BroMind

Local-first AI · Chrome · Firefox · Edge

Your AI runs on your machine, not someone else's cloud.

BroMind is a browser sidebar for the AI models you run yourself — Ollama, LM Studio, llama.cpp, and any OpenAI-compatible server. Chat, read the page, ask about a PDF. Nothing leaves your computer.

Get BroMind See how it works Chrome · Firefox · Edge · no account
The BroMind side panel: a chat with a local model about the current page.
No account No cloud No telemetry Minimal permissions Runs on your hardware

The difference

Cloud AI extensions read your pages on their servers. BroMind reads them on yours.

Typical AI sidebar

Your data takes a trip

  • Page content and prompts sent to a company's servers
  • Accounts, subscriptions, and opaque credit systems
  • Server-side processing you can't inspect — you take the privacy claims on faith
  • Broad permissions that quietly expand over time
BroMind

Your data stays home

  • Prompts and page text go only to your own model server
  • No account, no cloud, no telemetry of any kind
  • Check it yourself — watch your network monitor, nothing leaves
  • Localhost-only by default; broad access is opt-in, per site

Two minutes

Running in three steps

No terminal wrangling — BroMind finds your server and fixes the connection for you.

STEP 01

Run a local model

Install Ollama (or LM Studio, llama.cpp, Jan, vLLM…) and start it.

$ ollama run llama3.2
STEP 02

Open the panel

BroMind auto-detects your server, fixes Ollama's connection rejection, and can pull a starter model sized to your hardware.

Ctrl/⌘ + Shift + .
STEP 03

Ask anything

Chat, summarize the page, dig into a PDF, or rewrite a selection — all on your machine.

“Summarize this page”

What it does

A full assistant, kept honest

Chat with any local server

Ollama, LM Studio, llama.cpp, Jan, vLLM, LocalAI — anything OpenAI-compatible. Switch models mid-conversation.

Read the page — honestly

Ask about the current page with visible token budgeting. Long pages are condensed in parts, never silently truncated.

Selection toolbar

Highlight text to explain, summarize, translate — or rewrite it in place, right inside the field you're typing in.

Chat with PDFs

Attach a PDF and ask about it — extracted on-device, through the same private pipeline.

Prompt library

Save reusable prompts and drop them in with a “/” slash command. Edit any message and branch from there.

Yours to keep

Full local history with search and export. One-click wipe of everything. It's all on your disk.

Private by architecture

Your data has nowhere else to go

The point of local is that there's no server in the middle. Open your browser's network monitor and see for yourself — nothing leaves for anyone but the model on your own machine.

No backend, no telemetry.The developer receives nothing. There is no server to receive it.
Localhost-only permissions.At install it can only reach your own machine. Any website access is granted by you, per site.
Untrusted content is fenced.Page text is treated as data, never as instructions — a documented defense against prompt injection.
You can watch it work.Every request goes to your configured local endpoint and nowhere else — visible in any network inspector.

Read the full privacy policy →

Keep your AI to yourself.

Install BroMind, point it at the model on your machine, and start asking.

Chrome Web StoreSoon Firefox Add-onsSoon Edge Add-onsSoon
Privacy policy