Introduction
Docebo’s AI functionality is designed to streamline the creation, tagging, and skill assignment of learning content, in addition to supporting content search and assisting administrators with daily tasks through conversational agents.
This article provides an in-depth look at Docebo’s AI-driven tools, which utilize models from leading AI providers and are built with robust privacy and data security in mind. The following table summarizes Docebo’s AI technology and the providers supporting each feature.
AI features and models
Notes:
- While the table reflects all AI models available to support Docebo’s capabilities, specific models may not be actively used at any given time, depending on development needs and performance optimization.
- Docebo AI refers to open-source models that have been refined and customized by Docebo to better fit the platform’s needs. These models are hosted on AWS infrastructure, ensuring scalability and security.
- “LLMs” refers to: Anthropic and Nova via AWS Bedrock, OpenAI via Microsoft Azure, Gemini and Imagen via Google Vertex AI.
| Feature | Capability | Description | AI technology | Uses generative AI |
|
Creator |
Lesson generation | Automatically generates lesson content based on a prompt or topic | LLMs, AWS Comprehend | Yes |
| Text revision | Adjusts text by modifying length, tone, or phrasing | LLMs | Yes | |
| Assessment generation | Creates activities and assessments based on lesson content | LLMs | Yes | |
| Image tagging | Analyzes images to add text tags | LLMs | Yes | |
| Text translation | Provides multilingual support for content | LLMs | Yes | |
|
AI video presenter |
Creates a video where an AI avatar reads aloud the provided text, appearing as a virtual presenter | Colossyan | Yes | |
| AI roleplay | Allows learners to practice conversation role plays with an AI-powered virtual avatar | LLMs, AWS Transcribe, AWS Polly, Azure AI Speech, Tavus | Yes | |
| Auto skill assignment | Skill matching | Assigns skills based on similarity and confidence level | LLMs | Yes |
| Auto tagging | Keyword tagging | Analyzes content to identify key phrases and automatically generate tags for content organization | LLMs | Yes |
| Content recommendation | Personalized content suggestions | Utilizes individual user activity to recommend relevant content | Docebo AI recommendation model | No |
|
Harmony
|
Conversational agent | AI-powered assistant that helps find information and data, support learning, and get contextual help across the platform. | LLMs, AWS OpenSearch | Yes |
| Global search | Keyword search | Searches and prioritizes results to improve content discoverability | AWS OpenSearch | No |
Privacy and security compliance
Docebo AI models comply with strict data privacy and security standards. Third-party providers align with Docebo’s requirements, ensuring data privacy compliance for all AI functionalities.
- Data handling and usage: All data processed by these AI features remains secure within Docebo’s infrastructure. All customer data is kept private and specific to each customer. Inputs and outputs of the AI systems are managed internally, and third-party access is limited to necessary API calls only.
- Provider compliance: Providers meet industry compliance standards. Docebo performs regular assessments to strengthen data security and privacy compliance.
For more details about Docebo’s information security posture and certifications, please visit our Trust center page (opens in a new tab).
Security monitoring and guardrails
Docebo integrates AWS Bedrock Guardrails, Azure OpenAI Content Filter, and Google Vertex AI Content and Safety Filter to prevent unauthorized or harmful outputs generated by large language models (LLMs).
Customers maintain control over AI system settings, with options to adjust or disable features if necessary, with the exception of global search, which is a core functionality of the platform and cannot be disabled.
Continuous testing
Docebo AI models undergo thorough assessment during development to ensure quality and effectiveness. After release, ongoing automated tests monitor system functionality to ensure consistent performance and reliability.