CHAPTERS
How big v0 is: user growth and real deployments
Tom shares concrete adoption metrics for v0 and emphasizes that many outputs go beyond prototypes into deployed, real apps. The conversation frames v0 as a catalyst for “personal software,” including for non-engineers.
Live build #1: “Better LinkedIn feed” prompt → PRD-like plan and multi-file app
They start with a simple prompt to create a better LinkedIn newsfeed and watch v0 generate a full plan and project structure. The output resembles a lightweight PRD plus a multi-file implementation, showing how quickly v0 expands vague intent into a feature-rich scaffold.
Live build #2: Screenshot-to-app cloning with an Apollo-style homepage
They switch to a screenshot-driven workflow using an Apollo.io page to drive visual fidelity. The goal is to customize a SaaS homepage based on which product a user uses (e.g., meetings vs emails), and v0 mirrors styling while changing information hierarchy.
Iteration strategy: atomic changes, rollback, and “pull request” thinking
Tom explains how he prefers small, atomic iterations—similar to pull requests—so changes are easy to verify and revert. They highlight that v0 can also handle multiple changes at once, but incremental steps can reduce risk and improve control.
Under the hood: multiple models + Vercel’s “online model” that fixes LLM code
Tom describes v0’s approach to hiding model complexity and creating a guided “happy path.” He explains that raw LLM code often won’t run, so v0 uses additional layers and multiple models to iteratively correct and operationalize outputs.
From prototypes to usable apps: fixing routes, runtime errors, and exploring generated features
They explore the generated LinkedIn-like app and encounter missing routes/errors. Tom demonstrates interactive debugging by pasting the error and asking v0 to implement the missing route, showing a tight build-test-fix loop inside the tool.
Marketplace integrations: databases and AI providers in a few clicks
Tom introduces integrations that connect v0 projects to Vercel Marketplace services without heavy setup. He explains how users can attach databases (e.g., Neon) and AI providers, reducing friction to build functional, data-backed applications.
Adding an AI feature live: Groq-powered post composer + troubleshooting API keys
They attempt to add an AI-assisted post composer to the LinkedIn clone using the Groq integration. The demo hits issues (invalid API key, module export errors), and Tom shows how v0 plus version restore can help recover from breaking changes.
Tom’s personal software: feed filter tool, one-shot sneaker store, and “neighborhood button” app
Tom shares examples of what he builds as a CPO who doesn’t have hours to code anymore. He demonstrates real utility: internal tools, a one-prompt e-commerce store with realtime filtering, and a simple mobile-friendly app that tells you what neighborhood you’re in.
Workflow and scaling: browser creation vs IDE “eject,” Git-style iteration, and real infra
They discuss how teams take v0 output into editors like Cursor/Windsurf and the desire for a smoother back-and-forth workflow. Tom argues deployed outputs are “real software” backed by Vercel’s infrastructure, capable of scaling massively; the limiting factor is product value, not tooling.
Database demo end-to-end: Neon store, mock meetings, env vars, and a creation form
They complete the database loop by adding a Neon integration to store upcoming meetings, seeding mock data, and wiring the app to fetch it. They handle a missing NEON_DATABASE_URL env var and observe that v0 even generated a meeting creation form, though some flows remain unimplemented.
How v0 is built and staffed: a tiny team, Vercel leverage, and startup-within-a-startup
Tom explains the team structure behind v0 and why it can move fast. v0 benefits from Vercel’s existing foundations—accounts, marketplace, and deployment infrastructure—allowing a small team to ship quickly and focus on the core experience.
Vercel’s product operating system: product areas, frequent reviews, Notion→Linear, no rigid OKRs
Tom outlines how Vercel organizes into product areas with cross-functional leads owning roadmaps. Planning is reset roughly every six months but guided by frequent reviews; they use Notion for initiatives and Linear for execution, and they avoid heavy OKR processes in favor of “lowercase-a agile.”
How Vercel uses v0 internally: dogfooding, component generation, and demos across the company
Beyond prototyping whole apps, teams use v0 to generate smaller UI slices (e.g., a settings drawer) and then import them into production code. Tom describes a cultural shift where demo days now feature non-engineering teams showing real apps, like interactive TCO calculators built by account executives.
PM in the AI prototyping era: prototypes complement PRDs, and strategy matters more
They discuss how prototyping changes product work: PRDs can become thinner and more strategic while demos validate hypotheses earlier. Tom warns against a “feature factory” mindset—shipping is cheaper, but cohesion, prioritization, and problem framing remain core PM responsibilities.
Competitive landscape and React’s role: customer focus, framework as implementation detail, and role convergence
Tom explains his philosophy on competition—focus on customer value rather than chasing rivals’ feature sets. He reflects on React’s ubiquity (and training data advantages) while predicting more hybrid roles: design engineers and PM-engineers enabled by AI tools that reduce the “tax” of parts of the job.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome