The Room Podcast

In this episode of The Room Podcast, we speak with Carina Hong, founder and CEO of Axiom, about building AI that can reason through mathematics and generate verifiable code. Carina traces her path from math competitions in Guangzhou to studying mathematics and physics at MIT, becoming a Rhodes Scholar, and enrolling in Stanford’s JD/PhD program before leaving to build a company. She reflects on the difference between the quick rewards of competitions and the long, uncertain work of mathematical research, and why AI could help researchers move beyond problems that have kept them stuck for years.Carina shares the unlikely coffee-shop friendship that became a co-founder relationship, how Axiom brought together people from mathematics, machine learning, and hardware, and the encouragement that pushed her toward fundraising. She explains Axiom’s focus on verified code generation, the role of Lean in mathematical proofs, and her vision for quantitative superintelligence: systems that can generate new conjectures, develop theories, and connect mathematical reasoning to practical software.The conversation also explores the relationship between a research lab and a commercial business, the limitations of model benchmarks, open versus closed AI, and the potential for stronger guarantees in software and physical systems. Carina discusses team culture, the global AI landscape, and what she hopes Axiom will make possible next. She closes with a tribute to Blair, a friend from the Ross Mathematics Program whose intelligence, kindness, and patience shaped her approach to both research and teaching.Learn more about Carina Hong and explore Axiom.(04:34) Where did Carina grow up, and how did it shape her worldview?(05:28) Did a childhood in math competitions point toward entrepreneurship?(07:39) From MIT and Oxford to leaving Stanford’s JD/PhD program(11:30) The moment AI changed Carina’s view of mathematical discovery(12:43) What problem is Axiom solving, and what is its core product?(14:06) Meeting a future co-founder at a coffee shop(16:16) Turning an intellectual friendship into a company(16:41) Building a team across mathematics, AI, and hardware(18:18) What does quantitative superintelligence mean?(20:00) Humility, ambition, and learning as a founder(22:38) Who first encouraged Carina to raise capital?(24:05) What did not go as planned while building Axiom?(27:27) Is Axiom a research lab, a software company, or both?(29:33) Connecting mathematical research with verified software(34:22) How does verification differ from a benchmark score?(34:57) Model benchmarks and the gap between marketing and capability(36:08) Open versus closed models and formal verification(37:49) Employee projects, curiosity, and DrivingBench(39:20) Physical AI and systems that need stronger guarantees(41:13) Team structure, culture, and the tradeoffs of building a lab(43:57) How Carina thinks about the timeline for capable AI agents(48:02) Formalization and the relationship between human and machine mathematics(50:41) China, the United States, and the global AI landscape(53:52) What is next for Axiom?(56:57) Carina’s personal goals as the company grows(58:23) A woman who has had a profound impact on Carina’s lifeFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Podcasts!Brought to you by: Rippling, Gradial, and Perkins CoieWX Productions

In this episode of The Room Podcast, Madison and Claudia sit down in the studio with Cecilia Ziniti, co-founder and CEO of GC AI. Cecilia shares how growing up as an immigrant, working inside Yahoo’s early trust and safety team, and serving as counsel at technology companies shaped her path from law to entrepreneurship. Cecilia explains the moment a pre-ChatGPT language model convinced her that legal work could change, and how she and co-founder Bardia set out to build a tool for in-house legal teams. She talks about finding early customers, following unexpected signals from the market, fundraising, and why legal buyers quickly saw value in a product built around their actual work.The conversation turns to agents: how a legal team can delegate multistep work, why lawyers need a clear record of what an agent checked, and what AI may free them to do creatively and strategically. Cecilia also looks ahead to how legal teams could work in 2029, the questions of trust and responsibility that come with more capable agents, and the teachers in her family who shaped her approach to helping people adopt new technology.Learn more about Cecilia Ziniti and explore GC AI.(04:15) Where did Cecilia grow up, and how did it shape her worldview?(05:14) Did Cecilia always think she would become a founder?(06:34) Working at Yahoo before law school(07:50) What trust and safety meant in the early days of consumer technology(10:14) Helping the law keep pace with new technology at Cruise(11:46) What Amazon Alexa and autonomous vehicles taught Cecilia about AI(13:58) The pre-ChatGPT moment that led to GC AI(15:57) Meeting co-founder Bardia and deciding to build together(19:37) Finding early customers and following signals from the market(20:46) GC AI’s first investors and fundraising lessons(23:34) Building a legal AI product around what in-house teams need(26:50) What an investor called violent product-market fit(27:10) Staying focused when a new model launch shakes the market(28:53) How GC AI fits into everyday business and legal workflows(33:48) What becomes possible when lawyers can delegate more of the work(38:13) AI, legal strategy, and creative thinking(39:53) From AI interfaces to agents that complete multistep legal tasks(42:57) Why legal agents need trust, accountability, and a record of their work(44:14) Who takes responsibility when an agent acts on your behalf?(47:52) What could a legal team look like in 2029?(49:12) The remaining gaps in trust, data protection, and judgment(50:18) Curiosity and becoming an AI boss(52:40) What’s next for GC AI?(53:21) What’s next for Cecilia and the team?(53:54) A woman who has had a profound impact on Cecilia’s life and careerFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Podcasts!Brought to you by: Rippling, Gradial, and Perkins CoieWX Productions 

