JuStAI

A local AI assistant · runs on your NAS or PC

The AI for your home.

JuStAI runs on your NAS or PC at home and turns cheap network cameras into microphone, speaker and eyes. Say the wakeword in any room: the assistant there listens, sees, remembers and answers through the camera’s speaker. It runs on your own hardware, not with a provider.

Install in 5 minutesRead the docs
$ curl -fsSL https://raw.githubusercontent.com/naxeu/justai/main/scripts/install.sh | bash

Needs a Linux PC (x86_64) in your home network. The installer takes care of Docker.

🗣 “Hey Justy” is the built-in wakeword. You can pick another built-in one or upload your own. →

“Hey Justy, where are my car keys?”
On the shelf in the pool house. I saw them 20 minutes ago.

See it in action

A short introduction, less than a minute.

Scenes and voice in this video are AI-generated. The JuStAI interface and features shown are real.

Localwake word, speech and vision run on your machine
Cheap hardwarea network camera has mic, speaker and eye in one
3 languagesEnglish, German and Italian
Open sourceAGPL-3.0, built by vibe coding

Everything a room needs, in one small camera

No smart speaker per room. A network camera already has the microphone, speaker and eye, and costs very little. A Tapo C200, for example, is often available for under €20.

📷
Cheap camerae.g. TP-Link Tapo C200
+
🧠
JuStAIFree and open source
=
🏠
Smart roomHears you, sees, answers
🎤Microphonelistens for “Hey Justy”
🔊Speakeranswers in the same room
👁Eyepeople, pets and objects
🎯Pan and tiltfollows you through the room
🔒Privacy modeswitch off by voice

Tested with the TP-Link Tapo C200. Other ONVIF/RTSP cameras can be added; speaker output needs a camera with two-way audio.

Just say it

No apps, no menus. Talk the way you would talk to a person.

🗣 “Where are my car keys? Search the pool house.”
Found them on the shelf next to the towels, 20 minutes ago.
🗣 “Tell me when a car drives into the driveway.”
Will do. I will say it in the room you are in.
🗣 “The next time you see me from tomorrow 9 am, remind me to take out the rubbish.”
Done. I will remind you the first time I see you after 9.
🗣 “Where is Lucky?”
Lucky is on the sofa in the living room.
🗣 “Switch off the camera in the bedroom.”
The bedroom camera is off.
🗣 “Build a plugin that tells me when electricity is cheaper than 20 cents.”
I wrote and installed the plugin. I will announce cheap hours.

More examples →

How it works

  1. 1

    Put a camera in every room

    A cheap Tapo camera (or any ONVIF/RTSP camera, or just a phone) becomes the microphone, speaker and eye of the room. Add it in the admin interface.

  2. 2

    Say “Hey Justy”

    The assistant wakes up, understands you and answers on the speaker in the same room, never on its own voice. The wakeword is yours to choose.

  3. 3

    Let it do things

    It remembers, watches, reminds, controls devices and can even write new plugins for you.

Everything in one assistant

📷

Cheap cameras, smart results

A cheap network camera delivers video, microphone and speaker. JuStAI adds the brain: understanding, answering and remembering.

🏠

Rooms and endpoints

It knows where you are and answers in the same room. A deterministic output router keeps replies from wandering off.

👀

Sees and recognises

People, pets and objects through your cameras, with privacy masks and per-room switches.

🧠

Remembers

Facts, habits and “the next time you see me” reminders, stored on your own machine.

⚡

Automations by voice

Tell it a rule in plain language and it sets it up: events, jobs and conditions.

🧩

Plugins

Drivers for Tapo, ONVIF and RTSP cameras, AI providers and more. It can write plugins for itself.

📱

Tablets, screens and phones

Show answers, charts and camera images on tablets, smart-home screens and old laptops. Phones work as microphone, speaker and camera, installable as an app.

🗣

Your own wakeword

Say “Hey Justy” or anything you like. Pick a built-in wakeword or upload your own openWakeWord model.

🎯

Follows you

PTZ cameras can track a person around the room.

Your home stays yours

Wake word, speech recognition, speech output, embeddings and image analysis run locally. Only the language model needs a decision: use a Google API key, or stay completely offline with a local model such as Ollama.

Read about security and privacy →
  • Runs on your own computer
  • No cloud account for JuStAI itself
  • Camera streams stay in your network
  • Fully local mode with Ollama possible
  • Open source and auditable

Open source, built by vibe coding

JuStAI was built with AI coding assistants, and you can extend it the same way. No computer science degree needed.

🧩

Extensible

Add drivers, tools and integrations as plugins, or simply ask the assistant to build one for you.

🌱

It learns

It remembers facts, habits and preferences, and gains new behaviour and new tools over time.

💻

Open source

AGPL-3.0: free to use, change and share. Plugins can use any licence.

🎨

Vibe-code your own

Describe what you want to an AI coding assistant, test it, steer it. The docs show exactly how.

How it works →

Questions, bugs, ideas

Everything happens on GitHub, in English or German.

💬

Ask a question

Ask, share what you built, talk about ideas.

Open Discussions →

🐞

Report a bug

Something does not work? Tell us what happened.

Report a bug →

💡

Suggest a feature

Missing something? Describe what you would like to do.

Suggest a feature →

🔧

Build it yourself

Fix or extend it with a coding assistant and send a pull request.

Read the guide →

Install it

You need one computer that runs all the time and about 15 minutes. No programming skills required.

What you need

  • One or more network cameras, e.g. TP-Link Tapo C200 (often under €20, with microphone, speaker and pan/tilt). A phone or tablet works to start.
  • A PC or mini PC with an x86_64 processor (Intel or AMD), running Linux, e.g. Ubuntu 24.04
  • 4 GB RAM (8 GB recommended) and 15 GB free disk space
  • A network connection in the same home network as your phones and cameras
  • A language model: a free Google API key, or a local one (Ollama)

Raspberry Pi and other ARM computers are not supported. Windows and macOS are not tested.

  1. Open a terminal

    On your Linux computer press Ctrl+Alt+T, or search for “Terminal” in the menu.

  2. Paste this and press Enter

    The installer checks your computer, installs Docker if it is missing (it asks first) and downloads everything.

    $ curl -fsSL https://raw.githubusercontent.com/naxeu/justai/main/scripts/install.sh | bash

  3. Answer two questions

    It asks for a Google API key (you can leave it empty and add it later) and an admin password. Where do I get the key?

  4. Wait a few minutes

    It builds and starts everything. At the end it prints the address, user and password.

  5. Open the address in a browser

    Usually http://YOUR-COMPUTER:8080, user admin. Then continue with the first steps.

Full step-by-step guide →

Ready to give your home a voice?

Install in 5 minutesBuild a plugin