Nov 26, 2024 · 1h 28m · my-first-million
$100B Founder Breaks Down The Biggest AI Business Opportunities For 2025
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Tech entrepreneur Furqan Rydhan joins Sean Puri to break down the massive leverage of autonomous AI agents, explore contrarian opportunities across VR, crypto, and hardware, and share startup strategies from Founders Inc.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 32.8% of the talking time here. How this is scored →
The hosts as informed peer 4.8 Guest teaching 5.3 Guest disagreement 1.4 The hosts pushing back 1.8
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 25:20 Furqan asserts machines will never replicate human taste
Furqan takes a firm philosophical stand that machines will perpetually lag behind human taste and genuine creative variance.
Hardest push from the hosts ▶ 25:40 Shaan pushes back with TikTok algorithm proofShaan rejects Furqan's premise that taste is inherently human by demonstrating how TikTok's algorithmic curation already surpasses human self-curation.
Biggest teaching moment ▶ 36:10 Furqan reveals VR indie gaming revenue tractionFurqan dismantles the prevailing narrative of VR as a dead end by detailing hundreds of millions in revenue generated by grassroots indie VR titles.
The host holds their own ▶ 1:08:05 Shaan diagnoses Blab's fatal project selection flawShaan provides an incisive retrospective analysis on why superior execution without proper project selection caused them to miss the multi-billion-dollar B2B video market.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Reconnecting and Reviving the 'Cool Shit Hour' | 4 | 6 | 1 | 2 | Shaan sets up the episode as a revival of their show-and-tell dynamic, inviting Furqan to break down AI agents. Furqan explains how autonomous reasoning loops operate at ThirdWeb and on personal calendar automation, while Shaan asks clarifying operational questions. | |
| Sponsor Segment: HubSpot AI Strategies Report | 3 | 5 | 1 | 1 | Following a sponsor read on AI strategies, Furqan details developer frameworks versus no-code workflow tools like Leap, showing how digital processes can be stitched together with simple intent prompts. | |
| Mapping the AI Landscape and Google's Big Fumble | 5 | 5 | 2 | 2 | Shaan and Furqan dissect Google's failure to capitalize on the transformer architecture it invented with 'Attention Is All You Need', noting how the key authors departed to launch rival companies. | |
| Anthropic's Computer Use and the Pace of Frontier Models | 3 | 6 | 1 | 1 | Furqan explains why Anthropic's Computer Use feature and Sonnet models represent a shift from cloud-only AI to direct operating system workflow automation. | |
| Autonomous On-Chain Agents and One-Person Unicorns | 4 | 6 | 2 | 2 | Furqan outlines autonomous digital entities combining LLM reasoning, social distribution, and crypto wallets. Shaan pulls up live terminal logs from Virtuals to examine real-time agent decision-making. | |
| Generative World Engines, Media Synthesis, and Human Taste | 6 | 4 | 3 | 5 | Furqan claims AI cannot replicate human taste in content curation. Shaan directly challenges this premise, arguing TikTok's algorithm has proven machines can curate taste far better than humans admit. | |
| Polymarket's Breakout and the Power of Prediction Markets | 5 | 4 | 1 | 2 | Shaan and Furqan discuss Polymarket's post-election validation, comparing it to earlier prediction markets and examining the French whale who used independent polling to exploit mispriced odds. | |
| Why VR Is a Massive Sleeping Giant | 5 | 6 | 2 | 2 | Furqan delivers a contrarian thesis on VR as a sleeping giant, citing Quest sales surpassing PS5 and massive revenue generated by indie titles like Gorilla Tag. Shaan cites specific hardware unit data. | |
| Surviving Emerging Tech Waves and Specialized Talent | 6 | 5 | 1 | 1 | Furqan introduces his emerging tech philosophy that survival is the only rule. Shaan reinforces this by drawing parallels to Flexport founder Ryan Petersen's default-alive mindset. | |
| Founding Founders Inc and Contrarian Capital Bets | 5 | 5 | 1 | 1 | Furqan shares the founding philosophy behind Founders Inc and taking contrarian bets during market downturns, including signing an SF waterfront lease during peak COVID. | |
| Execution Speed vs. Project Selection: Lessons from Blab and AppLovin | 7 | 5 | 1 | 2 | Shaan reflects on their past startup Blab, identifying that high-speed execution failed due to poor project selection, whereas AppLovin paired rapid iteration with massive market alignment. | |
| The Resurgence of Consumer Hardware and AI Prototyping | 5 | 6 | 1 | 1 | Furqan explains how Raspberry Pi, 3D printing, and cloud AI drastically lower hardware prototyping costs, demonstrating the concept through an AI-powered conversational teddy bear. | |
| Industrial Robotics, Automated Forklifts, and The Garage | 4 | 6 | 1 | 1 | Furqan outlines Founders Inc's 20,000-square-foot Garage expansion and bets on retrofitting industrial machines, such as automated forklifts and power-washing drones, with open-source robotics. |
Statements from this episode (30)
Rydhan: Thirdweb has deployed 8 to 10 internal AI agents
“And we've deployed probably eight or 10 of these throughout Third Web to do a lot of different functions.”
Rydhan: AI agents give 37-person Thirdweb the output of 80 people
“So like, we're like 37 people right now. I really believe we're more like 80 people. That's what it feels like with a lot of these tools.”
Rydhan: Any digital task with a clear directive can be automated with AI agents
“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.”
Rydhan: Every company should deploy AI workflows like early cloud computing
“And I think everyone should be deploying them. Every company should have them. It is a superpower. It's like, I can't, you know, it's like a feeling of almost like cloud, like, oh, I don't need a whole data center team now.”
