Rasa

What it does

Rasa is an AI agent platform for product teams building customer-facing assistants. It extends large language models with structured flows, deterministic logic, and built-in recovery patterns. The platform covers dialogue management, enterprise search, natural language understanding, voice, and orchestration, with tools for testing, deployment, and governance.

Rasa is aimed at product and engineering teams that want to build customer-facing agents without handing everything to a managed black box. The workflow starts with a prompt that you can tune in the UI, in code, or with a coding agent. Rasa then runs the agent on simulated conversations in voice and text and shows what passed and what did not. You can run end-to-end checks, compare versions, and take an update live when it is ready. The platform also reads conversations, tags and clusters findings, and puts them in an inbox. On the build side, Rasa offers a visual studio workspace, a version-controlled codebase with CLI and SDKs, and MCP tools that connect coding agents to the shared codebase. Enterprise controls include single sign-on, role-based access control, PII data management, and secrets management. Deployment options include Kubernetes with Helm, and channel connectors cover REST and WebSocket. A free Developer Edition is available for developers and teams starting a project. It allows one bot per company, with up to 1000 external conversations per month or 100 internal conversations per month, and free community support in the forum. Enterprise pricing is only available by contacting sales.