In this episode of The Room Podcast, we speak with Sergi Bastardas, co-founder and CEO of Orbio AI, a platform using autonomous agents to coordinate the employee journey for frontline workforces. Sergi traces his path from a small town near Barcelona to joining Amazon Spain in its earliest days, and building a global supply-chain business. He shares how a family history of entrepreneurship, a curiosity about the wider world, and Amazon’s culture of operational excellence shaped the way he builds companies.He explains why HR is fundamentally a system of conversations and unstructured data and why Orbio was designed as an agent-native product rather than an AI add-on. From screening and scheduling to onboarding, retention, and exit interviews, Orbio’s agents work across voice, WhatsApp, text, and Teams while keeping people in control of employment decisions.In this conversation, Sergi breaks down the company’s expansion across languages and markets, what incumbent HR platforms would need to rebuild to become truly agent-native, and why he believes AI will empower frontline workers rather than simply replace them. He also reflects on fundraising, building a global company from Europe, balancing marriage with constant travel, and the coach who changed how he experiences the founder journey.Learn more about Sergi Bastardas on LinkedIn and explore Orbio AI.(03:52) Where Sergi grew up and how it shaped his worldview(06:48) Did Sergi always think he would become a founder?(08:13) Joining Amazon Spain in its earliest days(11:18) From Amazon and supply chains to the frontline workforce problem(13:04) Choosing experienced co-founders for Orbio(15:52) What AI-native HR means—and why agents are not a bolt-on(19:03) A day in the life of an Orbio agent(26:53) Launching first in Spanish and adapting voice agents across markets(28:07) Orbio’s Series A and the first investors who said yes(30:07) Building a global company from Europe(33:02) AI adoption across different countries and cultures(35:47) Voice infrastructure, LiveKit alternatives, and local-language models(37:29) Why incumbent HR platforms cannot easily replicate Orbio(40:07) The future of work with people, agents, and robotics(42:36) What’s next for Sergi personally(43:28) A woman who profoundly shaped Sergi’s life and careerFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music!Brought to you by: Rippling, Gradial, Ashurst Perkins CoieWX Productions 

In this episode of The Room Podcast, we speak with Caitlin Leksana, co-founder and CEO of Fazeshift, a company building AI agents for accounts receivable. Caitlin traces her path from growing up outside Washington, D.C., and studying mechanical engineering at Georgia Tech to water-desalination research, BCG, Harvard Business School, and entrepreneurship. She shares how a classmate became her co-founder, why their first crypto startup failed to find product-market fit, and how they found the conviction to pivot instead of walking away.Caitlin explains how her own messy invoicing spreadsheet led the team to hundreds of customer-discovery conversations and eventually to Fazeshift. She breaks down what changed when the founders started with a customer problem instead of a product idea, how Y Combinator helped them find momentum, and what she learned about fundraising, hiring, and building with a co-founder.The conversation also explores vertical AI in finance: why reliable automation starts with connecting messy data across ERPs, banks, inboxes, and industry-specific systems; how Fazeshift combines AI agents, browser workflows, and hands-on deployments; and why enterprise buyers want solutions to concrete problems rather than abstract AI capabilities. Learn more about Caitlin Leksana on LinkedIn and explore Fazeshift.(04:12) Where did Caitlin grow up, and how did it shape her worldview?(05:14) Did Caitlin always think she would become a founder?(07:01) Meeting her co-founder at Harvard Business School(09:00) Starting their first company and realizing it was time to pivot(11:17) How founders know when to stop pushing and change direction(12:40) Customer discovery and finding the accounts receivable problem(16:19) Building the MVP and joining Y Combinator(18:52) Finding Fazeshift’s first investors and navigating fundraising(21:50) Hiring mistakes and lessons from building after the pivot(24:10) How the Y Combinator community supported Fazeshift(29:43) Is Fazeshift a vertical AI company?(30:36) Selling the solution versus selling the AI(31:52) Why accounts receivable needed AI instead of traditional software(34:39) Who buys Fazeshift, and who uses it day to day?(36:54) Fazeshift’s product modules, pricing, and customer experience(39:16) Connecting data across ERPs, banks, inboxes, and customer systems(41:27) How Fazeshift deploys agents into messy enterprise workflows(43:50) Letting customer problems shape the product roadmap(45:07) Rebuilding browser agents as the technology improves(46:08) Partnerships with accounting firms, private equity, and industry software(48:38) How Caitlin’s role is changing as Fazeshift grows(49:47) A woman who has had a profound impact on Caitlin’s life and careerFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Podcasts!Brought to you by: Rippling, Gradial, and Perkins CoieWX Productions 

