Neon shifted decisively toward agent-native infrastructure over the last quarter. The platform moved beyond serving traditional developer workflows to become a first-class primitive for AI agents—from Stripe Projects automating database provisioning mid-agent execution, to the Cursor plugin giving agents structured access to schemas and credentials, to the MCP server standard letting agents discover and interact with Neon across different tools. Alongside this, Neon doubled down on production readiness for agent-generated apps: branching for isolated testing, autoscaling reports for visibility, and Auth integration with Vercel previews so agents can scaffold full-stack systems without manual configuration. The company also shipped zero-downtime patching prewarming to keep databases available during infrastructure maintenance, and expanded into data lakehouses with zero-ETL tooling for Postgres workloads moving toward enterprise analytics.
Neon shipped features cementing its role in agentic AI workflows. Stripe Projects integration enabled AI coding agents to provision databases directly within generated applications, while a case study with Specific highlighted the platform's ability to spawn thousands of databases at scale for agent-managed infrastructure. Separately, zero-downtime patching entered its first phase with prewarming, and query cancellation behavior in psql received attention.