Rasa Open Source Is in Maintenance Mode. Long Live CALM.

What it does

Rasa Open Source is a Python framework for automating text- and voice-based conversations across Slack, Facebook Messenger, Telegram, Twilio, and other channels. The repository is now explicitly in maintenance mode; the company is steering users toward Hello Rasa, a browser-based playground for prototyping AI agents powered by the CALM engine.

The interesting bit

CALM—Conversational AI with Language Models—splits responsibility between an LLM that interprets natural language and explicit code that enforces business rules through “flows,” replacing the old intent-classification pipeline with logical steps like collecting money or transferring funds.

Key highlights

• The legacy framework supports contextual, multi-turn conversations across a wide range of messaging platforms.
• Hello Rasa drops traditional NLU training; the LLM understands user input while developers define rigid business flows in the browser.
• A built-in copilot generates code and debugs flows on demand.
• The inspector exposes real-time decision-making to avoid black-box behavior.
• Agents prototyped in Hello Rasa can be exported to the commercial Rasa Platform for production scaling.

Caveats

• Rasa Open Source is in maintenance mode, so the repo is effectively a legacy artifact rather than an active open-source innovation hub.
• The README blurs the line between documentation for the legacy framework and marketing for the Hello Rasa product, which may confuse developers looking to contribute.

Verdict

Worth studying if you are maintaining an existing Rasa deployment or researching conversational-AI architecture; avoid if you need an actively maintained, fully open-source framework for a new greenfield project.