MariaDB has added vector search to its open-source database, but the MariaDB Foundation's executive chairman, Kaj Arnö, says the feature is being ignored by AI coding assistants. He told The Register at Percona Live that the world doesn't understand that MariaDB has vectors, and that developers using tools like Claude Code will not end up using MariaDB for vector search if the tool doesn't suggest it.
Arnö attributes this to LLM training data and documentation. Many models were trained before MariaDB's vector features existed, and frameworks and examples tend to favor established alternatives. In an unscientific check, The Register asked Claude which database to use for vector search: it suggested PostgreSQL first and did not mention MariaDB. When asked to compare MariaDB with PostgreSQL, however, it knew about MariaDB's capabilities.
Arnö compares the situation to search engine optimization, saying the world has moved from SEO to LLM optimization. He believes the only proper way to influence recommendations is to have the feature, document it, and get people using it. That creates a chicken-and-egg problem: without users generating examples and discussions, AI tools have less material to learn from. The article also notes a separate controversy over Galera clustering technology and its place in MariaDB's community edition, with Arnö defending the MariaDB server's open-source core.