Jun 23, 2025 · 55m · tbpn

We Interviewed the Best Startups From YC Demo Day (June 2025)

Garry Tan · 9m spoken Aaron Cannon · 6m spoken
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

Filmed live at Y Combinator Demo Day, this broadcast features YC CEO Garry Tan and a selection of standout founders demonstrating breakthrough startups in AI agents, physics-informed simulation, generative robotics, user research, education, and mobile app creation.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →

The hosts as informed peer 5.6 Guest teaching 4.4 Guest disagreement 1.1 The hosts pushing back 1.7
05100:0015:0030:0045:000:00–15:48 · The hosts as informed peer 5/10 Gary Tan on YC Batch Metrics, AI Trajectory, and Founder Agency John and his co-host converse easily with Gary Tan about batch growth metrics and non-GAAP revenue definitions. Gary educates the hosts on macro trends, including a 110% surge in young applicants and surprising CS graduate unemployment statistics.15:49–22:55 · The hosts as informed peer 5/10 Agentic Search and High-Scale Recruiting with Plato AI The hosts interview Plato AI's founders on agentic search and recruiting economics. John probes their parallel query architecture and pricing structure, while the founders explain how they index 200M profiles efficiently.22:57–34:08 · The hosts as informed peer 6/10 Outset on Scaling Qualitative User Research with AI Moderators Aaron Cannon outlines how Outset conducts scaled qualitative interviews. John shows strong familiarity with enterprise research dynamics, Azure model routing, and comparing Outset to traditional expert networks like GLG.34:09–39:26 · The hosts as informed peer 7/10 Godella on Physics-Informed Machine Learning for Engineering Simulations John demonstrates impressive domain knowledge in computational physics, specifically citing reinforcement learning with verifiable rewards, computational fluid dynamics (CFD), and drag modeling. The founders explain their mesh encoding approach.39:28–43:48 · The hosts as informed peer 5/10 Morpho AI on Generative Software for Custom Robot Hardware Design The hosts discuss Morpho AI's generative robotic design tool. John asks insightful questions about the OEM ecosystem versus integrators, while the founders explain why simulation-trained models are necessary due to lack of real-world CAD log datasets.43:49–49:49 · The hosts as informed peer 5/10 Vibegrade on In-Workflow AI-Assisted Grading and Student Feedback John pushes back on the premise of AI grading by pointing out the circular absurdity of students using ChatGPT to write essays and teachers using AI to grade them. The founders defend the system by highlighting the importance of actionable feedback.49:50–55:02 · The hosts as informed peer 6/10 Bloom on Instant Native Mobile App Creation and Distribution John demonstrates strong technical intuition by immediately identifying that Bloom leverages iOS App Clips to bypass App Store distribution friction. The hosts playfully tease the founder during an initial AirDrop demo glitch.0:00–15:48 · Guest teaching 5/10 Gary Tan on YC Batch Metrics, AI Trajectory, and Founder Agency John and his co-host converse easily with Gary Tan about batch growth metrics and non-GAAP revenue definitions. Gary educates the hosts on macro trends, including a 110% surge in young applicants and surprising CS graduate unemployment statistics.15:49–22:55 · Guest teaching 4/10 Agentic Search and High-Scale Recruiting with Plato AI The hosts interview Plato AI's founders on agentic search and recruiting economics. John probes their parallel query architecture and pricing structure, while the founders explain how they index 200M profiles efficiently.22:57–34:08 · Guest teaching 5/10 Outset on Scaling Qualitative User Research with AI Moderators Aaron Cannon outlines how Outset conducts scaled qualitative interviews. John shows strong familiarity with enterprise research dynamics, Azure model routing, and comparing Outset to traditional expert networks like GLG.34:09–39:26 · Guest teaching 5/10 Godella on Physics-Informed Machine Learning for Engineering Simulations John demonstrates impressive domain knowledge in computational physics, specifically citing reinforcement learning with verifiable rewards, computational fluid dynamics (CFD), and drag modeling. The founders explain their mesh encoding approach.39:28–43:48 · Guest teaching 4/10 Morpho AI on Generative Software for Custom Robot Hardware Design The hosts discuss Morpho AI's generative robotic design tool. John asks insightful questions about the OEM ecosystem versus integrators, while the founders explain why simulation-trained models are necessary due to lack of real-world CAD log datasets.43:49–49:49 · Guest teaching 4/10 Vibegrade on In-Workflow AI-Assisted Grading and Student Feedback John pushes back on the premise of AI grading by pointing out the circular absurdity of students using ChatGPT to write essays and teachers using AI to grade them. The founders defend the system by highlighting the importance of actionable feedback.49:50–55:02 · Guest teaching 4/10 Bloom on Instant Native Mobile App Creation and Distribution John demonstrates strong technical intuition by immediately identifying that Bloom leverages iOS App Clips to bypass App Store distribution friction. The hosts playfully tease the founder during an initial AirDrop demo glitch.0:00–15:48 · Guest disagreement 1/10 Gary Tan on YC Batch Metrics, AI Trajectory, and Founder Agency John and his co-host converse easily with Gary Tan about batch growth metrics and non-GAAP revenue definitions. Gary educates the hosts on macro trends, including a 110% surge in young applicants and surprising CS graduate unemployment statistics.15:49–22:55 · Guest disagreement 1/10 Agentic Search and High-Scale Recruiting with Plato AI The hosts interview Plato AI's founders on agentic search and recruiting economics. John probes their parallel query architecture and pricing structure, while the founders explain how they index 200M profiles efficiently.22:57–34:08 · Guest disagreement 1/10 Outset on Scaling Qualitative User Research with AI Moderators Aaron Cannon outlines how Outset conducts scaled qualitative interviews. John shows strong familiarity with enterprise research dynamics, Azure model routing, and comparing Outset to traditional expert networks like GLG.34:09–39:26 · Guest disagreement 1/10 Godella on Physics-Informed Machine Learning for Engineering Simulations John demonstrates impressive domain knowledge in computational physics, specifically citing reinforcement learning with verifiable rewards, computational fluid dynamics (CFD), and drag modeling. The founders explain their mesh encoding approach.39:28–43:48 · Guest disagreement 1/10 Morpho AI on Generative Software for Custom Robot Hardware Design The hosts discuss Morpho AI's generative robotic design tool. John asks insightful questions about the OEM ecosystem versus integrators, while the founders explain why simulation-trained models are necessary due to lack of real-world CAD log datasets.43:49–49:49 · Guest disagreement 2/10 Vibegrade on In-Workflow AI-Assisted Grading and Student Feedback John pushes back on the premise of AI grading by pointing out the circular absurdity of students using ChatGPT to write essays and teachers using AI to grade them. The founders defend the system by highlighting the importance of actionable feedback.49:50–55:02 · Guest disagreement 1/10 Bloom on Instant Native Mobile App Creation and Distribution John demonstrates strong technical intuition by immediately identifying that Bloom leverages iOS App Clips to bypass App Store distribution friction. The hosts playfully tease the founder during an initial AirDrop demo glitch.0:00–15:48 · The hosts pushing back 2/10 Gary Tan on YC Batch Metrics, AI Trajectory, and Founder Agency John and his co-host converse easily with Gary Tan about batch growth metrics and non-GAAP revenue definitions. Gary educates the hosts on macro trends, including a 110% surge in young applicants and surprising CS graduate unemployment statistics.15:49–22:55 · The hosts pushing back 1/10 Agentic Search and High-Scale Recruiting with Plato AI The hosts interview Plato AI's founders on agentic search and recruiting economics. John probes their parallel query architecture and pricing structure, while the founders explain how they index 200M profiles efficiently.22:57–34:08 · The hosts pushing back 1/10 Outset on Scaling Qualitative User Research with AI Moderators Aaron Cannon outlines how Outset conducts scaled qualitative interviews. John shows strong familiarity with enterprise research dynamics, Azure model routing, and comparing Outset to traditional expert networks like GLG.34:09–39:26 · The hosts pushing back 2/10 Godella on Physics-Informed Machine Learning for Engineering Simulations John demonstrates impressive domain knowledge in computational physics, specifically citing reinforcement learning with verifiable rewards, computational fluid dynamics (CFD), and drag modeling. The founders explain their mesh encoding approach.39:28–43:48 · The hosts pushing back 1/10 Morpho AI on Generative Software for Custom Robot Hardware Design The hosts discuss Morpho AI's generative robotic design tool. John asks insightful questions about the OEM ecosystem versus integrators, while the founders explain why simulation-trained models are necessary due to lack of real-world CAD log datasets.43:49–49:49 · The hosts pushing back 3/10 Vibegrade on In-Workflow AI-Assisted Grading and Student Feedback John pushes back on the premise of AI grading by pointing out the circular absurdity of students using ChatGPT to write essays and teachers using AI to grade them. The founders defend the system by highlighting the importance of actionable feedback.49:50–55:02 · The hosts pushing back 2/10 Bloom on Instant Native Mobile App Creation and Distribution John demonstrates strong technical intuition by immediately identifying that Bloom leverages iOS App Clips to bypass App Store distribution friction. The hosts playfully tease the founder during an initial AirDrop demo glitch.

