leader, but picked AppLovin's chaos instead. Adam offered higher-salary/low-stock or lower-salary/high-stock; Furqan took the low salary, moved back in with his parents, and bet on equity because it would be more fun and a higher rate of learning.

“And Adam gave me two options. He gave me a higher salary and a low stock. A lower salary and a higher stock option. And I was like, I kind of just made the decision down to like, well, if I go with Adam, I need to take the low salary, move back in with my parents, take the stock because I'm going to be in it for the ride.”

17-year-old turns $800 of car-mod parts into $1M e-commerce store

Furqan Rydhan, bored after a dot-com flameout, noticed PC enthusiasts wanted to 'trick out' their computers but had nowhere to buy parts in the US. He bootstrapped an accidental e-commerce store from $800 of inventory in his garage corner and hit $1M revenue in year one, at age 17.

“So I started an e-commerce store kind of accidentally. I was like, oh, like, these things are really expensive. If I buy them in bulk, I just get them cheaper. I'll sell off some, I'll have some more cash on the side. And I think we started, uh, $800 of stuff in the corner of my garage, and then within the first year we did $1 million of revenue.”

Steal thisFind a passionate niche community that has no good place to buy the gear it loves, then become that store.

17-year-old turns $800 of car-mod parts into $1M e-commerce store

Furqan Rydhan, bored after a dot-com flameout, noticed PC enthusiasts wanted to 'trick out' their computers but had nowhere to buy parts in the US. He bootstrapped an accidental e-commerce store from $800 of inventory in his garage corner and hit $1M revenue in year one, at age 17.

“So I started an e-commerce store kind of accidentally. I was like, oh, like, these things are really expensive. If I buy them in bulk, I just get them cheaper. I'll sell off some, I'll have some more cash on the side. And I think we started, uh, $800 of stuff in the corner of my garage, and then within the first year we did $1 million of revenue.”

Steal thisFind a passionate niche community that has no good place to buy the gear it loves, then become that store.

AppLovin's first liquidity event: over $1.4 billion

Furqan Rydhan, co-founder/CTO at AppLovin, notes the company had a big liquidity event a couple years prior valued at over $1.4 billion, on roughly $5M total funding.

“a couple of years back they had a big liquidity event, over $1.4 billion. So, and I think they've probably grown in multiple since then.”

AppLovin: from launch to $10K+ in revenue per DAY in the first month

Furqan Rydhan describes AppLovin's product-market fit: in the first month the ad product jumped to over $10,000 in revenue a day, with servers getting hammered as money flowed in.

“But the first, first month it was like, man, we just jump up to like $10,000+ in revenue a day. And then you just kind of keep skyrocketing up from that.”

Real-life Mario Kart: AR go-karting where you shoot weapons at rivals

Furqan Rydhan pitches an AR go-karting venue: longer track, goggles overhead, drive over digital boxes to pick up weapons, and shoot opponents to briefly shut down their cart. A location-based attraction like Sandbox VR or modern laser tag.

“I was like, dude, it would be cool if we could, you know, go to a go-karting place, but it's like Mario Kart instead, where you have these AR goggles over your head, and it's not about quick turns, and it's like a longer track where you can kind of like have some space. But I, you know, would digitally go over this box, and I now have a weapon that I could shoot at your car, and if I hit you successfully, your car kind of shuts down for a second and you're stuck. So like the real-life Mario Kart.”

Steal thisLayer AR game mechanics onto an existing physical activity (go-karting) to create a location-based attraction.

Always-on AirPods walkie-talkie for your inner circle

Shaan and Furqan riff on an AirPods app that creates an always-on, opt-in audio channel to a few close people. You speak a wake word and the other person hears you instantly, capturing a fleeting idea without picking up a phone or typing.

“So let's say I have my AirPods in, you have yours in, and I'm walking around, I'm at lunch, I have this idea for the podcast. I basically just say, yo, Sam, what we should do is we should do this. And then you basically receive But you're not sending back, um, unless you say whatever the keyword, like the, you know, like how Alexa works, right? Like it's like Sean or whatever it is. And that's way more lightweight than me picking up my phone and starting to type or phone call. Like who wants to do that? Right.”

Steal thisWhen new hardware ships (AirPods), ask what lightweight always-on software it now makes possible.

Lucid Drone: sell the incumbent an efficiency tool, don't try to replace them

Window-washing drone startup Lucid couldn't sell building owners directly, so they sold the existing window-washing service companies instead: lease our drone, do 5x more jobs per crew. They made the incumbent's business more valuable rather than disrupting it.

“So they went to the guys who were servicing the building. They're like, hey, I can make your business more efficient. Instead of sending 5 guys out per job, you could do 5 more jobs. Instead of renting this crazy scaffold, you just lease our drone. And it's like you make their business more valuable. You're not taking their business and then you don't have to go do all the building to building sales.”

Steal thisSell your tech to the incumbent service provider as an efficiency multiplier instead of competing with them.

Hard research is done: deep tech now just needs application engineers

Furqan argues drone flight and computer vision are now 'trivial' because the research work is done. You no longer need a CS PhD to build with them; the opportunity is for application engineers who combine these easier technologies into specific products.

“I think it's at the spot now where it needs application engineers and not that same, you know, you don't need to be a PhD with computer science and all these math degrees to go create this technology. You know, you take the work that Dave built and they've kind of established and then apply it to these things. And so drone flying, a drone is trivial now. Building a drone is trivial now., and then the software is easy to program a drone to fly. Object recognition and computer vision is easy now.”

Steal thisWait for hard research to commoditize, then be the application engineer who ships a specific product on top.

What product-market fit feels like: you lose control to the market

Furqan describes real PMF as binary, not a gradient. When it's truly working you stop trying to spot signals; it's so obvious the charts just fly up, like slamming the brakes and feeling the car take over. The market pulls so hard you're just catching up.

“And then when it's really working, it's not about spotting it. It's so obvious that there's no question that it's working. And so it's all the charts are flying up.”

He left a grand-slam startup after 3 years, not fully vested, to build games

Shaan flags Furqan as counterintuitive: most people ride a winner for 7 years and take the victory lap, but Furqan left AppLovin after ~3 years, leaving money on the table, because he wanted to build consumer products, not an ad platform.

“Cuz who would leave this company that's crushing it? Like most people, when you finally hit a grand slam, you wanna do the victory lap and people stay for 7 years and then they write this cheesy letter about how great the journey has been and they're finally ready for their next chapter. He basically left after 3 years. You weren't even fully vested, right?”

Mobile game LTV math: pay $8 for a user, make $20 in 7 days

Furqan recalls the money-printing economics he saw in ad tech: mobile game makers paid roughly $8 to acquire a US male 18-35, then made about $20 off them within 7 days, roughly 3x their money in a week.

“they're like, yeah, dude, we make like $20 on them, like, in like 7 days. And I'm like, and maybe those numbers are slightly off, but like, they were making a multiple to where you were like Wait, so you just have this machine where you just press this button and you put in money and 7 days later, 3 times more money come out?”

A+ execution: decide, then act now and announce it immediately

Furqan's biggest lesson from AppLovin CEO Adam: once a call is made (a hire, a deal, an org change), don't wait for a 'good time'. Adam acted now and communicated the change to the whole team the next morning with a new org chart. Speed creates clarity; logical, unemotional, fast.

“A lot of people kind of wait on it a little bit after deciding, and you kind of create some time. What's a good time to do it? For him, it was like, okay, we'll just go do this now. Like, what's stopping us?”

Steal thisOnce you've decided, execute and communicate it immediately instead of waiting for the right moment.