Rydhan: Claude's 'computer use' is the next frontier for AI agents
“Claude has created this new capability called computer use. I think that's the next area that AI agents are going to enter.”
Rydhan: Perplexity saves multiple steps over slow, ineffective Google Search
“I try not to do Google search anymore, mostly because it's slow and ineffective and perplexity really made it where it does a search. It reads the result as I would. It clicks into the links as I would, and then it tries to answer my question more purposefully…”
Rydhan: Google's failure to lead in LLMs is a lifetime-scale fumble
“I think the biggest fumble
You know, in our lifetime.
Decade fumble.
Yeah.
They built the technology for Transformers.
Right.
The thing that OpenAI and others have used to develop this large language model, they are the ones that were the first entry and like …”
Rydhan: OpenAI and Anthropic are clearly number one and two in AI
“And actually, it feels like open AI, number one, that's, I think, clear. I think Anthropic is number two. So then everyone else shows up. Is what it feels like to me, at least from people I'm seeing building or the technology they're using or the innovation th…”
Rydhan: AI will plateau, but it will still transform all digital work
“People are going to say it's going to top out. It will. I think it's going to be absolutely impressive wherever it like starts slowing down at. And we'll completely change the way how we do any digital work.”
Rydhan: Autonomous AI agents will not use traditional credit cards
“These like digital things, like these agents, they're not gonna have credit cards.”
Rydhan: Future billion-dollar companies will be built by 1 to 3 people
“I feel like we're going to get to a world where, and I think Sam Altman said this, which is like the one person doing dollar company, like this is happening. Like we're already experiencing another 10 X decrease in how many people you need and the abilities th…”
Rydhan: Decart's Oasis generates a playable game world in real time
“It's really like a game that's
Fully built in a generative AI model.
And so they built a game like Minecraft and you could just describe a world and it makes you a Minecraft world like that.
And like every step you take is like not a pre-programmed pixel.
It's…”
Rydhan: AI will never have true taste or replace human creativity
“Machines will never have true taste, right? Like, and I think forever that creativity is going to come from humans, and there's always a new thing.”
Rydhan: A fully AI-generated song will likely hit number one in 2025
“If we were doing over under 20, 25 I might pick the under, you know, in 25, like, that would be a good bet.”
Puri: An incorrect Polymarket forecast would have faced harsher backlash than polls
“I think if they were, I think if Polymarket was wrong, the reaction would have been much worse than the fact that the polls were wrong, what the reaction is to the polls, you know?”
Puri: Polymarket's 'French whale' made $20M to $30M betting on Trump
“And so he kind of went counter to narrative and he made, I think like Something like 20, thirty million dollars yesterday. Yeah. On that bet.”
Puri: Meta Quest has sold 20M units total and 1M Quest 3s
“Quest has sold over twenty million units, with the majority being Quest two. Quest three is currently sell has sold a million units at the 500 dollar price point.”
Rydhan: Meta Quest headset sales have outpaced PlayStation 5 console sales
“And I believe it's outpaced. It's outpaced PS five sales, for example.”
Rydhan: Indie VR multiplayer game Gorilla Tag hit $200M in revenue
“I think Gorilla Tech has a two hundred million dollars of revenue.”
Rydhan: Emerging tech's primary strategy is surviving until the market scales
“In these emerging tech things, so everything at Founders 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 indust…”
Puri: Specialized tech teams can become $100M assets without a hit product
“So then if, even if you don't have a hit product, if you just assemble like a plus talent, a super specialized, then as those platforms rise, your team itself becomes like a hundred million dollar asset.”
Rydhan: Delays between deciding and acting undermine startup execution
“It's like, you know, think, decide, act, and like how fast you run through that depends on like the moment you decide the delay on act is like usually a problem.”
Puri: Live streaming app Blab scaled to 4 million users before failing
“We built an app like that. It got to four million users, but it didn't become the next big thing.”
Puri: Not pivoting Blab to B2B video cost hundreds of millions
“And instead of looking at those clues and being like, huh, maybe we could do that. We missed that project selection choice. I remember that day. Cause that was a, probably a multi hundred million dollar fork in the road moment, you know, maybe you still had to…”
Rydhan: Twitch won because 10-hour gaming sessions solved audience timing
“Okay, so we know we don't, because people come on for an hour, they do a show, the epic content is not on long enough for people to show up and intersect with it. How come Twitch wins? And then it turned out that, oh, people play games for eight to 10 hours.”
Rydhan: Founders Inc. prototyped 60 AI hardware toys for under $100K
“In hardware, this might have cost like half a million to a million dollars, I think we did that for like 50 to a hundred grand. So like that difference, one person, three D printers, raspberry pies, some AI, and you could just sit there and deliver units and t…”
Rydhan: Raspberry Pi has sold roughly 60 million units
“Raspberry Pi was thought of as this hobbyist market of tinkerers that are going to buy a few of them. They've sold like, sixty million units or something ridiculous like that.”
Rydhan: Practical robotics startups will become $100M teams
“I'm more excited about all these startups that are going to form. They're going to build this expertise. They're going to be a hundred million dollar teams, whether it's their product or their knowledge.”
Rydhan: Founders Inc. hosted roughly 40 Vision Pro developers in residency
“When the Vision Pro came out, we did it. We had about 40 Vision Pro devs. We probably had the largest concentration of Vision Pro devs out of Apple.”
Rydhan: Founders Inc. bets on 'negative one to zero' pre-idea founders
“We want to bet on negative one to zero. Like Peter Thiel talked about the zero to one companies. It's like, we still one step even before that. Like you're at negative one, you're wandering the forest, you need a place to tinker and like come and get the ideas…”