# SQLite AI

> SQLite AI transforms SQLite into a distributed AI-native database for the Edge — combining the simplicity of SQLite with cloud-powered scalability, offline sync, synchronized agent memory, fault tolerance, automatic backups, and powerful extensions. AI at the Edge. Scale in the Cloud.

SQLite AI unifies on-device intelligence with global Cloud infrastructure. Build apps and agents that think locally, keep working offline, sync data and memory seamlessly, and scale globally. Backed by the creator of SQLite.

## Mission
- A powerful, lightweight AI engine ready to be used in any framework
- The default database for intelligent, edge-native applications
- A privacy-preserving AI that runs where your data lives
- Next generation offline-first apps for an exceptional user experience
- A programmable, distributed engine for real-time, intelligent apps
- A unified platform for agents, vectors, analytics, sync, AI, and JavaScript at the edge
- Synchronized memory for AI agents across devices, workers, and cloud backends

## Products

### SQLite Cloud
URL: https://www.sqlite.ai/cloud
Docs: https://docs.sqlitecloud.io

The easiest way to sync, share and scale SQLite databases. Designed for global scalability.

Features:
- High concurrency with strong consistency
- Dynamically scalable, multi-region deployment
- Conflict-free synchronization across nodes and devices
- CloudSync for offline-first local replicas backed by SQLite Cloud, PostgreSQL, or Supabase
- Synchronized agent memory for fleets of assistants, workers, and autonomous systems
- Realtime infrastructure: subscribe to changes, trigger webhooks, enable multiplayer collaboration
- Edge Functions: execute custom code on the same nodes as your data
- Unified database operations on clients, servers, and the edge
- Point-in-time recovery and continuous backups
- Global deployment and scaling
- Zero-setup sync between edge and cloud

### SQLite-AI Extension
URL: https://www.sqlite.ai/sqlite-ai
GitHub: https://github.com/sqliteai/sqlite-ai

A powerful, lightweight, and cross-platform AI engine built into SQLite. Embeds llama and whisper for local inference.

Capabilities:
- Local LLM inference and text generation
- Embedding creation for semantic search
- Speech recognition and translation via Whisper
- Multimodal support (image processing and embedding)
- Chat with models and stream tokens in real time
- All operations accessible through simple SQL queries

Technical highlights:
- Blazing-fast C/C++ implementation with hardware acceleration (Vulkan, Metal, CUDA, BLAS)
- Zero runtime memory allocations for maximum stability
- No third-party dependencies, fully self-contained
- Works entirely offline, no external servers needed
- Real-time performance optimized for mobile, IoT, and embedded devices
- Prebuilt binaries for iOS, Android, macOS, Windows, and Linux
- Zero-dependency design: nothing to install beyond the extension itself
- Reduces inference costs by up to 99%

Example usage:
  .load ./ai
  SELECT ai_model_load('/path/to/model.gguf', 'n_predict=512,n_gpu_layers=99');
  SELECT llm_text_generate('Write a short story about SQLite');

### SQLite-Agent
URL: https://www.sqlite.ai/sqlite-agent
GitHub: https://github.com/sqliteai/sqlite-agent

Run autonomous AI agents directly inside SQLite. SQLite-Agent uses SQLite-MCP for external tools, SQLite-AI for model inference and embeddings, and SQLite-Vector for optional vector indexing when agent output is stored in tables.

Features:
- Multi-step autonomous agents that plan and execute tool-using tasks
- MCP integration for web, API, and external tool access
- Table extraction mode that populates SQLite tables from a goal
- Schema-aware structured extraction based on target columns and types
- Auto-embeddings for BLOB columns named like embedding fields
- Optional vector index initialization for semantic search over inserted rows
- Cross-platform binaries and packages for native, mobile, Swift, Android, and Flutter targets

Example usage:
  .load ./mcp
  .load ./ai
  .load ./vector
  .load ./agent
  SELECT llm_model_load('/path/to/model.gguf', 'gpu_layers=99');
  SELECT mcp_connect('http://localhost:8000/mcp');
  SELECT agent_run('Find apartments in Rome under 100 EUR', 'listings', 8);

### SQLite Sync
URL: https://www.sqlite.ai/sqlite-sync
GitHub: https://github.com/sqliteai/sqlite-sync

Local-first SQLite extension for seamless, conflict-free data sync and real-time collaboration across devices, browsers, robots, and software agents using CRDTs (Conflict-free Replicated Data Types). Use it to synchronize application state, local SQLite replicas, and agent memory.

