Get started

Use a Redis deployment with RediSearch vector commands enabled.

pnpm add @anvia/core @anvia/redis redis

import { retrieveDocuments } from '@anvia/core/vector-store';
import { RedisVectorClient } from '@anvia/redis';
const storeClient = new RedisVectorClient({});
const store = storeClient.vectorStore({
    indexName: 'support_docs',
    keyPrefix: 'knowledge:support:',
    dimensions: 1536
});
await store.validate();
await store.upsert({
    documents: documents
});
const results = await retrieveDocuments({ store, model: embeddings, query, topK: 5 });

Documents must already contain embeddings. Without an injected client, the adapter uses REDIS_URL or redis://localhost:6379 and connects automatically.

Next

• Collections and indexing
• Search and filters
• Production