In this episode of The Room Podcast, we speak with Doug Tallmadge, co-founder and CEO of Gradial, a company building AI agents to help enterprise marketing teams execute work across their tools and campaigns. Doug traces his path from growing up in Cincinnati and studying physics at Dartmouth to joining SpaceX and helping build Starlink. He shares what working through seemingly impossible engineering challenges taught him and why finding a market for a startup can be a very different problem from building a product people already want. After ChatGPT launched, Doug and his longtime friends left their jobs to start a company. He explains how their broad ambition to bring AI into enterprise work became a focused business in marketing, where campaign briefs, creative assets, tickets, and manual handoffs create a significant operational burden. In this conversation, Doug breaks down how Gradial’s agents have evolved alongside the models, what it takes to earn enterprise trust, and why marketers should be able to plan their budgets without worrying about tokens. He also reflects on fundraising setbacks, managing the emotional swings of building a company, and the professor who backed an unconventional student project that helped launch his career.Learn more about Doug Tallmadge on LinkedIn and explore Gradial.(04:13) Where did Doug grow up, and how did it shape his worldview?(04:50) Did Doug always think he would become a founder?(05:40) Studying physics at Dartmouth and choosing SpaceX over finance(08:31) Building Starlink through existential engineering and business challenges(11:39) Discovering product-market fit versus engineering a product people already want(13:18) The moment Doug decided to start Gradial(14:39) Why enterprise work runs on tickets—and why Gradial focused on marketing(16:49) What goes into a marketing campaign, from brief to distribution(20:22) Starting a company with longtime friends and roommates(22:53) Narrowing the product after the first weeks of disillusionment(24:18) Gradial’s first investor and lessons from early fundraising(29:41) Managing the emotional swings of the founder journey(31:53) When the original plan failed and it was time to focus on marketing(34:19) How Gradial’s agent architecture evolved alongside the models(38:02) Better models and the future of long-duration marketing work(41:23) Enterprise security, procurement, and integration with existing systems(42:37) Helping marketers spend more time on creative work(43:36) Pricing AI agents and making marketing budgets predictable(48:15) What’s next for Gradial(49:36) What’s next for Doug outside the company(49:58) A woman who has had a profound impact on Doug’s life and careerFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music!Brought to you by: Rippling, Gradial, Ashurst Perkins CoieWX Productions

In this episode of The Room Podcast, we sit down with Ann Crady Weiss, co-founder and CEO of Hatch, to trace the path from her Silicon Valley upbringing and early career in law and technology to building one of the most recognizable names in sleep. Ann shares how her family’s connection to Intel shaped her entrepreneurial ambition, what she learned at Wilson Sonsini and Yahoo, and how the acquisition of her first startup prepared her to build again.Hatch began with a personal problem and evolved from a baby-focused hardware company into a broader sleep platform for children and adults. Ann explains how customer behavior led the team into the adult market, why emotional design and organic love became central to the brand, and how Hatch balances slower hardware cycles with a software-led organization and a growing library of bedtime content.The conversation also covers Hatch’s appearance on Shark Tank, the long relationship with True Ventures, an early move into Costco, and the discipline of building with real unit economics. Ann shares what she has learned as both a founder and venture partner, why timing and luck matter more than entrepreneurs often admit, and why Hatch emphasizes better sleep habits, phone-free routines, and contactless tracking rather than scores or constant monitoring—even as AI changes how products are built.Learn more about Ann Crady Weiss on LinkedIn and check out Hatch on their website.(04:13) Ann’s Silicon Valley childhood and family connection to Intel(05:05) How Ann’s uncle joined the early Intel team(05:35) Did Ann always think she would become a founder?(06:54) How law, Wilson Sonsini, and Yahoo shaped Ann’s early career(08:59) The acquisition of Ann’s first startup and the path to Hatch(10:45) The personal need and original product behind Hatch(12:47) Why Hatch expanded from baby products into the adult sleep market(14:53) How founders can pursue a large opportunity without losing focus(16:22) Building emotional connection through product and brand design(18:14) What founders should understand about hardware iteration(19:30) The Hatch Sleep Clock and a different approach to sleep tracking(22:47) Pitching on Shark Tank and partnering with Chris Sacca(26:03) True Ventures and the first investors to believe in Hatch (29:27) Setbacks, Costco, and lessons from expanding too early(32:09) Building a sustainable business with real unit economics(34:05) Hatch+ content and the case for phone-free sleep(37:37) Building customer trust and protecting bedroom privacy(39:29) How Hatch differs from wearables and other sleep trackers(41:05) What Ann has learned as a founder and venture partner(44:24) The future of sleep technology in the age of AI(46:47) Building products that make people feel calmer, not monitored(48:17) What’s next for Hatch(49:33) A woman who profoundly shaped Ann’s life and careerFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music!Brought to you by: Rippling, Gradial, Ashurst Perkins CoieWX Productions