Features:
- Offline-first by default: works seamlessly without a connection
- CRDT-powered sync: no conflicts, no data loss, no overwrites
- Built-in network layer: no servers or custom code required
- Row-level security: granular access control out of the box
- Cross-platform: iOS, Android, macOS, Linux, Windows, and WASM
- Zero setup: augment your tables and you're done
- Each device maintains its own database copy
- Automatic synchronization when reconnected
- No manual conflict resolution needed
- Works with SQLite Cloud, PostgreSQL, or Supabase as the central backend
- Syncs agent memory and task state across thousands of agents when needed

Example usage (3 lines):
  .load ./cloudsync
  SELECT cloudsync_init('my_data');
  SELECT cloudsync_network_sync();

### SQLite Vector
URL: https://www.sqlite.ai/sqlite-vector

A blazing fast and memory efficient vector search extension for SQLite. Handles millions of high-dimensional vectors on resource-constrained edge devices.

Features:
- Optimized C implementation with SIMD acceleration (AVX2, SSE2, Neon)
- No preindexing required: start searching instantly
- Massive datasets with minimal memory (30MB default)
- No external servers: guaranteed zero-latency at the edge
- Query millions of vectors in milliseconds
- No virtual tables or complex JOINs needed
- Prebuilt binaries for iOS, Android, macOS, Windows, Linux, and WASM
- Ideal for semantic search and retrieval augmented generation (RAG)

### SQLite-Columnar
URL: https://www.sqlite.ai/sqlite-columnar
GitHub: https://github.com/sqliteai/sqlite-columnar

Column-oriented analytics inside SQLite. SQLite-Columnar adds columnar virtual tables, stats, chunk metadata, range-aware helpers, and grouped aggregate functions for analytical workloads that need to stay embedded.

Features:
- Columnar virtual tables for analytical tables inside SQLite
- Separate per-column storage for scans over wide datasets
- Incremental columnar_analyze() metadata refresh
- Specialized aggregate helpers for sum, average, count, min, max, and grouped summaries
- Chunk zone maps for range filters that can skip irrelevant rowid ranges
- Good fit for local dashboards, telemetry, audit logs, ETL validation, and ML preprocessing
- Does not patch SQLite pager, btree, parser, VDBE, or shell

Example usage:
  .load ./columnar
  CREATE VIRTUAL TABLE sales USING columnar(id INTEGER, region TEXT, amount REAL);
  SELECT columnar_analyze('sales');
  SELECT columnar_sum('sales', 'amount');
  SELECT k, "sum", "avg", "count"
    FROM columnar_group_sum_avg_count('sales', 'region', 'amount');

### SQLite JS
URL: https://www.sqlite.ai/sqlite-js

Extend your SQLite database with custom JavaScript functions. Write cross-platform logic directly in the database.

Features:
- Define custom scalar, aggregate, window functions, and collation sequences in JavaScript
- Run JavaScript inside SQL with no extra setup
- Sync function definitions across devices using SQLite-Sync
- Write once, run anywhere: share logic across platforms, languages, frameworks
- Cross-platform and offline-ready: iOS, Android, macOS, Windows, Linux
- Perfect for distributed and edge-native apps
- Integrates with SQLite-Sync for automatic function synchronization across all connected clients

### SQLite-Memory
URL: https://www.sqlite.ai/sqlite-memory
GitHub: https://github.com/sqliteai/sqlite-memory

Persistent, searchable, and syncable memory for AI agents. Turns markdown content into queryable agent memory using semantic search and hybrid retrieval, then optionally synchronizes that memory across agents, devices, workers, and cloud backends through SQLite-Sync.

