Stateful flow functions with Dapr

This demo shows how to build a telegram chatbot for GPT4. The application is written in Rust and deployed as a Wasm serverless function on the flows.network platform. It utilizes the following SaaS or PaaS platforms.

• Telegram API to interact with bots on the telegram platform.
• OpenAI's Assistant API to interact with GPT4 in threads.
• The Catalyst Dapr-as-a-Service API to manage application states.

The Catalyst service provides a key-value store itself. But more interestingly, it can use almost any cloud-based storage service as its KV store backend, providing a standard API for multi-cloud applications.

Try the Telegram bot yourself

Send a message to the bot on Telegram

Deploy on flows.network

• Fork the GitHub repo for the flow function's source code
• In flows.network
  • Create a flow
  • Import the above forked repo
  • Set variables in the Advanced tab
    • telegram_token -- The access token for your telegram bot created by the botfather
    • OPENAI_API_KEY -- The OpenAI API key
    • openai_assistant_id -- The ID for the OpenAI assistant you created
    • catalyst_token -- The access token for your Catalyst account
    • catalyst_url -- The HTTPS access URL for your Catalyst project
    • catalyst_kvstore -- The name of your Catalyst application's kvstore
  • Deploy
• Send a message to your telegram bot