Polygres

Polygres is a managed PostgreSQL service that consolidates relational records, graph relationships, vector embeddings, and full-text search into a single project so AI applications can retrieve context from operational data without maintaining separate systems. It is aimed at teams building SaaS search, recommendations, customer-support investigation, and retrieval-augmented agent workflows on top of PostgreSQL, keeping relational facts, graph connections, and semantic search in one Postgres-native platform instead of relying on brittle ETL pipelines. The company is run by Dale and Dalton, the founders of Polygres , and Dale (@daleverett) lists the company as backed by @fdotinc . This launch introduces Polygres MCP, a Model Context Protocol server that lets AI clients like Claude, Cursor, and Codex connect to a Polygres workspace through browser sign-in rather than raw API keys. Users can scope access to a single project or every project they belong to, decide whether the client is read-only or allowed to make changes, and toggle individual surfaces such as databases, imports, AI search, graph, and debugging. Under the hood the server exposes 91 tools, but the client only sees the subset that matches the user's permissions and current project, and Polygres pauses to confirm the exact action before anything consequential runs. Shipping alongside MCP is Agent Skills 0.6.0, which extends Codex support and packages five Polygres skills covering setup, sync, write, memory, and graph retrieval. For founders and engineering teams already using Polygres as their retrieval layer, the launch matters because it turns the platform into something an AI agent can operate directly with the same access model a human teammate would have, rather than requiring custom glue code for every client.