Blink Bot
GitHub

Open-source AI companion

An AI that lives on your desk.

Chat, memory, voice and vision. Blink runs on your computer today and moves into a small desk robot when you are ready.

Get it on GitHub See it in action
  • GPL-3.0
  • Ukrainian & English
  • macOS · Linux · Raspberry Pi
The Blink dashboard: a chat where the bot checked the weather and searched the web, with its sources under the answer.
Shows its work Every tool it calls appears live: weather, search, memory.
Cites where it looked Links from the search sit right under the answer.
Pick the brain Any model behind one gateway, with as much thinking as you like.

Most assistants live in a browser tab. This one has a screen on your desk, a memory you can open and read, and a folder of its own. And it tells you what it is doing while it works.

The dashboard

Everything it knows, in one place.

  1. 01

    A chat that shows its work

    Replies arrive the way a person texts: a quick note, the tools it used, then the answer — with sources underneath.

    A conversation in the dashboard: the agent activity tree with weather and search, and the bot's answer with sources.
  2. 02

    A memory you can read

    What it learns about you and your people lives in plain Markdown notes on your own disk. Open them, fix them, delete them.

    The memory panel: a list of notes and an open note about a friend called Olya.
  3. 03

    Make it yours

    A name, a personality, themes, accents and wallpapers. Dark graphite and terracotta out of the box.

    A new conversation over a painted sky, with the bot's greeting.
  4. 04

    From the couch, too

    A native Android app for your chats, files and voice input, connected to the same bot.

    A conversation in the native Blink Android app, with its drawer closed.
A conversation in the dashboard: the agent activity tree with weather and search, and the bot's answer with sources. The memory panel: a list of notes and an open note about a friend called Olya. A new conversation over a painted sky, with the bot's greeting. A conversation in the native Blink Android app, with its drawer closed.

The device

And this is its screen.

A small screen on your desk with a clock, the weather and a drawer full of apps — everything you need within reach.

18
apps in its store
3
screen skins
320×240
pixels on the screen
The bot's screen: a large pixel clock and the date. The bot's screen: today's weather, an hourly graph and the week ahead. The bot's screen: the app drawer laid out as a honeycomb.

Raspberry Pi · screen · mic · speaker · camera

And more

Small things, done properly.

Talks and listens

Speech recognition with Whisper and a natural voice with Piper — both run on your own machine.

Lives in your messengers

Write to it in Telegram or Discord — the conversation shows up in the dashboard too.

Sees through a camera

Face and motion detection over a webcam, so it notices when you sit down.

Plays music on its screen

YouTube and YouTube Music right on the device — just ask.

Works on a schedule

Morning digests, reminders and recurring jobs run on their own.

Has a folder of its own

It reads and writes files in its workspace and draws diagrams you can edit.

How it works

Three layers, one personality.

  1. The body Raspberry Pi

    Screen, mic, speaker and camera. Reflexes in under 50 ms.

    < 50 ms
  2. The home Your computer or a home server

    Memory, tools, voice and the dashboard. Answers in 100–500 ms.

    100–500 ms
  3. The thinking Language models

    Any model through the OpenClaw gateway — swap it whenever you like.

    0.3–3 s

Get started

Three commands to a living bot.

No hardware needed to begin: the whole bot runs on your computer first, and the robot can wait.

You will need Python and access to a language model. The README walks you through both.

zsh — claude-bot
  1. 1Clone the repository git clone https://github.com/Imlokzu/blinkbot.git
  2. 2Start the bot — it sets up its own environment cd "blinkbot/Virtual Bot" && ./start.sh
  3. 3Open the dashboard in your browser open http://127.0.0.1:8100/dash/

Build your Blink Bot.

Free, open source, and already waiting on GitHub.

Open on GitHub