Viewing Podcast: Podcast
AI
Arts
Business
Crypto
Finance
Health
History
Interviews
Investing
Macro
Misc
News
Politics
Product
Programming
Science
Social
Startups
Technology
VC
How Crosby is Building an AI Law Firm on Deal Velocity not Billable Hours
Duration: 00:49:59
September 2, 2025
- Crosby is structured as an AI-first law firm, rather than legal software, to allow for unique telemetry and feedback loops by having domain experts working alongside engineers.
- Crosby is innovating on pricing by moving away from the billable hour and instead billing by the document, incentivizing efficiency and predictability in legal work costs.
- The company aims to build a better legal infrastructure by generalizing benchmarks in contracts and automating parts of the legal process, ultimately providing more people with access to legal services.
n8n CEO Jan Oberhauser on Building the Universal AI Automation Layer
Duration: 00:35:44
August 26, 2025
- NADN is focused on enabling users to build AI-powered applications, rather than simply adding AI features to an existing automation tool.
- A key turning point for NADN's growth was shifting the marketing strategy to prioritize community adoption over lead generation, leading to exponential growth as the community realized NADN had become an AI-focused tool.
- NADN maintains an open-source ethos by allowing free use of its source code, but prohibits commercialization to ensure the sustainability of the company, prioritizing data privacy and security over cost savings through open source.
Scaling the ‘Cursor for Slides’ to $50M ARR: Gamma founder Jon Noronha
Duration: 00:30:05
August 19, 2025
- Gamma's initial vision was to reinvent presentations, moving beyond being a mere PowerPoint clone to creating a new format for visual communication.
- Facing middling traction pre-AI, Gamma found a solution to the "blank page" problem by leveraging AI to generate rough drafts, shifting user effort from creation to editing.
- Gamma emphasizes a "lean paranoid mindset" and experimentation, prioritizing cost efficiency and AB testing across multiple models, like Claude, Gemini, and GPT, to optimize prompt engineering.
Delphi’s Dara Ladjevardian: How AI Digital Minds Can Scale Human Connection
Duration: 00:39:05
August 12, 2025
- Deli is creating "digital minds," AI representations of individuals, that go beyond mimicking speech to capture thinking and reasoning processes.
- The company's origin stems from the founder's personal need to connect with a digital version of his late grandfather, highlighting the potential for AI to facilitate human connection and mentorship.
- Deli envisions a future where AI-powered digital minds become commonplace, shifting online interactions from information feeds to personalized conversational experiences.
Vercel CEO Guillermo Rauch: Building the Generative Web with AI
Duration: 01:00:59
August 5, 2025
- Vercel's Vzero is democratizing software creation by attracting non-developers like designers and marketers, who can now generate functional front-end apps from text prompts.
- AI is transforming software development metrics, providing immediate feedback loops based on code functionality and user engagement, which is a significant improvement over traditional developer experience measurement.
- The future web will be generative, with applications created on-demand for individual users, potentially rendering traditional, downloadable software obsolete.
OpenAI’s IMO Team on Why Models Are Finally Solving Elite-Level Math
Duration: 00:30:10
July 30, 2025
- OpenAI's model achieved a gold medal performance at the International Math Olympiad (IMO), marking a significant milestone in AI development.
- The team emphasized the use of general-purpose techniques for scaling compute and handling hard-to-verify tasks, rather than bespoke solutions tailored only to math competitions.
- While the model excels at competition math, the next frontier involves tackling problems requiring significantly longer reasoning times, with the ultimate goal of facilitating novel research breakthroughs.
OpenAI Just Released ChatGPT Agent, Its Most Powerful Agent Yet
Duration: 00:37:36
July 22, 2025
- The new ChatGPT agent combines Deep Research and Operator functionalities, unifying the architecture to allow seamless transitions between tasks like browsing, text analysis, and code execution within a shared state environment.
- The agent is trained using reinforcement learning across thousands of virtual machines, enabling it to discover optimal strategies for tool usage without explicit programming, learning to effectively switch between different tools to solve tasks.
- A prioritized safety approach involves monitoring the agent's activities, implementing guardrails, and conducting red teaming exercises to mitigate risks such as data exfiltration, biohazards, and falling victim to scams, with ongoing efforts to improve robustness and rapidly respond to newly discovered attacks.
DeepMind's Pushmeet Kohli on AI's Scientific Revolution
Duration: 00:41:13
July 11, 2025
- AlphaFold represents a major breakthrough in protein structure prediction, enabling faster understanding of biological mechanisms and drug development across various diseases.
- The impact of AlphaFold is already evident in scientific research, highlighted by recent papers solving decades-old problems, like the mechanism of the nuclear pore complex.
- While AlphaFold has limitations, like predicting protein dynamics, future advancements could focus on modeling interactions with small molecules to further enhance drug discovery capabilities.
Mapping the Mind of a Neural Net: Goodfire’s Eric Ho on the Future of Interpretability
Duration: 00:47:07
July 8, 2025
- Goodfire aims to "unlock the black box" of AI, allowing for intentional design rather than solely relying on data-driven growth.
- Mechanistic interpretability seeks to "unscramble" neurons to reveal clean, interpretable concepts and can enhance good AI behavior while removing harmful actions.
- The speaker predicts that the interpretability field in general and Goodfire in particular will "figure out" the features, circuits, and patterns, and therefore confidently understand neural nets by 2028.
ElevenLabs’ Mati Staniszewski: Why Voice Will Be the Fundamental Interface for Tech
Duration: 00:59:53
July 1, 2025
- 11 Labs has maintained a competitive edge against larger AI labs by staying focused on audio-specific AI advancements, particularly in research model development.
- The company's origin stemmed from identifying the poor quality of dubbed foreign films in Poland, inspiring them to leverage AI for high-quality voice replication and speech generation.
- 11 Labs is seeing growth in voice agent applications, believing voice will become a primary interface for technology, especially in education and breaking down cross-lingual communication barriers.