In this episode of season 15 of The Room Podcast, we speak with Gavriel Cohen, co-founder and CEO of NanoCo and creator of NanoClaw, an open-source framework designed to make powerful AI agents safer and more practical for real-world use.NanoClaw emerged from Gavriel’s own experience using AI agents to run an AI-native marketing agency. After connecting an agent he called “Andy” to his company’s sales pipeline, Gavriel saw the potential for agents to operate less like chatbots and more like persistent digital coworkers—but also recognized the security challenges that come with giving autonomous agents access to sensitive tools, data, email, and business systems.In this conversation, Gavriel shares the core idea behind NanoClaw: rather than weakening an AI agent, give it powerful capabilities while putting clear boundaries around what it can access—what he describes as putting the agent “in a box.” He also explains why he believes the future of work will involve humans becoming managers of increasingly capable teams of agents, rather than simply replacing people with AI.Learn more about Gavriel Cohen on LinkedIn and explore NanoCo.(04:08) Gavriel’s childhood and how it shaped his worldview(05:33) Did Gavriel always think he would become a founder?(06:40) Moving from New York to Israel and studying at Tel Aviv University(08:08) The aha moment behind NanoClaw and Gavriel’s entry into AI agents(12:52) Why the AI agent market was getting security wrong(15:32) What is a “Claw” and how do AI agents fit into everyday work?(20:03) The most surprising ways people are using NanoClaw(23:50) Raising $12 million and finding NanoCo’s first investors(37:34) Why Gavriel turned down acquisition interest and kept building independently(40:18) NanoClaw’s architecture decisions and what still isn’t ready for primetime(45:13) A woman who has had a profound impact on Gavriel’s life and careerFor The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music!Brought to you by: Rippling, Gradial, Ashurst Perkins CoieWX Productions

In this season 15 opener of The Room Podcast, we speak with Russ d'Sa, co-founder and CEO of LiveKit, a platform for building real-time voice, video, and AI agents. LiveKit began as an open-source project for live audio and video. As developers adopted it, the team built cloud infrastructure to help them scale. A second turning point came when OpenAI used LiveKit to build a voice interface for ChatGPT, revealing how technology built to connect people could also connect people with AI.Russ traces his path from growing up around computers in Canada and Silicon Valley to founding a startup, working at 23andMe and Twitter, and building LiveKit. He shares the unexpected relationships and product decisions that shaped the company, including why he calls it the “accidental AI company.” In this conversation, Russ explains why speaking with an AI requires a different kind of infrastructure from typing to one. He also looks ahead to physical AI and reflects on how becoming a father has changed the way he thinks about success and his values.Learn more about Russ on LinkedIn and explore LiveKit.(03:54) Where did Russ grow up, and how did it shape him?(06:42) Did Russ always expect to become a founder?(07:30) What screen did Russ watch the documentary on?(09:09) Has Russ read *Tomorrow, and Tomorrow, and Tomorrow*?(09:17) What was Russ’s first Y Combinator company?(09:24) Was Russ's company a gaming company?(11:13) What year did Russ start that company?(11:40) Did Russ join Twitter after his first startup?(11:51) How did Russ make the transition to 23andMe?(13:22) How did 23andMe eventually lead Russ to Twitter?(16:51) Was Russ's Twitter role in software or content?(22:45) What was the original idea behind LiveKit?(30:17) Did LiveKit finally arrive at the right time?(30:37) Who was LiveKit’s first investor?(35:35) What has gone differently than planned while scaling LiveKit?(37:34) Where does LiveKit fit between AI models and the software built around them?(43:07) What makes building voice agents different from building text agents?(47:51) What still prevents AI voice conversations from feeling fully natural?(52:08) What is Russ excited about on LiveKit’s roadmap, and what can he share?(55:43) Which woman has had a profound impact on Russ's life and career?For The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on InstagramFollow us on TikTokCheck out our guide to podcasting here!Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music!Brought to you by: Rippling, Gradial, Ashurst Perkins CoieWX Productions

