# IQAI — Multi-Model AI Dashboard

A sleek AI chat dashboard powered by [Replicate](https://replicate.com). Chat with multiple AI models simultaneously, tag them with @mentions, and augment responses with live web search.

## Features

- **Multi-model chat** — Select multiple models and query them in parallel
- **@mention routing** — Type `@claude`, `@grok`, `@gemini` to direct messages to specific models
- **Web search** — Inject DuckDuckGo search results as context before querying models
- **Model manager** — Add, edit, or remove Replicate models from the UI
- **System instructions** — Per-model system prompts per conversation
- **Account info** — View your Replicate account details in the sidebar
- **Conversation history** — Persisted locally across sessions

## Quick Start (Docker)

```bash
cp .env.example .env
# Edit .env and set your REPLICATE_API_TOKEN

docker compose up --build
# Open http://localhost
```

## Development

```bash
# Backend
cd backend && npm install && node server.js

# Frontend (in another terminal)
cd frontend && npm install && npm run dev
```

## Adding Models

Click **Manage Models** in the sidebar. Paste any Replicate model API URL to auto-fill the fields:

```
https://api.replicate.com/v1/models/owner/model-name/predictions
```

## Default Models

| Model | Tag | Type |
|-------|-----|------|
| Claude Opus 4.6 | @claude | Text |
| Grok 4 | @grok | Text |
| Gemini 3.1 Pro | @gemini | Text |
| DeepSeek R1 | @deepseek | Text |

## Architecture

```
iqai/
├── backend/          Express API server (port 3001)
│   ├── routes/       chat, models, account, search
│   └── data/         models.json (persisted config)
├── frontend/         React + Vite + Tailwind
│   └── src/
│       ├── components/
│       ├── store/    Zustand state management
│       └── utils/    API client
├── docker-compose.yml
├── Dockerfile.backend
├── Dockerfile.frontend
└── nginx.conf
```