Rasa Platform
AI model by Rasa
Build, test, and deploy reliable AI agents with full infrastructure control.
Rasa Platform is a self-managed conversational AI environment for enterprise teams to build and scale reliable AI agents. It combines a visual flow builder with developer tools, allowing cross-functional teams to collaborate on agents that run securely within their own infrastructure.
Model specs
The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.
- Price
- Free
Category: Conversational AI
- Modalities
- Text, Voice → Text, Voice
- Fine-tuning
- LLM fine-tuning recipes supported
- Deployment Support
- Kubernetes through Helm
- Concurrency Management
- Redis
- Data Pipeline
- Kafka
- Observability Standard
- OpenTelemetry
- Self-managed deployment on-premise or private cloud
- Visual-to-code synchronization for team collaboration
- Kubernetes deployment support through Helm charts
- Automated simulations for agent performance evaluation
- PII data management and Vault-powered secrets management
Overview
Rasa Platform is an enterprise-grade conversational AI environment designed for product teams to build, test, and deploy reliable AI agents. It serves businesses in regulated industries like finance, healthcare, and government by offering self-managed deployment on-premise or in private clouds, ensuring data remains within the customer's infrastructure. The platform features a visual-to-code synchronization that allows non-technical members to use a visual builder while developers work in a structured codebase. It includes the Mantle orchestrator, which combines agentic conversation with reusable skills, tools, and guardrails. Rasa Platform supports large-scale deployment on Kubernetes and integrates with enterprise infrastructure like Redis for concurrency and Kafka for data pipelines.
Build and Design
- Visual and Code-Based Design: Product teams can design AI assistants using a visual builder or directly in code.
- No-Code Flow Builder: The platform includes a no-code AI assistant flow builder for non-technical users.
- Content Management System: Teams can manage content and responses through a centralized system.
- Open Prompt Editing: Developers can edit prompts directly within the user interface.
- Custom Actions Server: The platform provides a custom actions server via the Rasa SDK for extending functionality.
Test and Optimize
- Automated simulations allow teams to evaluate agent performance against common questions or difficult customer scenarios.
- End-to-End Testing Panel: A dedicated testing panel allows for end-to-end testing of assistants before launch.
- Conversation Analytics: The platform provides a conversation view to analyze interactions and identify areas for improvement.
- Version Control and Rollback: Users can observe changes and roll back to previous versions if needed.
- LLM Fine-Tuning: The platform supports LLM fine-tuning recipes to optimize model performance.
Deploy and Scale
- Kubernetes Deployment: Deployments are supported on Kubernetes using Helm charts for enterprise scaling.
- Multi-Node Concurrency: The platform uses Redis to manage multi-node concurrency for high-volume traffic.
- Kafka Data Pipeline: A conversational data pipeline is powered by Kafka for data streaming.
- OpenTelemetry Observability: Standardized observability is provided through OpenTelemetry integration.
- Channel Connectors: Built-in connectors are available for REST and WebSocket channels.
Governance and Security
- Access Control: The platform includes Single Sign-On (SSO) and Role-Based Access Control (RBAC) for secure team management.
- Data and Secrets Management engine: PII data management and secrets management powered by Vault are included from day one.
- Voice Integration: An IVR Connector to AudioCodes VoiceAI Connect is available as an optional purchase.
Use cases
- Banking teams build agents to verify identity, freeze cards, and dispute charges without manual call queues.
- Retailers deploy agents to track packages, issue refunds, and pull order history automatically for customers.
- IT departments use agents on Slack to reset access, open tickets, and check requests against role permissions.
- Government agencies provide citizens with simpler ways to access public services through automated text and voice agents.
- Healthcare organizations build compliant agents to support patient needs while keeping sensitive data within their own infrastructure.
Specifications
- Modalities
- Text, Voice → Text, Voice
- Sizes and variants
- Developer Edition, Business (Rasa Pro + Studio), and Enterprise
- Access
- Self-Managed (On-premise or Private Cloud) or Managed Service
- Fine-tuning
- LLM fine-tuning recipes supported
- Languages
- Language-agnostic natural language understanding and multilingual support
- Deployment Support
- Kubernetes through Helm
- Concurrency Management
- Redis
- Data Pipeline
- Kafka
- Observability Standard
- OpenTelemetry
- Secrets Management
- HashiCorp Vault
- Access Control
- SSO and RBAC
- Orchestrator
- Mantle
- Developer Interface
- Visual Studio and Code-based (Markdown/Python)
How it works
- Build Skills: Teams design the agent's skills in markdown or using the visual Studio builder, defining tools, personality, and voice.
- Test and Evaluate: Run automated simulations and evaluations to test the agent against common or difficult customer queries to get a performance score.
- Release and Deploy: Deploy the agent via the release pipeline, which runs end-to-end checks and compares versions before going live.
- Improve and Iterate: Use production analytics and real conversation signals to identify patterns and apply recommended changes in the next build cycle.
Plans
- Developer Edition: 0
- Free License: A free license that can be used locally or in production for one bot per company.
- Conversation Limits: Includes up to 1000 external conversations or 100 internal conversations per month.
- Community Support: Access to free community support through the Rasa forum.
- Enterprise
- Full Platform Access: Full access to the Rasa Platform including the visual studio and advanced orchestration.
- Enterprise Scaling: Designed for large-scale deployment with pre-built enterprise security and governance features.
- Success Management: Advanced success planning, best practice guidance, and regular business reviews.
FAQ
Where does the Rasa Platform run?
Rasa Platform is designed for self-managed deployment on-premise or in a private cloud by default, allowing it to run within a customer's existing security infrastructure.
How does Rasa Platform handle agent testing?
The platform supports automated simulations where Rasa can play the role of a customer to test common questions or difficult scenarios, providing a score and recommended fixes.
Can non-developers use Rasa Platform?
Yes, the platform includes a no-code AI assistant flow builder and a visual studio workspace, allowing non-technical product team members to build and monitor agents alongside engineers.
Added by AntSpace from public sources (rasa.com).