Features:
- Markdown-aware chunking for human-readable knowledge bases
- Vector similarity search with local embeddings
- FTS5 full-text search for keyword recall
- Hybrid retrieval combining semantic and keyword search
- Optional offline-first sync between agents via SQLite-Sync
- Shared knowledge and task memory for multi-agent systems
- Local autonomy first: agents can read and write memory without network latency
- Privacy-preserving: all data stays local by default

### SQLite-MCP
URL: https://www.sqlite.ai/sqlite-mcp
GitHub: https://github.com/sqliteai/sqlite-mcp

Connect SQLite to MCP servers and call tools directly from SQL. Bridges database state to external tool execution.

Features:
- Connect to Model Context Protocol servers from SQL
- List available tools and inspect schemas
- Invoke tools and store results in the database
- Combine tool output with local data in SQL queries
- No additional application layer required

### Adam Agent
URL: https://www.sqlite.ai/adam
GitHub: https://github.com/sqliteai/adam

Embeddable cross-platform AI agent library written in C. Adam provides a complete agent loop with cloud and local LLMs, tool calling, long-term memory, voice, sessions, structured output, research mode, database extensions, and WASM support.

Features:
- Agent loop with automatic tool calling until final response
- Cloud providers and local GGUF models through the same interface
- Built-in tools for file I/O, shell, calculator, SQL, web fetch/search, HTTP POST, memory, research, and multi-agent workflows
- Long-term memory with hybrid BM25 and vector search through SQLite-Memory and SQLite-Vector
- SQLite and PostgreSQL extensions that expose Adam as SQL functions
- Sessions, streaming, voice, guardrails, response cache, thread pool, and filesystem sandboxing

## Platform Features (All Plans)
- Serverless edge functions
- Embeddable databases
- REST API Access
- CRDT Distribution
- Offline sync for local SQLite databases
- Agent memory sync with SQLite-Memory and SQLite-Sync
- Role-based authentication
- Data sharding
- SDKs in multiple languages
- JSON storage
- AES-256 encryption
- 100% API compatible SQLite client
- Webhooks for reactive applications
- Full text search (coming soon)
- Vector database (coming soon)

## Use Cases
- Solutions overview: https://www.sqlite.ai/solutions
- Real-time chat and messaging
- Offline-first field apps that keep writing locally and sync when connected
- Personal finance apps
- Multiplayer games
- Fitness tracking across wearables, mobile, and web
- IoT device control and sensor data sync
- AI agents that synchronize memory, notes, embeddings, and task state
- E-commerce platforms with live inventory updates
- Collaborative document editors
- Learning management systems
- Calendar and event scheduling with offline editing
- Mobile and embedded AI applications
- Robotics and automotive systems
- Privacy-preserving on-device AI
- PostgreSQL and Supabase teams that want local SQLite replicas without replacing the central backend
- Managed SQLite Cloud backends with auth, APIs, RLS, encryption, webhooks, Edge Functions, Drizzle Studio, MCP Server, vector search, and AI features

## Pages
- Homepage: https://www.sqlite.ai
- SQLite Cloud: https://www.sqlite.ai/cloud
- Solutions: https://www.sqlite.ai/solutions
- SQLite-AI Extension: https://www.sqlite.ai/sqlite-ai
- SQLite-Agent: https://www.sqlite.ai/sqlite-agent
- SQLite Sync: https://www.sqlite.ai/sqlite-sync
- SQLite Vector: https://www.sqlite.ai/sqlite-vector
- SQLite-Columnar: https://www.sqlite.ai/sqlite-columnar
- SQLite JS: https://www.sqlite.ai/sqlite-js
- SQLite-Memory: https://www.sqlite.ai/sqlite-memory
- SQLite-MCP: https://www.sqlite.ai/sqlite-mcp
- Adam Agent: https://www.sqlite.ai/adam
- Pricing: https://www.sqlite.ai/pricing
- Sync Live Demo: https://www.sqlite.ai/sync-demo
- Changelog: https://www.sqlite.ai/changelog
- Support: https://www.sqlite.ai/support
- Privacy Policy: https://www.sqlite.ai/privacy
- Cookies Policy: https://www.sqlite.ai/cookies
- Documentation: https://docs.sqlitecloud.io
- Blog: https://blog.sqlite.ai
- Dashboard: https://dashboard.sqlitecloud.io
- GitHub: https://github.com/sqlitecloud