Evaluation record · rasa

Rasa (Open Source & Rasa Pro)

vRasa Pro 3.16.x; Rasa Open Source 3.x (maintenance mode)

Rasa Technologies

Agentnlpconversationalopen-source

Strong

About This Agent

Open-source-rooted conversational AI framework for building contextual assistants with full control over NLU, dialogue management, and deployment. NOTE: classic Rasa Open Source (Apache 2.0) is in maintenance mode; Rasa's strategic direction is CALM, its LLM-native dialogue approach, delivered via the commercially licensed Rasa Pro (3.16.x as of mid-2026) and the no-code Rasa Studio. A free Rasa Pro Developer Edition covers 1,000 conversations/month; paid tiers start at ~$35,000/year.

Last Evaluated: July 9, 2026

Official Website

Trust Vector Analysis

Dimension Breakdown

🚀Performance & Reliability

intent classification

Intent accuracy benchmarking

Evidence

Rasa NLU — Customizable NLU pipeline with multiple classifiers

highVerified: 2026-07-09

dialogue management

Conversation flow testing

Evidence

Rasa Core — Machine learning-based dialogue policies and story-based training

highVerified: 2026-07-09

entity extraction

Entity extraction testing

Evidence

Entity Extractors — Multiple entity extractors (CRF, DIET, SpaCy, Duckling)

highVerified: 2026-07-09

custom actions

Custom action testing

Evidence

Custom Actions — Full Python SDK for custom business logic and API calls

highVerified: 2026-07-09

context handling

Context retention testing

Evidence

Slot Management — Slot-based context tracking with influence on dialogue

mediumVerified: 2026-07-09

latency

Performance benchmarking

Evidence

Performance — Latency depends on infrastructure and model complexity

mediumVerified: 2026-07-09

🛡️Security

self hosted control

Deployment security assessment

Evidence

Deployment Options — Full control with on-premises or private cloud deployment

highVerified: 2026-07-09

authentication

Authentication capabilities review

Evidence

Security Features — Token-based authentication, requires custom implementation

mediumVerified: 2026-07-09

data privacy

Data flow analysis

Evidence

Open Source — All data stays on your infrastructure, no third-party sharing

highVerified: 2026-07-09

open source transparency

Open source assessment

Evidence

GitHub Repository — Apache 2.0 license, 18k+ stars, full source code available

highVerified: 2026-07-09

encryption

Security features review

Evidence

Documentation — Encryption implementation is user's responsibility

mediumVerified: 2026-07-09

🔒Privacy & Compliance

data retention

Privacy architecture review

Evidence

Tracker Store — Full control over conversation storage and retention

highVerified: 2026-07-09

gdpr compliance

Compliance capabilities assessment

Evidence

Self-Hosted — GDPR compliance achievable with proper configuration

highVerified: 2026-07-09

on premise deployment

Deployment options assessment

Evidence

Deployment — Complete on-premises deployment with no external dependencies

highVerified: 2026-07-09

pii handling

PII protection assessment

Evidence

PIIEntityExtractor — PII detection available, anonymization requires custom actions

mediumVerified: 2026-07-09

no data sharing

Data flow analysis

Evidence

Open Source Model — No telemetry or data sharing with third parties

highVerified: 2026-07-09

👁️Trust & Transparency

documentation quality

Documentation completeness review

Evidence

Rasa Docs — Excellent documentation with tutorials and best practices

highVerified: 2026-07-09

open source code

Open source assessment

Evidence

GitHub — Apache 2.0, 18k+ stars, active community contributions

Rasa documentation (product split) — Rasa Open Source is in maintenance mode; active development has moved to CALM in the commercially licensed Rasa Pro (3.16.x) and Rasa Studio; legacy OSS remains Apache 2.0