speaking balance: gold is the hosts, purple is the guest (3 minute bins)

0:00 · the hosts 0% · guest 100%0:00 · the hosts 0% · guest 100%3:00 · the hosts 0% · guest 100%3:00 · the hosts 0% · guest 100%6:00 · the hosts 0% · guest 100%6:00 · the hosts 0% · guest 100%9:00 · the hosts 0% · guest 100%9:00 · the hosts 0% · guest 100%12:00 · the hosts 0% · guest 100%12:00 · the hosts 0% · guest 100%15:00 · the hosts 0% · guest 100%15:00 · the hosts 0% · guest 100%18:00 · the hosts 0% · guest 100%18:00 · the hosts 0% · guest 100%21:00 · the hosts 0% · guest 100%21:00 · the hosts 0% · guest 100%24:00 · the hosts 0% · guest 100%24:00 · the hosts 0% · guest 100%27:00 · the hosts 0% · guest 100%27:00 · the hosts 0% · guest 100%30:00 · the hosts 0% · guest 100%30:00 · the hosts 0% · guest 100%33:00 · the hosts 0% · guest 100%33:00 · the hosts 0% · guest 100%36:00 · the hosts 0% · guest 100%36:00 · the hosts 0% · guest 100%39:00 · the hosts 0% · guest 100%39:00 · the hosts 0% · guest 100%42:00 · the hosts 0% · guest 100%42:00 · the hosts 0% · guest 100%45:00 · the hosts 0% · guest 100%45:00 · the hosts 0% · guest 100%48:00 · the hosts 0% · guest 100%48:00 · the hosts 0% · guest 100%51:00 · the hosts 0% · guest 100%51:00 · the hosts 0% · guest 100%54:00 · the hosts 0% · guest 100%54:00 · the hosts 0% · guest 100%
Sharpest disagreement ▶ 47:01 Daniel rejects the cynical take on AI student workflows

Daniel directly refutes the common criticism that AI grading is pointless when students also use LLMs, asserting that students genuinely seeking to learn need faster feedback loops.

Hardest push from the hosts ▶ 46:40 John challenges the utility of AI grading AI output

John frames the contemporary homework workflow as an absurd loop where teachers generate prompts with ChatGPT, students answer with ChatGPT, and teachers grade with ChatGPT.

Biggest teaching moment ▶ 2:34 Gary Tan shares startling demographic and labor market shifts

Gary informs the hosts that YC applications from 18-to-22-year-olds surged by 110% alongside an unexpected spike in CS graduate unemployment relative to humanities majors.

The host holds their own ▶ 36:39 John demonstrates deep mechanical engineering simulation knowledge

John breaks down physics ML by referencing computational fluid dynamics, airfoil drag calculations, and inference speedup over deterministic physics equations.

