Welcome SQLite AI

Marco BambiniMarco Bambini
3 min read

Introducing SQLite AI: The Database for Local, Private, and Intelligent Apps

When we started this journey, our mission was simple: make it easy to sync, share, and scale SQLite databases. We believed that developers deserved the power of the cloud without the complexity, and that SQLite — the most deployed database engine in the world — could play a bigger role in modern app development.

Today, after several months of intense development, we’re thrilled to announce the next chapter:

SQLite AI — The Local Brain for Your Smart Apps

We envision a future where every app, device, or robot can run AI workloads locally — privately, securely, and without needing massive infrastructure or a constant internet connection.

Edge-native AI isn’t a dream anymore — it's here, and it's powered by SQLite AI.

Why SQLite + AI?

SQLite is already embedded in every phone, browser, and OS. It’s fast, reliable, and developer-friendly. But what if your database could do more than just store data?

What if it could:

  • Classify images or transcribe audio on-device

  • Embed, search, and retrieve vector data

  • Run machine learning models at the edge

  • Learn and adapt — all within a few lines of SQL

With SQLite AI, that’s exactly what’s possible.

Local-First. Intelligent. Open Source.

We've built a set of SQLite extensions that bring AI capabilities to your local database:

  • SQLite-Vector: Store and search vector embeddings using L2, cosine, or dot-product distances — ideal for recommendation engines, semantic search, and retrieval-augmented generation.

  • SQLite-AI: Run on-device LLMs and transcribe audio in real time with Whisper and LLaMA models.

  • SQLite-JS: Define SQL functions in JavaScript, opening the door for dynamic logic and edge scripting.

  • SQLite-Sync: A local-first extension to Seamlessly sync data across devices and the cloud with minimal setup.

You can now build local-first, intelligent applications with cross-platform SQL that works on iOS, Android, Linux, Windows, and macOS — with no server required.

Built for Edge AI

Whether you're building:

  • A smart assistant for field workers

  • A voice-enabled app for offline environments

  • A private, intelligent notebook for your thoughts

  • Or a robot that learns from its surroundings

SQLite AI gives you a foundation to build, ship, and scale edge-native AI applications — securely and efficiently.

Fully Open Source

Everything we’ve built is now available on GitHub, under permissive open-source licenses. We believe the future of AI should be open, local, and in your hands:


Start Building Today

You can get started with just a few lines of SQL:

-- Augment a table with CRDT for a full offline-first experience
SELECT cloudsync_init('my_table');

-- Run a nearest-neighbor search (top 10)
SELECT e.id, v.distance 
FROM embeddings AS e
JOIN vector_quantize_scan('embeddings', 'vector', ?, 10) AS v
ON e.id = v.rowid;

-- Perform inference on device
SELECT llm_text_generate('Tell me a joke');

-- Define new logic in JavaScript
SELECT js_eval('return input.toUpperCase()', 'hello world');

We’re incredibly excited to share SQLite AI with the world. This is just the beginning.

Let’s build a smarter, more private, and decentralized future — together.

0
Subscribe to my newsletter

Read articles from Marco Bambini directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Marco Bambini
Marco Bambini