highVerified: 2026-07-09

conversation debugging

Debugging tools assessment

Evidence

Debug Mode — Interactive learning and conversation debugging tools

highVerified: 2026-07-09

community support

Community engagement analysis

Evidence

Community — Active forum, Discord, and community contributions

highVerified: 2026-07-09

explainability

Explainability features assessment

Evidence

Policy Insights — Policy confidence scores available but limited explainability

mediumVerified: 2026-07-09

⚙️Operational Excellence

ease of integration

Integration complexity assessment

Evidence

Connectors — Built-in connectors for Slack, Facebook, Telegram, REST API

highVerified: 2026-07-09

scalability

Scalability testing

Evidence

Scaling Guide — Horizontal scaling possible with container orchestration

mediumVerified: 2026-07-09

cost predictability

Pricing model analysis

Evidence

Open Source — Free open source, costs only for infrastructure

Rasa Pro licensing — Free Rasa Pro Developer Edition (up to 1,000 conversations/month, 100/month for internal employee agents); Growth tier from ~$35,000/year; Enterprise custom; legacy OSS remains free (Apache 2.0) but in maintenance mode

highVerified: 2026-07-09

monitoring

Monitoring features assessment

Evidence

Monitoring — Basic metrics endpoint, requires external monitoring tools

mediumVerified: 2026-07-09

production readiness

Production readiness assessment

Evidence

Deployment Guide — Requires DevOps expertise for production deployment

mediumVerified: 2026-07-09

training pipeline

Training capabilities assessment

Evidence

Training — Flexible training pipeline with customizable components

highVerified: 2026-07-09

Strengths

+Open-source roots (Apache 2.0 legacy) with complete control over code and data; CALM in Rasa Pro adds LLM-native dialogue while keeping business logic deterministic
+On-premises deployment option for maximum privacy and compliance
+No vendor lock-in or usage-based pricing, only infrastructure costs
+Highly customizable NLU pipeline and dialogue policies
+Active community with 18k+ GitHub stars and extensive resources
+Machine learning-based dialogue management for contextual conversations

Limitations

!Rasa Open Source is in maintenance mode: new capabilities (CALM, LLM-native dialogue, Rasa Studio) ship only in the commercially licensed Rasa Pro
!Requires significant ML and DevOps expertise for deployment
!Limited enterprise features out-of-box (auth, monitoring, analytics)
!Steeper learning curve compared to managed cloud services
!Production deployment and scaling requires infrastructure knowledge
!No built-in analytics dashboard, requires external tools
!Training data quality critical for good performance

Metadata

license: Apache 2.0 (legacy Rasa Open Source, maintenance mode); Rasa Pro is commercially licensed

supported models

0: DIET

1: TED Policy

2: Transformer Embedding Dialogue

3: CALM with pluggable LLMs (Rasa Pro)

programming languages

0: Python

deployment type: Self-hosted (on-premises, cloud, containers)

tool support

0: Custom actions (Python)

1: REST API

2: Webhooks

pricing model: Legacy OSS free (maintenance mode); Rasa Pro: free Developer Edition, Growth from ~$35,000/year, Enterprise custom

github stars: 20700+

first release: 2016

supported channels: Slack, Facebook, Telegram, Web, Custom REST

enterprise version: Rasa Pro with CALM + Rasa Studio (commercial license)

pricing: Free Rasa Pro Developer Edition (1,000 conversations/month; 100/month for internal employee agents), Growth tier from ~$35,000/year, Enterprise custom

version: Rasa Pro 3.16.x; Rasa Open Source 3.x (maintenance mode)

Use Case Ratings

customer support

Excellent for custom support bots with full control

code generation

Not designed for code generation use cases

research assistant

Conversational focus, limited research capabilities

data analysis

Can integrate with analytics via custom actions

content creation

Focused on conversations, not content generation

education

Good for educational chatbots with customization

healthcare

On-premises deployment ideal for HIPAA compliance

financial analysis

Self-hosted option meets compliance requirements

legal compliance

Would need significant custom development

creative writing

Not optimized for creative tasks