Paul Copplestone

Paul Copplestone is the co-founder and chief executive of Supabase, the open-source application platform built around PostgreSQL. He has helped make a decades-old relational database central to modern developer infrastructure, including authentication, realtime applications, semantic search, and AI-assisted software development.

Before Supabase, Copplestone founded startups including ServisHero and Nimbus For Work. In December 2019, he pitched Ant Wilson on building an open-source database company in a Singapore coffee shop. They interviewed developers the following January, began developing the product in February, and joined Y Combinator’s Summer 2020 batch. When the pandemic disrupted their relocation plans, they built Supabase from Singapore as a remote-first company and raised seed funding that August, a founding sequence Copplestone recounted publicly.

Supabase combined managed PostgreSQL with authentication, generated APIs, realtime updates, storage, and server-side functions, offering developers an open-source alternative to Firebase without sacrificing SQL or data portability. Its expansion into AI accelerated after an outside contributor integrated pgvector, an existing PostgreSQL extension created by Andrew Kane. Copplestone subsequently collaborated with that contributor on Supabase Clippy, a documentation assistant that stored embeddings in Postgres, retrieved relevant passages, and supplied them to a language model.

• Keep embeddings with operational data. Storing vectors alongside application records can reduce database round trips and extend PostgreSQL’s row-level security for AI retrieval to user-specific data. Copplestone demonstrated how triggers, range partitions, and selective indexing can also classify incoming images and isolate irrelevant content within the database. He considers specialized vector databases useful for some workloads, without treating them as mandatory infrastructure for every AI application.

• Evaluate the entire retrieval system. Copplestone has contrasted IVFFlat with HNSW vector indexing, showing improvements in throughput and retrieval accuracy while cautioning that isolated benchmarks omit network requests, access controls, and workload-specific tradeoffs. His Postgres vector-search demonstration also explored Citus-based sharding for larger embedding collections.

• Defend open infrastructure through portability. His case for open data standards connects PostgreSQL, OpenTelemetry, and Apache Iceberg through permissive licensing, practical self-hosting, and vendor-neutral governance.

• Extend Postgres at infrastructure scale. Copplestone introduced Multigres, a scaling initiative led by Vitess co-creator Sugu Sougoumarane, addressing pooling, availability, and horizontal sharding. He also opened the relevant OrioleDB patent under a nonexclusive license covering proprietary forks.

By April 2026, Supabase had reached 100,000 GitHub stars, reflecting the scale of a platform that extends existing open-source database infrastructure instead of replacing it.