tisement-agent 0.2.0

WhatsAI — deployable AI customer service agent by Tisement Media. Connect LLM + WhatsApp webhook and serve 24/7.

WhatsAI by Tisement Media

AIM: A fully deployable AI customer service agent built for small businesses.

Philosophy

Customers want their business on WhatsApp, not just for marketing, but for the actual customer service. Small businesses don't have the staff to answer 24/7. WhatsAI lets small businesses engage leads, answer FAQs, book orders, and handle real conversations inside WhatsApp — all from a live webhook and an LLM backend that Tisement Media can configure per client.

Installation

pip install tisement-agent

Setup

whatsai setup

You'll be prompted for:

1. LLM provider + API key (OPENAI_API_KEY, ANTHROPIC_API_KEY, or custom endpoint)
2. WhatsApp Business API credentials (WHATSAPP_PHONE_NUMBER_ID, WHATSAPP_ACCESS_TOKEN)
3. Optional: Facebook Messenger and Instagram
4. Client config path

Run Webhook Server

whatsai run --config ./clients/savannah-grill

Test Locally

whatsai chat --config ./clients/savannah-grill

Project Structure

tisement-agent/
  tisement_agent/
    cli/              — Typer CLI entry points
    core/             — Webhook router, LLM client, conversation store
    platforms/        — WhatsApp, Messenger, Instagram adapters
  clients/            — Client configs + KBs (not shipped, user-local)
  tests/

Deployment

The server is a single FastAPI app. Deploy to Fly.io, Railway, Railway, Render, Heroku, or any VPS.

Environment variables for production:

• WHATSAPP_PHONE_NUMBER_ID
• WHATSAPP_ACCESS_TOKEN
• WHATSAPP_WEBHOOK_VERIFY_TOKEN
• LLM_API_KEY
• LLM_PROVIDER (openai | anthropic | custom)
• LLM_MODEL_NAME
• PORT (auto-set on most platforms)

License

MIT © Tisement Media