Rasa
Open-source conversational AI platform and enterprise framework with full channel coverage, dialog orchestration, trustworthy-AI positioning, and self-hosted or SaaS deployment.
Rasa is a framework and platform for building conversational AI agents, text and voice assistants that a company hosts and controls itself. It has long been one of the best-known open-source options in this space, rooted in machine learning for understanding language and managing dialogue, and its current generation centers on an approach the company calls CALM, short for Conversational AI with Language Models, which marks a shift away from the older method of hand-labeling intents and training phrases.
The idea behind CALM is a deliberate separation of concerns. A language model handles the fluent part of a conversation, interpreting what a user means and gracefully handling things like topic changes, corrections, and clarifications, while the actual steps the agent is allowed to take are defined by the developer as Flows, structured business processes broken into clear steps. The model understands the user, but it does not invent the business logic. That design keeps execution deterministic and debuggable and is intended to make the agent resistant to hallucination, prompt injection, and jailbreaking, which is a large part of why regulated industries gravitate toward it. CALM can run with frontier models or with smaller fine-tuned models for lower latency and cost, and it can operate fully on-premises with no calls to external LLMs.
The broader Rasa Platform splits into a pro-code framework and infrastructure called Rasa Pro, with enterprise security, analytics, and observability, and a no-code interface called Rasa Studio for business users, so technical and non-technical teams can work on the same agents. A browser-based playground lets teams prototype CALM agents quickly, and a free developer edition lets them build and run within set limits. Agents connect to backend systems through custom actions and standards like the Model Context Protocol, coordinate with other agents, and deploy across web, mobile, messaging channels, and voice with built-in turn-taking and latency control.
What consistently distinguishes Rasa is control. It is built to be self-hosted on a team's own infrastructure or in the cloud, giving full ownership over data, models, and behavior, which suits organizations with strict privacy, compliance, or reliability requirements that are not comfortable handing conversations to a fully managed third party.
Vendor details
Canonical URL
https://rasa.com
Category
Multi-agent platform
Company status
independent
Use cases & customers
Target customers
Deployment options
In practice
You need a chatbot that can't improvise its way into a compliance problem. Rasa's CALM lets a language model interpret the user while you define the exact steps it can take as Flows, keeping execution deterministic and auditable.
Your data can't leave your infrastructure. Rasa is built to be self-hosted on-premises, can run with small fine-tuned models and no calls to external LLMs, and gives you full ownership of models and behavior.
Your technical and business teams both need to work on the assistant. Rasa pairs a pro-code framework with a no-code Studio, plus a browser playground to prototype CALM agents and a free developer edition to start.
Sources & related URLs
Research sources
Agentic Index coverage score
12.5 / 14 capabilities · 89%
| Integrations & Tool CallingOfficial docs 2026-06-08 | Full |
|---|---|
| Workflow OrchestrationDialog orchestration docs 2026-06-08 | Full |
| Knowledge Grounding & RAGOfficial docs 2026-06-08 | Full |
| Human Oversight & GuardrailsTrustworthy AI docs 2026-06-08 | Full |
| Security, Identity & GovernanceEnterprise docs 2026-06-08 | Full |
| Observability & AuditabilityOfficial docs 2026-06-08 | Full |
| Memory & State PersistenceStateful dialog docs 2026-06-08 | Full |
| Deployment & Data ResidencySelf-hosted and SaaS docs 2026-06-08 | Full |
| Prebuilt Agents, Templates & PacksOfficial docs 2026-06-08 | Partial |
| Triggers & Channel CoverageChannel coverage docs 2026-06-08 | Full |
| Model Flexibility & RoutingMulti-model OSS docs 2026-06-08 | Full |
| APIs, SDKs & MCP ExtensibilityOfficial docs 2026-06-08 | Full |
| Testing, Debugging & OptimizationOfficial docs 2026-06-08 | Full |
| Browser & Computer UseOfficial docs 2026-06-08 | Unable to verify |
The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Recent platform changes
Rasa Pro 3.17.3 was issued to urgently backport the removal of the LangChain dependency cluster, addressing a critical arbitrary local file read vulnerability within the LangSmith TracingMiddleware. It also fixes missing input and output spans in Langfuse tracing.
Bears on: Security / enterprise
View sourceRasa Pro 3.18.0 removes the LangChain dependency cluster to patch a critical file-read vulnerability, migrating vector stores to native SDKs and mandating a model retrain. The release also adds a new Capabilities API endpoint for dynamic conversational flows and a local document retrieval backend for the Rasa copilot.
