Superbase in the Age of AI Builders

Overview

In this episode of The Light Cone, Superbase co-founder and CEO Paul Coppstone discusses how AI agents are changing software development, how Superbase is adapting, and why more people may become builders and entrepreneurs.

Superbase’s Growth and Shift to Agents

• Superbase reports 13 million signups, up from roughly 3 million a year earlier.
• It creates more than one million databases per week; databases under management have grown from about 8 million to nearly 50 million.
• Revenue is expected to grow around 4× this year.
• The company says agents account for roughly 70% of new activity.
• Superbase has evolved from a dashboard-first product, to an API-first platform, and now to a code-first product built for agents and CLI workflows.
• Its open-source foundation, support for the Postgres ecosystem, and agent-friendly error messages help agents use it and recover when something goes wrong.

A New Kind of Builder

• AI tools are enabling people without software-engineering backgrounds to build useful applications.
• Domain experts can turn long-standing ideas into products without needing to persuade an engineer to join them.
• Examples include a Brazilian taxi driver who built a local ride-hailing app on his phone, and a former fintech founder building personalized finance software for himself.
• Superbase sees expanding access to building as a major source of entrepreneurship.

Personal Software and Personal Data

• Coppstone describes building a personal “control tower” for email, finances, health, workouts, and other life information.
• A personal database lets agents combine context from different sources—for example, relating changes in weight to travel or identifying transactions using personal notes.
• Keeping data in a database the user controls can make it easier to use different agents and applications without locking personal information into one tool.
• File uploads can also be stored and interpreted by an agent, which determines their relevance and what to do with them.

Scaling for Many More Databases

• Superbase expects AI could bring vastly more people into software building, creating a need to serve a much larger number of small projects and databases.
• Its strategy is to support both ends of the market:
• Multigresses for projects that need to scale toward petabytes.
• Superbase Spark, based on an architecture compatible with Postgres and SQLite, for prototypes and smaller workloads.
• The goal is to let projects start simply and scale without needing to leave the platform.

Building and Testing for Agents

• Superbase evaluates how different models work with its platform and with Postgres.
• The company is improving error messages so they explain how to fix problems, rather than merely reporting them.
• Coppstone says a key is to win twice: first, be selected by an agent; then, help it recover when it gets stuck so it does not switch to another tool.
• Internally, engineers can choose their AI tools freely. Adoption has increased as models improve, alongside more emphasis on testing and evaluation.

Advice for Founders

• Build in durable categories: Coppstone views databases as a long-lasting foundation, even as AI tools change.
• Adapt quickly rather than assuming the future is predictable.
• Use your own product, partner with important ecosystem players, and preserve team capacity to test new opportunities.
• Not every experiment will work, but successful ones can help a company ride a new wave.

Main Takeaway

AI is lowering the barrier to building software. Superbase’s bet is that individuals, domain experts, startups, and agents will all need accessible, flexible databases—and that the next generation of builders should be able to create and customize tools for themselves.