Embeddings is all you need

Embeddings is all you need is an in-browser embedding classifier that performs intent classification using a tiny model.

Embeddings is all you need is an intent classification tool that operates entirely in the browser using WebAssembly. It runs completely client-side without a server or a large language model. The system uses a tiny 23 megabyte MiniLM-L6-v2 embedding model to process voice commands and determine intent via cosine similarity. It achieves fast response times of under 50 milliseconds after warm-up. This tool is designed for developers and users who need lightweight, private, and local intent recognition.

Embeddings is all you need at a glance

Best for: Software engineer
Runs on: Web
Pricing: Free
Released: June 2026
Category: AI productivity tool
Website: lusob.github.io

Key features

100% in-browser execution
No server required
No large language model used
Under 50 ms response time after warm-up
Intent classification via cosine similarity
Uses MiniLM-L6-v2 23 MB WASM model

Use it for

Classify user intents for voice commands
Process voice input locally without external servers

Alternatives to Embeddings is all you need

Docling — Provides advanced PDF understanding and local execution for sensitive data.
Use the AI you already pay for — Use existing AI subscriptions to process documents into audio.
WebAI by Opera — Runs local open-weight models directly in your browser without cloud costs.
I had 100 tabs open — Replaces tab hoarding with a private local search command bar.
Trim 1 — Runs on one CPU core in milliseconds with no GPU or network calls required.
Jab — Runs locally in the browser in two milliseconds with no network requests.
Xberg — Extracts documents, code, images, and audio using one native-Rust engine.
Remio — Local-first AI memory indexing unlimited files and meetings with complete data control.
LocalBolo — Private, on-device dictation that works in any app without cloud data usage.
Lucid — Converts dictated text into structured notes and diagrams using a local model.
IWE — Works natively across Mac apps, text editors, terminal CLI, and MCP servers.
Clotr — Runs locally with no network or accounts required.

All Embeddings is all you need alternatives