Loose, opinionated hallway talk about AI products and craft.
Nabeel and Fraser trade notes on AI tools and the people using them, from Claude Code and parallel agent orchestration to dentists vibe-coding their own apps. The through-line is product judgement: when frustration is a product problem rather than a model problem, and why craft still matters.
PRESS PLAY
Hallway Chat is a running conversation between Nabeel and Fraser about AI products: the ones they use, the ones that delight them, and the ones that fall short. The register is casual and quick. Thoughts trail off, jokes arrive mid-argument, and more than one episode ends on some version of the same question: should we be done? The hands-on material is specific. One conversation opens with one of the pair deep in a Claude Code rabbit hole, building hundreds of little apps with no room left for new television. The same episode turns to Conductor, which runs many Claude Code and Codex agents in parallel, and to Charlie at Conductor's tweet inviting users to prompt a feature rather than request one. Another includes a self-imposed rule: build an app only through prompts, never reading the code, to learn what the experience feels like for someone who has never coded. When a login flow breaks, the takeaway is that code editors show the visual layer but not the logic underneath. People outside the tech world get attention too, notably dentists in Australia building their own custom apps, which prompts a joking pitch to call an episode 'What Is It With Dentists and Vibe Coding?' Opinions are firm and argued in the open. When OpenAI released a prompting guide for GPT-5, the verdict is blunt: the broadest horizontal product since Google should not make users relearn their technique, so this is a product problem, not a model problem. Elsewhere the mood turns reflective, setting Steve Jobs as a soulful product humanist against a current crop of tech heroes described as dealmakers or people who can summon AI gods from Common Crawl. Craft keeps resurfacing. One conversation pictures a 1930s Paris for AI makers, with builders egging each other on like Picasso and Matisse, and contrasts it with incubators that ask about weekly growth rather than whether the product got better. It offers a pointed definition of an AI wrapper as a bet on the founder's intuition rather than on a better model, and finds a lesson in thoughtful design in a backpack's keychain. Andrew Mason from Descript joins one episode to talk about Airtable's AI features, the difficulty of open world agents, and teaching software by showing it examples instead of writing formulas. Startup business is never far away, with asides on handling a decision honestly and confidentially after diligence, on why some founders dislike SPVs, and on the case for being the ubiquitous default in dev tools and databases. Long AI memory and privacy get serious attention as well. Beneath the critique sits real enthusiasm, down to a closing story about hearing technology that carries no stigma and a reminder that the rate of change is easy to lose track of. For coding agents, 'The two types of Agentic Engineering, and their teams' is a natural first listen; for design and taste, 'Making the perfect backpack. Craft in the age of AI.' fits better.
Made for: Suits a listener who already experiments with tools like Claude Code or Codex and wants candid product judgement on AI software, with occasional talk about founders and startup funding mixed in.
What sets it apart: It judges AI through a product lens rather than a research one, openly arguing that a frustrating model is a product problem, and pairs hands-on tool talk with reflections on craft and the loss of product-minded tech heroes.
Fraser & Nabeel explore what it means to build great products in this new world of AI.Two former founders, now VCs, have an off-the-cuff conversation with friends about the new AI products that are worth trying, emerging patterns, and how founders are navigating a world that’s changing every week. Fraser is the former Head of Product at OpenAI, where he managed the teams that shipped ChatGPT and DALL-E, and is now an investor at Spark Capital. Nabeel is a former founder and CEO, now an investor at Spark, and has served on the boards of Discord, Postmates, Cruise, Descript, and Adept. It's like your weekly dinner party on what's happening in artificial intelligence.
Dentists are vibe coding; Nabeel and Fraser make you wonder who gets to build software next.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A grounded look at long-running agents, one-shot apps, and prompt-driven feature requests.
Agentic engineering comes through as a practical shift in how software is made. The hosts draw a useful line between long-running agents and one-shot app building, then ground it in a multiplayer product that had once looked like a ten-person, seven-or-eight-month effort and…
From dentists to captioning glasses, the episode asks whether technology still serves human experience.
