R

Chatbot builders

Rasa: Open-Source Platform for AI Agents

rasa.com · checked 21 Oct 2026

Open-source, self-hostable platform for building and running AI agents in voice and chat.

How it plugs in

—Public API
YesOpen source
YesFree tier

Signal on Chatbot builders8.2#1 of 257

Rasa provides a self-hostable foundation for enterprise agents working across voice and chat. Teams can create agents in code or visually, link business systems, ground responses in company knowledge, and preserve context across turns.

Rasa home page

Review

In a line

Rasa gives organizations an open-source way to build, deploy, and operate voice and chat AI agents. It is made by Rasa, founded in 2016 and based in San Francisco.

What it does

Agents may be authored with code or assembled in Rasa's visual assistant flow builder. Rasa handles language understanding for multilingual agents, manages dialogue state and memory, and can answer from connected knowledge sources.

It supports bot generation, knowledge-base answers, human handoff, and custom actions. Through an action server, those actions can invoke APIs, query databases, or execute business logic. For quality work, Rasa offers end-to-end tests, an interactive panel, conversation review, rollback controls, and model fine-tuning recipes.

How it connects

Rasa is open source under the Apache 2.0 license, so it can be used in commercial software without a license charge. An optional HTTP API can train models, send messages, and run tests when enabled with the --enable-api option. A Python SDK supports custom actions; other languages can use the HTTP action-server contract.

Channels available out of the box include Slack, Facebook Messenger, Twilio, Telegram, Microsoft Bot Framework, Cisco Webex, Rocket.Chat, Mattermost, and Google Hangouts. Backend connections include Zendesk, Salesforce, ServiceNow, HubSpot, and databases such as MongoDB, DynamoDB, and Oracle.

It runs on Linux, macOS, Windows, the web, and self-hosted setups. Deployment can be on your own premises or private cloud, through a managed service, or on Kubernetes and OpenShift with Helm charts and multiple replicas.

Good fit

Rasa suits teams that need control over agent code, data location, and business logic. With self-hosting, customer data remains on infrastructure chosen and operated by the customer.

It fits multilingual voice or chat projects that need enterprise search, context across turns, custom integrations, and scaling. Teams can also review documentation covering installation, tutorials, API use, deployment, and SDK guidance.

Watch for

Reviewers often find sophisticated Rasa projects difficult to learn. The visual builder does not remove the need for code in every scenario, especially when custom actions or complex integrations are required.

Check system details before deployment. Legacy open-source documentation lists Python 3.7 through 3.10 and setup guidance for Ubuntu, macOS, and Windows. The documentation also says Duckling containers and the ConveRTFeaturizer lack native Apple Silicon support.

Support varies by tier: Basic support offers web-portal break-fix help, while premium support adds 24/7/365 response, customer-success staff, and planning services.

Pricing

Rasa follows a freemium model. Its Free Developer Edition covers one bot per company, allows up to 1,000 external or 100 internal conversations each month, and includes community forum support.

Rasa Pro and the Business plan have no listed price. Rasa Pro adds CALM dialogue management, enterprise search, channel connectors, observability, and PII data management. The Business plan adds Rasa Studio, a no-code flow builder, conversation analysis, SSO, and role-based access control.

Enterprise pricing is available on request and includes the full platform, premium support, large-scale deployment, and enterprise security features.

Verdict

Choose Rasa if your organization wants an open-source foundation, self-hosting, multilingual language understanding, and direct control over integrations and data. The free developer plan can support a small starting point.

Look elsewhere if you need a purely no-code tool, want a short learning path for a complex project, or require a published price for advanced plans.

Our take

Rasa is a practical choice for organizations that need to host an AI agent themselves, connect it to internal systems, and shape its behavior with code. Its open-source license and free developer edition lower the barrier to starting. The trade-off is complexity: advanced work may demand Python, action-server logic, deployment skills, and careful platform checks. Paid plans do not publish standard prices, so budgeting requires a conversation with Rasa. Teams seeking a simple, fully visual build process may find another chatbot product easier.

  • Commercial software can use the Apache 2.0 open-source license without a license charge.
  • Agents can be authored through code or Rasa's visual assistant flow builder.
  • Connections reach major chat channels, business systems, and databases.
  • Deployment options include self-hosting, managed service, and scalable environments.
  • The free plan covers one bot with monthly conversation limits.
  • Reviewers often find sophisticated Rasa projects difficult to learn.
  • Some custom actions and integrations require code or an action server.
  • Apple Silicon does not natively support Duckling containers or ConveRTFeaturizer.
  • Free plan limits usage and includes community forum support only.
  • Pick it ifTeams building multilingual enterprise agents with self-hosting needs; Projects needing custom business logic and internal data connections; Organizations comfortable managing code and deployment
  • Skip it ifYou need a fully no-code experience for complex agent behavior.; You require clear public prices for paid enterprise plans.; Your deployment depends on native Apple Silicon support for listed components.

Pricing

The free developer plan covers one bot and limited monthly conversations.

  • Free Developer Editionone bot per company · up to 1,000 external conversations/month or 100 internal conversations/month · community forum supportFree
  • Rasa ProCALM dialogue understanding and management · Enterprise search · Channel connectors · Observability · PII data managementOn request
  • Business planRasa Pro plus Rasa Studio · No-code AI assistant flow builder · Conversation analysis · SSO · Role-based access controlOn request
  • Enterprisefull Rasa Platform access · premium support · large-scale deployment · enterprise security featuresOn request

As rasa.com lists them, checked 21 Oct 2026.

Questions

Is Rasa open source?

Yes. Rasa Open Source uses the Apache 2.0 license and may be used in commercial software without charge.

Does Rasa have a free plan?

Yes. One bot per company is included, with up to 1,000 external or 100 internal conversations per month.

Can Rasa connect to business software?

Yes. Backend connections include Zendesk, Salesforce, ServiceNow, HubSpot, MongoDB, DynamoDB, and Oracle.

Where can Rasa run?

Rasa supports Linux, macOS, Windows, web, self-hosted, on-premises, private-cloud, managed, Kubernetes, and OpenShift deployments.

Does Rasa offer an API?

An optional HTTP API can train models, send messages, and run tests when enabled with the --enable-api option.

What support is available?

Basic support provides web-portal break-fix help. Premium support adds 24/7/365 response, customer-success staff, and planning services.