Side Eye

Local AI side panel chat with page context and read-aloud

Side Eye adds a local-first AI chat panel beside the tab you're browsing. Chat on-device with Gemma 4 via LiteRT-LM (default) — no local server or cloud account required. Or connect LM Studio or Ollama on your machine for larger models, vision, and MCP tools. Stream markdown replies, use multiple chat tabs, and optionally include page context (URL, selection, readable excerpt) when you ask about the current page. FEATURES • Side panel chat with streaming responses • LiteRT-LM on-device profile — Gemma 4 (E2B or E4B) runs in Chrome with WebGPU; one-time model download from Hugging Face • LM Studio and Ollama profiles (localhost or LAN URLs you configure) • Page context toggle and domain ignore list for privacy • Image attachments for vision-capable local models (LM Studio / Ollama; text-only for LiteRT-LM) • Quick actions and page-type suggestions • Optional on-device text-to-speech (Supertonic) • Dark/light theme and customizable UI PRIVACY Chat history and settings stay in your browser. LiteRT-LM inference runs entirely on your device. Page text is read only when you enable context and is sent to your chosen AI — on-device (LiteRT-LM) or your configured local server (LM Studio/Ollama). No analytics or ads. REQUIREMENTS • Google Chrome 114+ (Side Panel API) • LiteRT-LM (default): Chrome with WebGPU; network access for one-time Gemma 4 model download • Optional: LM Studio and/or Ollama for local server inference • Optional: network access for one-time TTS model download Privacy: https://getsideeye.com/privacy-policy.html