Introduction
Docebo’s AI functionality is designed to streamline the creation, tagging, and skill assignment of learning content. This article provides an in-depth look at Docebo’s AI-driven tools, including AI authoring, Docebo Shape, auto skill assignment, and auto tagging. These capabilities 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.
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.
| Feature | Capability | Description | AI technology providers | Uses generative AI |
| AI authoring | Text revision | Adjusts text by modifying length, tone, or phrasing | Anthropic via AWS Bedrock, OpenAI via Microsoft Azure | Yes |
| Assessment generation | Creates activities and assessments based on lesson content | Anthropic via AWS Bedrock, OpenAI via Microsoft Azure | Yes | |
| Image tagging | Analyzes images to add text tags | Anthropic via AWS Bedrock, OpenAI via Microsoft Azure | Yes | |
| Docebo Shape | Summarization | Breaks down complex text into digestible segments for structured learning | Docebo AI | No |
| Keyword extraction | Extracts key terms and phrases to support scene-based learning | Docebo AI | No | |
| Semantic image matching | Matches text with relevant images to enhance visual engagement in learning scenes | Docebo AI | No | |
| Text translation | Provides multilingual support for content | AWS Translate | No | |
| Audio generation | Converts text to speech to create voiceovers | AWS Polly | No | |
| Auto skill assignment | Skill matching | Assigns skills based on similarity and confidence level | Anthropic via AWS Bedrock, OpenAI via Microsoft Azure | Yes |
| Auto tagging | Keyword tagging | Analyzes content to identify key phrases and automatically generate tags for content organization | Google Cloud Natural Language AI | No |
| Content recommendation | Personalized content suggestions | Utilizes individual user activity to recommend relevant content | Docebo AI recommendation model | No |
| Global search | Content search | Searches and prioritizes results to improve content discoverability | Elasticsearch engine | No |
Privacy and security compliance
Docebo AI models comply with strict data privacy and security standards. Third-party providers (AWS, Microsoft Azure, Google) 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 Guardrail and Azure OpenAI Content Filter to prevent unauthorized or harmful AI-generated results.
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.