Episode Summary
Executive Summary: The conversation argues that AI agents, VR, and hardware robotics are entering a new phase where small teams can outperform much larger ones. The guest explains how he is using agents for sales, calendar, and workflow automation, why generative AI will reshape content and software, and why underappreciated sectors like VR and consumer hardware are becoming viable again thanks to cheaper prototyping tools and better models.
Main Topics: AI agents as digital workers (Priority: 5/5): The guest defines AI agents as LLM-based systems with reasoning loops that can plan, research, decide, and act on clear digital tasks. He shares examples at Thirdweb and in his personal life, including signup triage, personalized outreach, and calendar management. Workflow automation and small-team leverage (Priority: 5/5): He emphasizes that tools like agents and no-code workflow builders let a 37-person company function like an 80-person team by automating repeatable digital work and adding judgment to processes. The AI platform landscape (Priority: 4/5): The discussion compares OpenAI, Anthropic/Claude, and Perplexity, highlighting their different strengths: foundational models, computer use/browser-like interaction, and search-plus-reasoning respectively. VR as a sleeping giant (Priority: 5/5): The guest argues VR has quietly matured through Quest adoption, social games, and productivity use cases, and that teams can now build profitable businesses with small headcounts on platforms like App Lab. Hardware and robotics are becoming tinkerable again (Priority: 5/5): He describes how Raspberry Pi, 3D printing, NVIDIA Jetson, and cloud AI have lowered the barrier for consumer hardware and robotics, enabling small teams to prototype toys, interfaces, drones, and warehouse automation. Project selection, survival, and long-term compounding (Priority: 4/5): A recurring theme is that emerging tech rewards staying alive long enough for the wave to arrive. The guest contrasts good execution with bad project selection and argues that founders should survive, iterate, and avoid burning out too early. Inversion on narrative and contrarian bets (Priority: 4/5): He repeatedly points to opportunities where public sentiment is negative but product reality is improving, such as VR, crypto, San Francisco real estate, and Google search versus newer AI tools.
Key Arguments: AI agents are useful when the task is digital, clearly directed, and can be expressed as a mission with constraints; they can research, reason, and act without step-by-step human supervision. A company can gain major leverage by deploying a handful of specialized agents for sales, scheduling, email, and other repetitive work. Personal agents are already practical for individuals; his calendar and email workflows now run 24/7 with a few hours of setup. OpenAI and Anthropic are the leading model platforms, while Perplexity wins by combining search with reasoning in a way that saves multiple manual steps. Google missed the developer mindshare shift because the AI leader narrative moved elsewhere, and that is hard to reverse once it happens. VR is no longer just a hype cycle; it now has real distribution, real businesses, and a growing generation of users who find headsets more natural than desktops or TVs. Small teams can succeed in VR because the platform is still early enough that quality supply is scarce, so a few talented builders can become top apps quickly. Consumer hardware is becoming feasible for tiny teams because off-the-shelf components and AI dramatically reduce prototyping cost and complexity. Robotics and industrial automation are entering a similar phase: machines like drones, forklifts, and warehouse systems can be upgraded with cheap compute and vision systems. The winning strategy in emerging tech is to survive the valley of doubt until the platform inflects; timelines are hard to control, but staying in the game is controllable. Project selection matters as much as execution; many of the speaker's earlier product efforts failed because they chose the wrong opportunities despite strong teams.
Data Points: Thirdweb team size: 37 people - Used to illustrate how AI tools make the company feel like a much larger organization. Implied effective team size with AI: about 80 people - Guest says automation makes the 37-person team feel closer to 80. Deployment count of agents at Thirdweb: 8 to 10 agents - Different agent workflows were deployed across the company for multiple functions. Calendar setup time: about 15 minutes - He connected his calendar and built a query interface in one short session. Additional calendar automation setup: another 45 minutes to 1 hour - He added commands to block time and query meeting patterns. Weekly meeting load: 28 hours - An AI query of his calendar showed too many meetings last week. Quest headset sales: over 20 million units - Used to argue VR has meaningful distribution. Quest 3 sales: 1 million units - Mentioned at a $500 price point. Quest 2 global usage: 5 to 10 million monthly active headsets - Cited as evidence of a real active platform base. Gorilla Tag revenue: $200 million - Example of a breakout VR-native social game. Fluid weekly active users: about 5,000 - Productivity/browser VR app still attracting users via App Lab. Magical Toys units shipped: 60 to 70 units - Prototype AI teddy bear shipped over several months. Magical Toys prototype budget: $50,000 to $100,000 - Speaker contrasts this with older hardware costs. Historical hardware startup cost: $500,000 to $1,000,000 - Approximate cost he says similar prototyping would have required before. Founders/industry residency for Vision Pro: 40 Vision Pro devs - He says they concentrated a large developer cohort around a new platform. AI/agents concept bet: negative one to zero - He frames the studio’s thesis as investing before zero-to-one, during the phase where an idea is still forming.
Pivotal Quotes: "AI agents are using LLMs or AI systems ... but then giving it reasoning loops." — Furkan: Core definition of what makes an AI agent different from a chatbot. "We want to bet on negative one to zero." — Furkan: Describes the studio’s strategy of investing before a market becomes obvious. "Technology takes a long time. When it hits though, it happens very fast." — Furkan: Summarizes his thesis on emerging tech adoption and survival.
Implications: Listeners should expect more software work to be automated, and more opportunities to emerge in VR, consumer hardware, and robotics for tiny teams. The edge will increasingly come from being early, staying alive, and pairing strong taste with fast iteration.
About My First Million
Sam Parr and Shaan Puri brainstorm new business ideas based on trends & opportunities they see in the market. Sometimes they bring on famous guests to brainstorm with them.