AI-native

Chinese tea, in your chat.

Each bot is a thin webhook over the Content API: send a tea name — get its full card; /random to discover. The user's locale picks the language (up to 72). One backend, one handler per platform.

How a message flows

user: "Da Hong Pao"
  → GET /api/v2/search?q=Da Hong Pao        (resolve slug)
  → GET /api/v2/preview/da-hong-pao?lang=ru     (localized card · free)

Platforms

Platform	Status	Needs
Telegram	live	@TheTeaSupportBot — DM a tea name
Discord / Slack	planned	app token (self-serve)
LINE	planned	Messaging API channel
WhatsApp / Messenger	planned	Meta Business + app review
WeChat 公众号	planned	verified China business

Launch the Telegram bot

Message @BotFather → /newbot → copy the token.
Deploy the worker (code in the repo, bots/telegram/):

cd bots/telegram
npx wrangler deploy
npx wrangler secret put BOT_TOKEN     # paste token
# point Telegram at the worker:
curl "https://api.telegram.org/bot<TOKEN>/setWebhook?url=https://thetea-telegram-bot.<acct>.workers.dev"

Then DM it: Biluochun, 大红袍, /random, /tea longjing. The other platforms reuse the same handler — the work is mostly account approval, not code.

Other ways in

The same data is a REST API, a hosted MCP server, an AI skill, an OpenAPI spec and an llms.txt Account.