AI-Native Analytics for the Supabase Era

Every modern product team needs three things to ship: a database, a frontend, and answers. The first two have been rebuilt for the AI-native era. Analytics has not.

For most early-stage teams, getting a straight answer about their own product is still a weeks-long project. Someone asks a simple question. What follows is the same broken sequence: instrument the app, pipe the data somewhere, build a warehouse, hire someone to make sense of it all. By the time the dashboard ships, the question has already changed. The answer was sitting in the production database the entire time.

Analytics done the old way is event-based, and event-based is backwards. Teams shipping in 2026 are done paying the legacy BI tool tax.

Supabase has made this especially clear. Four years ago, Felicis invested in Supabase (opens in new tab), believing that Postgres-as-a-service would become the default backend for modern software. It has: more than 7 million developers, over a million active databases, and an ecosystem powering the first generation of AI-native app builders like Lovable, Bolt, and the long tail of founders shipping products without standing up their own infrastructure. These teams are not paying for Amplitude. They are not hiring data engineers. They are building fast and need answers at the same speed.

Enter Dreambase.

Dreambase is led by two founders who know this problem inside and out. Andy Keil (opens in new tab) spent a decade inside early-stage B2B SaaS as a zero-to-one product leader, most recently as Head of Product at QuotaPath, where he felt how punishing the old analytics sequence is for a small team. Kyle Ledbetter (opens in new tab) spent more than a decade leading design and UX at Teradata, MicroStrategy, and eBay, watching enterprise analytics get heavier, slower, and further from the person trying to get an answer. The company they built is the exact inversion of the last twenty years of the category. I was introduced to Andy by Scott Buxton, Supabase's CFO. Scott told me what I later saw for myself: the Supabase ecosystem needed what Andy and his co-founder were building, and Supabase wasn't going to build it themselves.

Dreambase is a featured integration in the Supabase marketplace. Connect a Supabase project and Dreambase reads the schema, generates production-ready dashboards, and returns the first report in under 90 seconds. No tracking libraries, no ETL jobs, no warehouse. Just the database, an AI-native semantic layer the team calls Topics, and an Analyst Agent tuned for Postgres that takes a plain-English question and returns an accurate query without a SQL cursor in sight. Supabase Report Cards, free for every project, assess security, performance, and architecture on day one. Paid tiers scale from there.

For the founder who needs to know how the new feature is doing, Dreambase is the difference between analytics and flying blind. Schema-first, agent-assisted, and free of instrumentation debt. The work that used to require a full data team now happens in the background.

Felicis is proud to lead the seed in Dreambase, alongside Active Capital, FirstMile Ventures, Darkmode Ventures, Angel Collective, Earl Grey Capital, and Mercury Fund. Strategic angels from Supabase, Perplexity, Cloudflare, Expo, and Reforge have also invested, an unusually specific group of operators that signals exactly where the ecosystem is headed.

Every AI-native company needs an intelligence layer sitting directly on top of its production data. The teams that start from the right place in the stack, the database not the event stream, will move faster, ship more, and understand their customers better than those who retrofit. Supabase solved the database. Dreambase is how the database answers back.