Hallway Chat uses a loose conversation about vibe coding to reach a larger question: what technology is supposed to do for people. It moves from dentists making custom apps to a wider contrast between older tech heroes, with Steve Jobs held up as a soulful product humanist, and…
A sharp conversation about who should adapt when AI products keep changing.
Nabeel and Fraser look at a recurring cost of AI's rapid product cycle: every three to six months, new capabilities can force users, startups, and CEOs into another round of reorientation.
Airtable makes the bigger shift feel concrete: teaching AI through examples, outputs, and user habits.
Hallway Chat uses Airtable to open up a more interesting question: how AI products should learn from people. Nabeel, Fraser, and Andrew Mason of Descript move from Airtable's place in a side project around a physical board game club to the promise of showing an AI a few finished…
A magnetic keychain opens a thoughtful case for the human craft AI products still need.
A stolen backpack leads into a sharp, tactile argument for why AI products still need human craft. The Boundary Backpack gives the episode its clearest anchor, especially when its magnetic keychain turns a small irritation into a quietly elegant fix.
Frontier technology, seen at civilizational scale.
Water tech, with teeth and numbers.
Consumer brand strategy without the gloss.
Southern startups, capital, and conviction.
["Hallway Chat's Nabil and Fraser argue that customers are moving from feature requests to prompts, and Claude Code is increasingly the thing that turns those prompts into code.", "Fraser says Camp began as a fork of Chorus and then swapped in Convex so real-time multiplayer work was easier.", "Gemini becomes the most vivid example: Fraser uses it as a tailor for shirts, sending photos back when the first size looked wrong.", "The episode closes on engineering orgs, with a two-person pod, AI fluency, and CEO-level example treated as the new baseline."]
Hallway Chat centers on dentists as an unexpected vibe-coding market, with Nabil arguing that a solo dental practice has the overhead, cash flow, and software pain to adopt custom tools.
Hallway Chat centers GPT-5 and Sonnet 3.7 as a product-design problem, not a prompting puzzle, with Nabeel and Fraser arguing that tool use and code execution now matter more than model folklore.
["Andrew Mason argues that Airtable's AI works because it puts chat on the left and the spreadsheet on the right, making the agent visible instead of hiding it in a help bubble.", "Nabeel says he and Andrew are using Airtable for a physical board game club, which gives the discussion concrete examples like reservations, members, and a 50-page PDF import.", "The conversation contrasts 'open world agents' with 'agents on rails,' then uses Notion AI, review states, and nested history to show why trust and auditability matter.", "The episode closes on Descript's 'AI gene therapy' idea, where people move from skeptical to converted to rewired through repeated use rather than a one-day training."]
Hallway Chat's "Making the perfect backpack.
Hallway Chat hosts Nabil and Fraser debate whether the best AI models will become broadly affordable within five or six years, as smartphones did for many users.
Hallway Chat examines AI and games through GDC observations, including an AI dinner that grew from four or five attendees to 25.
Hallway Chat examines why AI-app experimentation lacks the shared discovery systems that shaped Facebook and mobile development.
Hallway Chat hosts Nabil and Fraser debate whether an AI product that is 10% better can eventually beat ChatGPT's 400 million weekly active users.
Hallway Chat hosts Nabil and Fraser interview Granola CEO Chris Pedregal about building an AI note-taking product after Socratic was acquired by Google and reached roughly four billion annual queries.
Two VCs discuss the challenges of building AI products and review a new AI note-taking app called Granola.
Two former founders and VCs discuss the rapid commoditization of AI models and the points of competition in the AI product landscape.
Fraser and Nabeel interview Josh from the Browser Company about building consumer AI products, focusing on the importance of tiny interaction details and the craft behind ArcSearch.
Two former founders turned VCs discuss the state of startups and AI products in 2020, reflecting on their long hiatus and the importance of commitment.
Two VCs interview Fred Wilson about his discovery of social media and discuss AI product trends.