# `data.vector_store_unsupported`

Vector field on a store that isn't PostgreSQL.

Area 21, [AI](../topics/ai.md); not declinable; the next step is usually the agent's.

A field holding a vector, a vector type's or an `Embedding`'s, in an
entity whose `store:` isn't PostgreSQL. Vectors are stored and searched
by pgvector, and there is no fallback. It is said in place of Redis
entities' own `toolchain.not_built_yet`, since it stays when they are
built.

```jadpo
application {
    connections: {
        drafts: {
            driver: Driver.redis
            url: config.redis_url
        }
    }
}

config {
    redis_url: RedisUrl
}
```


```jadpo
use support/types { ArticleMeaning }

/// A question someone is still typing, kept for an hour.
entity OpenQuestion {
    id: Uuid
    question_meaning: ArticleMeaning

    store: drafts
}
```

```text
error: `OpenQuestion.question_meaning` is a vector, which `drafts` can't store
 --> domains/support/entities/open_question.jadpo:6:5
  |
6 |     question_meaning: ArticleMeaning
  |     ^^^^^^^^^^^^^^^^ a vector, in Redis
  reason: A vector is kept in a column of pgvector, an extension of PostgreSQL, and `drafts` is a Redis connection. There is no fallback, since a search held anywhere else would behave differently from the one that was tested.
  next: Remove `store: drafts` to keep `OpenQuestion` in `main`, a PostgreSQL store (agent)
  rule: data.vector_store_unsupported
  help: https://jadpo.dev/docs/errors/data.vector_store_unsupported

check failed: 1 error, 3 files
```