The AI-Native Client-Side Database

Columnist-DB is the only client-side columnar database combining lightning-fast full-text search, vector embeddings, and persistent AI memory — all in one engine.

Key Highlights

Query 100k embeddings in <50ms
Multi-device synchronization
Full TypeScript support

Open Source

MIT License
◇ Hybrid Search with Vector Embeddings
import { Columnist } from 'columnist-db-core'
// Define your schema
const schema = {
memories: {
id: { type: 'string', primaryKey: true },
content: 'string',
embeddings: 'vector',
tags: 'string[]'
}
};
// Initialize database
const db = await Columnist.init('my-app', { schema })
// Store data with vector embeddings
await db.insert('memories', {
id: 'mem_1',
content: 'Important project notes',
embeddings: [0.1, 0.2, 0.3],
tags: ['project', 'important']
});
// Hybrid search (text + vector)
const results = await db.search('memories', {
query: 'project notes',
vector: [0.1, 0.2, 0.3],
similarityThreshold: 0.7
});
// Query 100k embeddings in <50ms ⚡

Columnar Storage Architecture

Optimized for analytical queries and vector operations with compression. IndexedDB persistence with offline capabilities and ACID compliance.

AI Memory & MCP Integration

Universal AI memory layer with 16 MCP tools. Store conversations, documents, and custom content with semantic search and cross-session persistence.

TypeScript First & React Hooks

Fully typed with Zod schema validation and type-safe migrations. Seamless React integration with hooks for real-time queries and updates.