Bears on: MCP / tool calling / API
View sourceRasa Pro 3.17.0 added native Langfuse integration for LLM observability, auto-instrumenting paths like message processing and MCP tool calls. The release also brought token-by-token streaming for custom actions and a new API endpoint for dynamic capability discovery.
Bears on: Observability / auditability
View sourcePricing
Free developer edition · Growth from $35k/yr
subscription (by conversations)
Included quota
Developer Edition is free with full access to Rasa Pro and CALM, roughly 1,000 conversations a month (about 100 a month for internal employee assistants) and community support. Growth adds the full Rasa Platform, including Rasa Studio, higher volume up to 500,000 conversations a year, and basic support, from about $35,000 a year. Enterprise adds premium support, prebuilt enterprise security, and large scale deployment at custom pricing.
What is public
Public: the free Developer Edition and its conversation limits, that the Growth tier starts around $35,000 a year for teams under 500,000 conversations annually, and that Enterprise is custom. Not public: exact Growth and Enterprise dollar figures beyond the starting point, which are quoted by sales.
Billing mechanics
An annual platform subscription tiered by conversation volume. The free Developer Edition gives individual developers full access to Rasa Pro and CALM within a low monthly conversation limit. Growth is an annual license starting around $35,000 for teams under 500,000 conversations a year, and Enterprise is a custom annual contract. Rasa is self hosted or run as SaaS, so model and infrastructure costs sit with the customer.
Cost watchouts
Rasa is self hosted, so infrastructure and, when using frontier models, LLM inference costs are additional to the platform license and are borne by the customer. Exceeding the conversation tier moves you to a higher and more expensive plan. Growth and Enterprise pricing beyond the starting point is quoted by sales.
Variable cost rationale
The Rasa platform fee is a flat annual subscription tiered by conversation volume, which is predictable, but Rasa is self hosted, so model inference and infrastructure costs sit with the customer and scale with usage. CALM can run on smaller on premises models with no external LLM calls to hold marginal cost down, so exposure is moderate and largely within the customer's control.
Additional watchouts
Beyond the free Developer Edition, paid pricing is quote based, with Growth starting near $35,000 a year. Self hosting means the customer owns infrastructure and model costs. The free tier is capped at a low monthly conversation volume.
Overage / add-ons
Conversation volume is bounded by the plan tier; exceeding it requires moving to a higher tier or a custom contract agreed with sales.
Sales call required
Mixed (some tiers require a call)
Free / trial
A free Developer Edition gives individual developers full access to Rasa Pro and CALM within a monthly conversation limit, with community support.
Lowest paid plan
Growth from about $35,000 a year for teams under 500,000 conversations annually, above a free Developer Edition.
Commercial notes
Positioned as the control first, self hosted option for regulated industries: full ownership of data, models, and behavior, with CALM keeping execution deterministic. The tradeoff is engineering ownership and a platform license that starts in the tens of thousands annually once you move past the free Developer Edition.
Key ambiguities
Only the Growth starting figure of about $35,000 a year is public; the full Growth and Enterprise pricing is quote based and depends on volume, support, and security requirements. Self hosted infrastructure and model costs are separate.
Support SLA / resale
Developer Edition has community support. Growth includes basic support. Enterprise includes premium support with enterprise SLAs.
Missing data
Exact Growth pricing above the $35,000 starting point and Enterprise dollar figures are not public and are quoted by sales.
Related vendors
- CrewAI — Open-source multi-agent orchestration framework for building…
- Relevance AI — Platform for building and deploying a workforce of AI agents with a…
- Agent Zero — Open-source multi-agent framework with its own Linux-VM agent,…
- Akka — Enterprise agentic AI platform that builds and runs multi agent…
- Artian — Enterprise multi agent AI platform for financial services: a Planner…
- Databricks Mosaic AI — Databricks Mosaic AI platform for building, deploying, and…
Alternatives to Rasa
The closest documented capability profiles to Rasa among multi-agent platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Akka12.0 / 14A lighter documented profile than Rasa
- CrewAI13.0 / 14Adds documented Browser & Computer Use
- Databricks Mosaic AI12.0 / 14A lighter documented profile than Rasa
- Distyl AI11.5 / 14A lighter documented profile than Rasa
- Mindra11.5 / 14A lighter documented profile than Rasa
- Dataiku12.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded