Rasa

Rasa is a development platform for building enterprise-grade conversational AI agents. It operates using a hybrid architecture that combines a patented dialogue management system with its CALM (Conversational AI with Language Models) framework, enabling structured flows and deterministic logic. The platform addresses complex dialogue challenges through a language-agnostic NLU, enterprise search for RAG, and a custom actions server via the Rasa SDK. For infrastructure, it supports on-premise, private cloud, and air-gapped deployments, with Kubernetes integration managed through Helm charts and secrets management powered by Vault for enhanced security. The platform is built for enterprise development teams, data scientists, and conversation designers creating sophisticated agents for customer support, sales, and operational efficiency. Developers use the Rasa SDK to connect to external APIs, while designers use the no-code Rasa Studio to build flows, manage multilingual responses, and analyze user interactions. The system facilitates end-to-end testing and observability via OpenTelemetry, allowing teams to debug conversations, manage multi-LLM integrations, and roll back changes directly from the UI, fitting into a continuous integration and deployment cycle.