In this bonus episode of The Room Podcast, we speak with Jai Das, President, Partner and Co-Founder of Sapphire Ventures, one of the world's leading enterprise venture capital firms backing category-defining B2B software and AI companies. Sapphire Ventures has invested in companies including monday.com, Box, Square, MuleSoft, Fitbit, ExactTarget, and many others, while building one of the industry's most recognized enterprise-focused venture platforms. Before co-founding Sapphire Ventures, Jai's career spanned software engineering at Oracle, founding startups during the dot-com era, corporate venture investing, and leading the transformation of SAP Ventures into the independent firm it is today. In this conversation, Jai reflects on the lessons learned across multiple technology cycles, from the early internet to today's AI revolution, sharing how his experiences as an engineer, founder, and investor shaped Sapphire's long-term strategy and investment philosophy. Learn more about Jai on LinkedIn - https://www.linkedin.com/in/jaidas Explore Sapphire Ventures on their website - https://sapphireventures.com/ (05:38) Jai's upbringing in India and journey to Silicon Valley (08:20) Did Jai always think he would become a founder (09:37) What was it like being an engineer at Oracle in the 1990s? (11:51) Did Oracle miss its opportunity in cloud computing? (13:10) Jai’s startup journey after Oracle and the path to SAP Ventures (17:53) How did Jai’s first venture investing experience shape your career? (18:50) What was early corporate venture capital like at Intel and SAP? (20:57) How should founders think about taking money from corporate VCs? (23:33) How did SAP Ventures become Sapphire Ventures? (25:06) How did Sapphire differentiate itself in a crowded VC market? (28:19) When did Sapphire experience its biggest growing pains? (32:40) Who was Sapphire Ventures' first LP? (34:18) Why did Sapphire build a fund-of-funds strategy? (35:59) What was the first portfolio company that believed in Sapphire? (37:37) Why has the developer become such an important buyer? (39:58) How has AI changed who the modern "developer" really is? (41:50) How does the rise of AI agents influence your investment thesis? (44:42) How does venture capital stay relevant in the age of AI? (48:11) What is Jai most excited about right now? (49:37) Who is a woman that has had a profound impact on Jai’s life and career? For The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on Instagram Follow us on TikTok Check out our guide to podcasting here!  Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music! WX Productions

In this bonus episode of The Room Podcast, we speak with Packy McCormick, founder of Not Boring and Not Boring Capital, an investment firm focused on backing ambitious technology companies tackling large, complex industries. Through his writing, investing, and storytelling, Packy has built one of the most influential platforms in venture capital, helping founders explain difficult technical businesses to customers, talent, and investors. Not Boring began as a newsletter during the early months of the pandemic and has since grown into a media platform reaching hundreds of thousands of readers, alongside an early-stage venture fund investing in companies across energy, aerospace, infrastructure, AI, and enterprise software. Rather than chasing trends, Packy focuses on founders building category-defining businesses that require deep conviction and long-term thinking. In this conversation, Packy shares how storytelling became his competitive edge as an investor, why the best founders think several moves ahead of everyone else, and how developing a unique perspective can create opportunities that traditional venture firms often overlook. Learn more about Packy on LinkedIn: https://www.linkedin.com/in/packymccormick/ Explore Not Boring at https://www.notboring.co/ (04:11) Packy's upbringing and entrepreneurial family (05:01) Did Packy always think he would become a founder? (05:50) Early career in finance and startups (09:16) Starting Not Boring during the pandemic (12:48) The first article that proved Not Boring was working (14:04) Growing the newsletter and launching Not Boring Capital (15:57) The thesis behind Not Boring Capital (17:23) The first investor to back Not Boring Capital (17:48) What Packy looks for in exceptional founders (21:57) Building a venture firm around a personal platform (23:56) The future of venture capital (26:36) How storytelling creates an investing advantage (28:45) Investing in complex businesses (31:08) An investment that didn't go as planned (33:19) How AI is changing writing and investing (35:23) Advice for building an audience today (37:05) What's next for Not Boring (38:20) A woman who had a profound impact on Packy's life and career For The Room Podcast in your inbox every week, subscribe to our newsletter.Follow us on Instagram Follow us on TikTok Check out our guide to podcasting here!  Don't forget to subscribe to our channel on YouTube, Spotify, and Apple Music! WX Productions