Latest Interviews
Showing 1–10 of 10 interview transcripts.
Clear all filters- a16z1h 20m
How Decagon Runs 90% of Its Agents on Open-Source Models
Sarah Wang, Kimberly Tan, Jesse Zhang, Ashwin Sreenivas
Decagon Labs challenges the prevailing 2026 narrative that AI agents will replace all software by establishing a dedicated "model factory" that prioritizes open-source fine-tuning over frontier models to achieve superior latency for voice interactions. The company differentiates its enterprise deployment through a "glass box" approach and its Duet Autopilot system, which automates the creation of agent procedures and reduces production timelines from years to months. Despite rapid international expansion and the lowering of technical barriers, Decagon contends that the primary constraint remains human talent acquisition while affirming that AI will drive new service demand rather than eliminating careers or rendering legacy SaaS obsolete.
- a16z54 min
Building Agents at Home: Homeschooling, Parenting and More | The a16z Show
Katherine Boyle, Sarah Wang, Jesse Genet
Former Y Combinator founder Jessie utilizes a fleet of eleven autonomous agents built on the Open-Claw framework to simultaneously manage a homeschooling household for four young children and develop advanced technical products. By adopting a "benevolent neglect" parenting strategy and delegating daily tasks to specialized software agents running on isolated Mac Minis, she successfully reverses typical trade-offs between deep tech work and active family engagement. This operational model demonstrates how non-technical founders can achieve a Cambrian explosion in personal capability, potentially reshaping future domestic economics by making remote, AI-driven parenthood viable without traditional startup pressures.
- a16z1h 5m
Ben Horowitz and Ali Ghodsi: How to Run a $100 Billion Business
Ben Horowitz, Ali Ghodsi, Sarah Wang, Erik Torenberg
In 2016, Databricks CEO Ali Ghodsi executed a critical strategic pivot from open-source distribution to a B2B enterprise sales model to overcome the open source paradox and secure proprietary revenue. This transformation required hiring non-PhD sales veterans and forging a high-stakes Microsoft partnership that aligned Databricks' technical capabilities with Microsoft's massive distribution channel. Under Ghodsi's leadership, the company maintained a rigorous acquisition strategy prioritizing cultural fit over immediate financial metrics while retaining top engineering talent through competitive compensation and a private equity structure.
- a16z53 min
From Vibe Coding to Vibe Researching: OpenAI’s Mark Chen and Jakub Pachocki
Mark Chen, Jakub Pachocki, Anjney Midha, Sarah Wang
OpenAI researchers Mark and Jakob Sutskever outline a strategic roadmap centered on GPT-5, which aims to mainstream advanced reasoning and automate scientific discovery by merging the capabilities of instant-response and deep-thought models. This approach shifts evaluation metrics from solving static competition problems to generating economically relevant insights and extending autonomous time horizons to several hours through reinforcement learning. The organization distinguishes itself by balancing protected fundamental research teams with product accountability, prioritizing talent that persists through failure to overcome current limitations in coding autonomy and physical robotics.
- a16z1h 39m
Dylan Patel on the AI Chip Race - NVIDIA, Intel & the US Government vs. China
Dylan Patel, Erik Torenberg, Sarah Wang, Guido Appenzeller
NVIDIA and Intel have established a transformative $5 billion strategic partnership to jointly develop custom data center and PC products, a move that significantly boosted NVIDIA's stock and acknowledged the industry's shift from CPU to GPU dominance. This alliance contrasts sharply with China's aggressive pursuit of semiconductor self-sufficiency, where Huawei navigates US export bans and high-bandwidth memory bottlenecks through domestic innovation and alleged smuggling channels. Simultaneously, hyperscalers like Oracle and AWS are capitalizing on surging demand with massive infrastructure deals and repurposed capacity, while the market faces technical complexities in deploying next-generation Blackwell architectures and optimizing hardware for specific inference workloads.
- a16z43 min
GPT-5 and Agents Breakdown – w/ OpenAI Researchers Isa Fulford & Christina Kim
Isa Fulford, Christina Kim, Erik Torenberg, Sarah Wang
OpenAI leadership and key architects Christina and Issa unveil GPT-5 as a significantly more useful frontier model that prioritizes cross-domain utility, massive coding improvements, and reduced hallucinations over isolated benchmark scores. The update integrates advanced reasoning capabilities and autonomous agent functions to lower barriers for non-technical users while shifting evaluation metrics toward real-world usage data and specialized synthetic datasets. By optimizing pricing and balancing helpfulness against sycophancy, the release aims to establish a new baseline for general intelligence that transforms complex, multi-step workflows into accessible daily tools.
- a16z35 min
How to Build a Thriving AI Ecosystem with Lisa Su, CEO of AMD
AMD Chair and CEO Lisa Su outlines the company's transformation into a leading AI infrastructure provider through strategic acquisitions of Xilinx and ZT Systems, alongside a commitment to open standards like UALink. Su emphasizes that overcoming current supply bottlenecks in advanced packaging and high-bandwidth memory requires a shift from efficiency to supply chain resiliency supported by the CHIPS Act. This approach enables AMD to partner with major hyperscalers and startups to deliver heterogeneous computing solutions capable of training next-generation models with trillions of parameters.
- a16z37 min
How Waymo Is Using GenAI to Build a Better Driver
Dmitri Dolgov, David George, Sarah Wang
Dimitri Dolgov traces the evolution of autonomous driving from the 2007 DARPA Urban Challenge to Waymo's current deployment of generalizable AI across four major US cities. By combining foundational machine learning breakthroughs like Transformers with massive synthetic simulation, Waymo has demonstrated a 76% reduction in property damage and operates 24/7 in diverse weather conditions. The company now focuses on scaling complex urban logistics while maintaining a hybrid market strategy that partners with Uber and prioritizes rigorous safety validation over short-term commercial shortcuts.
- a16z30 min
Bringing AI to the Masses with Adam D'Angelo, CEO of Quora
Adam D'Angelo, David George, Sarah Wang, Max Wiethe, Dan Morehead, Matt Rowe
Former Quora CEO Adam D'Angelo launched Poe as a multi-platform aggregator to navigate the shift from human-centric knowledge publishing to an instant, multi-model chat paradigm driven by generative AI. The platform differentiates itself by supporting diverse fine-tuned models and an open API that enables independent creators to deploy specialized tools while sharing revenue to offset high inference costs. D'Angelo anticipates a bifurcated market where Poe serves the downstream application layer of accessible AI, eventually evolving into a symbiotic network that complements Quora's human expertise with scalable automated agents.
- a16z17 min
Universally Accessible Intelligence with Character.ai's Noam Shazeer
Co-founder Noam Shazeer details Character.ai's vertically integrated strategy to accelerate AGI development by prioritizing generalizable dialogue models over domain-specific applications, a choice that has driven over 20 billion messages and sustained two hours of daily engagement per user. The company leverages current scaling laws to rapidly reduce training costs while projecting a model tens of IQ points smarter by year-end alongside multimodal enhancements like voice and long-term memory. Shazeer frames the technology as a pre-AGI breakthrough comparable to the Wright Brothers' first flight, predicting ubiquitous, affordable intelligence within a year as global H100 infrastructure expands, while dismissing immediate existential risks in favor of a four-month pause for safety verification.