the scores for every segment, with the reasoning behind each
ChapterTopicThe hosts as informed peerGuest teachingGuest disagreementThe hosts pushing backWhy
Gary Tan on YC Batch Metrics, AI Trajectory, and Founder Agency 5512 John and his co-host converse easily with Gary Tan about batch growth metrics and non-GAAP revenue definitions. Gary educates the hosts on macro trends, including a 110% surge in young applicants and surprising CS graduate unemployment statistics.
Agentic Search and High-Scale Recruiting with Plato AI 5411 The hosts interview Plato AI's founders on agentic search and recruiting economics. John probes their parallel query architecture and pricing structure, while the founders explain how they index 200M profiles efficiently.
Outset on Scaling Qualitative User Research with AI Moderators 6511 Aaron Cannon outlines how Outset conducts scaled qualitative interviews. John shows strong familiarity with enterprise research dynamics, Azure model routing, and comparing Outset to traditional expert networks like GLG.
Godella on Physics-Informed Machine Learning for Engineering Simulations 7512 John demonstrates impressive domain knowledge in computational physics, specifically citing reinforcement learning with verifiable rewards, computational fluid dynamics (CFD), and drag modeling. The founders explain their mesh encoding approach.
Morpho AI on Generative Software for Custom Robot Hardware Design 5411 The hosts discuss Morpho AI's generative robotic design tool. John asks insightful questions about the OEM ecosystem versus integrators, while the founders explain why simulation-trained models are necessary due to lack of real-world CAD log datasets.
Vibegrade on In-Workflow AI-Assisted Grading and Student Feedback 5423 John pushes back on the premise of AI grading by pointing out the circular absurdity of students using ChatGPT to write essays and teachers using AI to grade them. The founders defend the system by highlighting the importance of actionable feedback.
Bloom on Instant Native Mobile App Creation and Distribution 6412 John demonstrates strong technical intuition by immediately identifying that Bloom leverages iOS App Clips to bypass App Store distribution friction. The hosts playfully tease the founder during an initial AirDrop demo glitch.

Statements from this episode (9)

Assertion Not checkable as stated
Tan: YC Spring 2025 batch is 90% AI, batch revenue growth hit 12%
“It's like 90% AI. It's about 10% 11% hard tech, which is awesome. And then the really crazy stat is, you know, how the last four batches, about the last year you know, the batch itself as a whole has been growing revenue by 10%. This time it's 12.”
Garry Tan Jun 23, 2025 ▶ 1:52
Assertion Not checkable as stated
Tan: 18-to-22-year-old YC applicants and acceptances increased 110% year-over-year
“The wild stat that we're seeing is actually the number of 18 to 22 year olds applying to YC and getting is up a 110%. Year on year, right?”
Garry Tan Jun 23, 2025 ▶ 2:51
Prediction Not checkable as stated
Tan: Platform neutrality could enable hundreds of thousands of 10-person $1B companies
“We actually need the platforms to allow other people to enter, but if we do that, we can actually have tens of thousands or hundreds of thousands of companies, each of which can get to a billion dollars net revenue each, you know. And like, you could do it pro…”
Garry Tan Jun 23, 2025 ▶ 5:30
Assertion Not checkable as stated
Tan: Roughly 30% of YC companies pivot during their batch
“Something like 30% of the companies change their idea during the batch, and that's actually great.”
Garry Tan Jun 23, 2025 ▶ 9:37
Assertion Not checkable as stated
Tan: Y Combinator is currently accepting startups at a 0.8% rate
“We're accepting companies at a 0.8% rate right now.”
Garry Tan Jun 23, 2025 ▶ 14:43
Assertion Not checkable as stated
Tan: YC alumni Axiom generated hundreds of millions in pure profits
“And hundreds of millions of dollars in pure profits.”
Garry Tan Jun 23, 2025 ▶ 15:32
Disclosure
Cannon: Outset's early enterprise customers include Microsoft, Nestle, and Weight Watchers
“So there's a lot of opportunity, and yeah, we work with, like, Microsoft, Nestle, Weight Watchers, so we've been super enterprise focused, and yeah, we're a couple years in.”
Aaron Cannon Jun 23, 2025 ▶ 24:48
Insight
Cannon: Users share more candidly with AI interviewers than human moderators
“In fact, they like, have even expressed that like, they'll share more because there isn't a person. It's this idea of like social desirability bias. You're like, I don't want to come off as a person that is X, Y, and Z to another human.”
Aaron Cannon Jun 23, 2025 ▶ 29:30
Insight
Cannon: Avoid AI moderators if user research is a sales opportunity
“Yeah, there's basically, I, like, I think a good reason not to use an AI moderator is, like, if you could sell to that person. Probably don't outsource that, right? Like, maybe to build the relationship yourself.”
Aaron Cannon Jun 23, 2025 ▶ 30:30
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 500 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.