# Jarvis AI

[Canonical page](https://www.mohtasham.dev/work/archive/jarvis-ai)

A terminal-style personal assistant with voice input, spoken replies, search, weather, and command history.

2025. Prototype. Role: Independent build.

Jarvis AI puts a small personal assistant inside a terminal instead of a chat app. It accepts normal questions, but it also has explicit commands for the jobs that should be predictable.

Holding the spacebar starts voice input. ElevenLabs reads answers aloud, with browser speech as a fallback, while separate services handle web search and weather.

The terminal keeps command history and keyboard controls for clearing the screen or stopping an active response. New commands can be added without changing the rest of the interface.

## Facts

- Built-in command groups: 4. 

- Keyboard controls: 4. 

- Input modes: 2. 

## How it works

1. Type or hold: Enter text or hold spacebar to speak.

2. Route: Match a command or pass the question to the model.

3. Fetch: Call search or weather services when needed.

4. Answer: Print the result in the terminal.

5. Speak: Read it aloud through ElevenLabs or browser speech.

## Languages

- TypeScript: 89407 bytes

- CSS: 4168 bytes

- JavaScript: 474 bytes

[Public GitHub repository](https://github.com/MohtashamMurshid/jarvis-ai)
