AI2AI Chat

A research tool for running structured conversations between two AI models, built for academic research at Maastricht University. Each bot is configured independently — different provider, model, system prompt, and generation parameters — and you watch them talk to each other. Useful for studying how models respond to each other, testing prompt strategies, or collecting dialogue data for research.

Live: ai2aichat.com

---

What it does

You set up two AI bots side by side. Each one has its own provider, model version, system prompt, temperature, and token limit. You send an opening message and the two bots respond to each other in turn. You can let them run automatically or step through each response manually.

The main things you can configure and control:

Per bot:

• Provider — OpenAI, Anthropic, Google Gemini, or Mistral
• Model version (e.g. GPT-4o, Claude Sonnet 4.6, Gemini 1.5 Pro, Mistral Large)
• System prompt — defines the bot's role, persona, and rules
• Temperature (0–2) and max output tokens
• Custom name, bubble colour, and text colour for the conversation view

Conversation controls:

• Auto-interact mode — the bots reply to each other automatically up to a set number of messages per bot (max 25 each)
• Manual mode — click to trigger each response one at a time
• Response delay — add a fixed pause between replies, with an optional length-based variance that scales with message length
• Repetitions — run the same conversation from scratch multiple times in sequence (useful for collecting varied outputs to the same prompt)
• Save to history — toggle whether the conversation is written to the database

Export:

• Plain text transcript
• CSV with per-message metadata (sender, model, temperature, word count, response time in ms)
• Screenshot of the conversation panel

Data view: Alongside the chat view there is a live data table showing each message with its sender, model version, temperature, word count, and response time. This updates in real time during a conversation.

Conversation history: Past conversations are saved to a Supabase database per user account. You can browse them, preview the full exchange, and reload any conversation back into the chat panel to continue or re-examine it.

Other:

• Dark mode (follows system preference on first visit, then remembers your choice)
• User account settings — update display name, email, or password; delete your conversation history; or permanently delete your account

---

Supported models

Provider	Models
OpenAI	GPT-4o, GPT-4o Mini, GPT-4 Turbo, GPT-4, GPT-3.5 Turbo
Anthropic	Claude Sonnet 4.6, Claude Opus 4.6, Claude Haiku 4.5, Claude 3.5 Sonnet, Claude 3.5 Haiku, Claude 3 Opus
Google	Gemini 1.5 Pro, Gemini 1.5 Flash, Gemini Pro
Mistral	Mistral Large, Mistral Medium, Mistral Small, Mistral 7B (open)

API keys are entered in the bot configuration panel. They are stored in your browser's local storage and cleared automatically when you sign out. Nothing is transmitted to or stored on this server — requests go directly from your browser to each provider's API.

---

Running locally

Requirements: Node.js 18+ and npm, and a Supabase project.

git clone https://github.com/JonasHeller1212/AI2AI-Chat.git
cd AI2AI-Chat
npm install

Create a .env file in the project root:

VITE_SUPABASE_URL=https://your-project.supabase.co
VITE_SUPABASE_ANON_KEY=your-anon-key

Start the dev server:

npm run dev

The app runs at http://localhost:5173.

Set up the database schema:

Open the Supabase SQL editor for your project and run the contents of supabase/migrations/001_initial.sql.

Deploy the Edge Function (required for account deletion):

Install the Supabase CLI, then:

supabase login
supabase link --project-ref your-project-ref
supabase functions deploy delete-account

The function uses the built-in SUPABASE_SERVICE_ROLE_KEY secret — no additional configuration is needed.

To build for production:

npm run build

Output goes to dist/.

---

Deploying to Netlify

Connect the repo to Netlify and it will build and deploy automatically on every push to main. The netlify.toml in the repo root handles the build command and SPA routing.

One-time setup:

1. In your Netlify site settings, add two environment variables:

   • VITE_SUPABASE_URL
   • VITE_SUPABASE_ANON_KEY

2. Run the database migration in the Supabase SQL editor (supabase/migrations/001_initial.sql).

3. Deploy the Edge Function (see the local setup instructions above). The function must be deployed from the Supabase CLI — it cannot be configured through Netlify.

---

Tech stack

• React 18 + TypeScript 5
• Vite 5
• Tailwind CSS 3
• Supabase — auth and conversation storage
• Lucide React — icons
• html2canvas — screenshot export

Built by

Jonas Heller — Assistant Professor of Marketing, Maastricht University.