Supabase co-founders Paul Copplestone and Ant Wilson joined swyx to explain how the company is rethinking backend infrastructure as AI agents generate vastly more software. Recorded ahead of Supabase Select 2026, the conversation examined product work, strategic acquisitions, and operational changes aimed at supporting long-running processes, agent-friendly schemas, and a much larger universe of applications.
Key products and moves
-
Supabase Compute: an approach for running long-lived workloads alongside the database. Compute assumes AI agents and automation will create many continuous tasks that benefit from being close to data and storage.
-
Multigres: a highly available Postgres infrastructure project currently running in private alpha. Multigres focuses on improving Postgres reliability and availability for production-scale use.
-
OrioleDB and the Postgres storage engine: the discussion referenced work on storage-layer innovations within the Postgres ecosystem, including OrioleDB.
-
Supabase acquires Turso: Supabase acquired Turso to gain scale-out capabilities. The acquisition is positioned as a way to broaden the company’s approach to scaling and distribution.
The interview was recorded before Supabase Select 2026.
Why this matters
AI-driven agents will change the shape of backend needs in several ways:
- computation needs to be closer to data to reduce latency for many concurrent, persistent tasks;
- databases must scale reliably and preserve state for longer-lived workflows;
- schemas and interfaces must be easier for agents to discover and understand.
Supabase’s products and acquisitions respond to those pressures: Compute brings compute closer to storage, Multigres targets Postgres availability, and Turso provides options for scale-out deployment patterns.
Other topics covered
The conversation also delved into:
- the path from prototypes to petabyte-scale deployments;
- how AI affects Supabase’s growth and activation, especially agent-driven activation patterns;
- community-led distribution and crowdfunding within an open source strategy;
- balancing open source values with enterprise sales;
- the question of which layer will be stickier—application builders or the database itself;
- durable workflows and how to design infrastructure for them;
- code-first infrastructure and declarative schemas to help automation;
- whether AI-generated software will converge to common patterns.
Internal AI usage and company organization
Copplestone and Wilson also described how Supabase uses AI internally and discussed hiring practices for a remote-first company.
Structure and timing
The interview was structured into chapters, opening with a segment titled “Six years in the making” and continuing for over an hour. Notable segments included a deeper look at Supabase Compute, Multigres and Postgres reliability, the Turso acquisition and scale-out strategy, and the company’s priorities going forward.
Conclusion
Supabase is actively reshaping its infrastructure to prepare for a future with many more AI-created applications. By combining Compute, Multigres, storage-layer work like OrioleDB, and the Turso acquisition, the company aims to bring computation closer to data, improve Postgres reliability, and support varied scaling models. The discussion also underscored the role of open source, community distribution, and remote-first hiring in Supabase’s broader strategy.
