DiscussionInsightful

Why Everyone Is About To Have Their Own Database

Y Combinator

Paul Copplestone, CEO of Supabase, discusses how the platform has evolved from a dashboard company to an API-first company and now to a code-first company optimized for AI agents. With 13 million users and 1 million databases created weekly, Supabase is experiencing explosive growth driven by both software engineers and non-technical builders who can now create sophisticated applications with AI assistance.

Summary

Paul Copplestone of Supabase discusses the platform's dramatic evolution and explosive growth over the past year. The company has grown from 3 million to 13 million signups, with databases under management growing from 8 million to nearly 50 million. This growth is driven by two major factors: abandoned software engineers building with AI, and a new category of non-technical builders empowered by AI tools.

The conversation explores Supabase's three eras: first as a dashboard company (built in 100 days during YC with an Airtable-like UI on top of Postgres), then as an API company (when customers needed programmatic database creation), and now as a code-first company optimized for AI agents. Approximately 70% of new activity comes from agents, a dramatic shift that required rearchitecting the CLI and infrastructure-as-code approach.

Copplestone describes the changing profile of builders on the platform. Historically, startups required technical co-founders, but now non-engineers with domain expertise can build sophisticated applications—a fintech founder creating personal finance software, a 50-year-old taxi driver in Brazil building a mini-Uber app. The key insight is that knowing what to build matters more than being able to build it, which was previously the bottleneck.

The discussion covers how Supabase is preparing for massive scale, potentially needing to handle 1.2 billion new databases monthly (extrapolating from ChatGPT's user base). To achieve this, they've acquired Temso to provide SQLite-compatible architecture via Supabase Spark for prototyping workloads, while Postgres handles scaling to petabytes. This allows them to serve both ends of the spectrum simultaneously.

Copplestone shares his personal experience building internal tools (called Control Tower) combining finance, health, email triage, and CRM in a single database. This reflects how databases are becoming stores of record for personal life data that agents can access and correlate. He emphasizes that error messages must be designed for agents to self-remediate rather than requiring human intervention.

The conversation also features insights from Gary Marcus on building Gbrain (retrieval-augmented generation system) on top of Supabase and Postgres with PgVector for graph and semantic search capabilities. Marcus describes how agents can autonomously improve performance through eval loops and how these primitive approaches may evolve into general frameworks.

Copplestone discusses internal adoption at Supabase: 60% of engineers now write 80% of their code with AI agents, up from 10% six months prior. This cultural shift required robust testing practices and eval suites to ensure code quality. The company is investing more in testing and agent-specific use cases than traditional feature development.

The conversation touches on market dynamics where traditional vertical SaaS founders struggle to raise Series A funding despite reaching $5-10M ARR, while many can now achieve $50M+ revenue with 15 people profitably. This is changing venture capital dynamics and allowing builders to remain profitable without external funding.

Copplestone emphasizes that building in durable categories (databases have existed 50 years and will for 50 more) and maintaining adaptability are critical. The pace of change is accelerating—each new model (Opus, o1, etc.) creates inflection points. He stresses dogfooding, maintaining team capacity for experimentation, and partnership with key ecosystem players. Many experimental bets fail, but the successful ones enable riding major waves.

Key Insights

  • 70% of Supabase's new activity comes from AI agents, not human developers, representing a fundamental shift in who the primary users of database infrastructure are.
  • Non-technical builders with domain expertise (like a taxi driver building an Uber competitor or fintech founder creating personal finance software) are now creating sophisticated applications because knowing what to build matters more than being able to build it.
  • Supabase's engineering team went from 10% to 60% of engineers writing 80% of their code with AI agents in just six months, driven by convincing evidence that models produce better work, requiring corresponding investments in testing rather than traditional feature development.
  • Error messages must be redesigned for agents to self-remediate by providing specific instructions on how to fix problems, rather than generic errors, preventing agents from switching to competing platforms when stuck.
  • Vertical SaaS founders struggling to raise Series A at $5-10M ARR while simultaneously achieving $50M+ revenue with 15 people profitably represents a fundamental market shift where external funding is becoming unnecessary rather than essential.

Topics

AI agents as primary users of developer infrastructurePlatform evolution: dashboard → API → code-first architectureDemocratization of software building for non-technical domain expertsDatabase scale requirements (billions of databases per week)Error messages and developer experience redesigned for agentsPersonal databases as stores of life records for agent accessTesting and evaluation frameworks for AI-generated codeStartup economics: profitability without Series A fundingBuilding in durable categories and maintaining adaptabilityDogfooding and partnership strategy for riding technology waves

Transcript

[0:00] I'm genuinely excited about what the future could be for a lot of people. The thing that I really love about Superbase is that like I see a lot of entrepreneurship um because of the platform. I certainly think that it'll be the case that everyone in the world with access to AI will become a builder of some sort. Like you can build anything, you can customize it. You want to make tweaks to some app, some UI, then you don't have to be beholden to to someone. I'm pretty pumped about 2027 actually. [0:37] Welcome back to another episode of the light cone. Today we're joined by Paul Coppstone, co-founder and CEO of Superbase, the open-source Postgress…

Full transcript available for MurmurCast members

Sign Up to Access

More from Y Combinator

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.