Rasa

About Rasa

Rasa is a developer-focused platform for building, deploying, and managing enterprise AI agents. It combines large language models with structured business logic so organizations can create conversational agents that are more predictable, controllable, and suitable for complex workflows. Rasa’s roots go back to 2016, when the company introduced Rasa NLU, and the platform has since evolved toward LLM-powered agent development.

Rasa is designed for organizations that want more control over their AI stack. Instead of relying entirely on a hosted black-box conversational service, teams can control the underlying logic, models, integrations, deployment environment, and agent behavior. Rasa supports deployment on an organization’s own infrastructure, private cloud, or offline environments, including air-gapped deployments.

The current Rasa Platform combines Rasa Pro, its pro-code development framework, with Rasa Studio, a no-code interface for business users. This allows technical developers and non-technical teams to work with the same AI-agent platform.

CALM and Agentic AI

A central part of modern Rasa is CALM (Conversational AI with Language Models). CALM separates natural-language understanding from the execution of business logic. This allows an AI agent to communicate naturally while still following predefined flows and business rules.

Rasa describes CALM as a way to make agents more reliable for complex enterprise tasks. Flows define the business processes an agent can perform, while language models help the system understand how users express their requests.

This approach is useful when businesses need AI to do more than answer questions. Agents can handle structured processes such as customer-service journeys, account requests, employee helpdesk tasks, searches, and voice interactions.

AI Agent Development

Rasa provides a pro-code framework for developers who need detailed control over their AI applications. Teams can define flows, create custom actions, connect external systems, configure models, and manage the underlying project files.

Rasa also provides Rasa Studio, a visual interface that allows business users to build and modify conversational flows without working directly with code. Changes made through Studio can be converted into deployable Rasa configurations.

This combination makes Rasa suitable for organizations where developers and business teams need to collaborate on the same AI-agent projects.

Multi-Agent Orchestration

Rasa supports multi-agent orchestration, allowing organizations to coordinate multiple specialized agents. Agents can share conversational state and memory while handing tasks between different skills or domain-specific agents.

This architecture can be useful for complex enterprise applications where a single agent would otherwise need to manage too many responsibilities.

Voice AI

Rasa supports voice and IVR experiences in addition to text-based conversations. Its voice capabilities are designed for real-time conversations, including turn-taking, interruptions, and background-noise handling.

The current Rasa Platform includes an IVR connector for AudioCodes VoiceAI Connect, available through an additional purchase.

Enterprise Search and Knowledge

Rasa includes enterprise search capabilities that allow AI agents to retrieve relevant information when handling customer or employee requests.

The platform can combine knowledge retrieval with business logic, allowing agents to use enterprise information while still following defined workflows. This is particularly useful for support, internal helpdesk, and search-agent applications.

Integrations

Rasa is built around an open and extensible integration model. Its platform supports REST and WebSocket channel connectors, custom action servers, MCP-based integrations, and connections to external agents through A2A.

Rasa also supports integration with external APIs and enterprise systems. Its current documentation highlights MCP tools and integrations as part of its modern agent architecture.

Deployment and Security

One of Rasa’s strongest advantages is deployment flexibility. Organizations can deploy agents on their own infrastructure, private cloud, or managed environments. Rasa also supports offline and air-gapped deployments where required.

The platform includes features such as PII data management, secrets management, dependency vulnerability protection, observability, and multi-node concurrency.

This makes Rasa particularly attractive to organizations that have strict data, security, or compliance requirements.

Observability and Analytics

Rasa includes tools for monitoring and evaluating AI-agent behavior in production. Teams can trace conversations, inspect decision paths, analyze workflow performance, and identify problems affecting agent reliability.

Rasa Studio also provides conversation views and tools for analyzing and optimizing assistants.

Pricing

Rasa currently offers a Free Developer Edition and an Enterprise offering. The Developer Edition can be used locally or in production, with a limit of 1,000 external conversations per month or 100 internal conversations per month per company.

The Enterprise offering is aimed at organizations deploying conversational AI at scale and includes full access to the Rasa Platform, enterprise security features, increased automation capabilities, and premium support. Enterprise pricing is contact-based.

The broader Rasa Platform combines Rasa Pro with Rasa Studio. Rasa Pro provides the pro-code infrastructure, while Rasa Studio adds no-code development and business-user capabilities.

Review

Rasa is a strong choice for organizations that prioritize control, customization, security, and deployment flexibility. Its biggest distinction is that it gives developers significant control over how agents reason, follow business logic, connect to systems, and operate in production.

The combination of CALM, multi-agent orchestration, enterprise search, voice capabilities, observability, and flexible deployment makes Rasa well suited to complex enterprise applications.

Another major advantage is its deployment model. Organizations can run Rasa on their own infrastructure or private cloud, which can be important for businesses with strict data-governance requirements.

The main drawback is that Rasa is more developer-oriented than many simple chatbot builders. While Rasa Studio provides a no-code interface, organizations building sophisticated applications may still need technical expertise to configure integrations, business logic, deployment, and infrastructure.

Rasa’s free Developer Edition is useful for getting started, but its conversation limits mean larger production deployments will generally require an enterprise arrangement.

Best suited for: enterprises, developers, contact centers, customer-service teams, financial services, healthcare organizations, government services, internal helpdesks, and businesses that require highly customizable AI agents with strong control over infrastructure and business logic.