Details about AI models, services, and learning
At Kontent.ai, we believe in full transparency about how our AI accelerates your content creation and operations. We detail which AI models we use, how they process your data, and other resources about the AI governance policies that guide our responsible use of these technologies.
Models and services in use
| Service | Model | Publisher |
| Translation | DeepL API | DeepL |
| Translation fallback, text transforms, and generation | GPT-4o | OpenAI |
| Text embeddings (search and recommendations) | text-embedding-3 | OpenAI via Azure AI |
| Image descriptions | Azure AI Vision | Microsoft |
| Image tagging | Azure AI Vision | Microsoft |
How are they used
Below is an overview of how each AI model and service is integrated into Kontent.ai to power the AI capabilities, streamline operations, and enhance user experiences.- DeepL is primarily used to translate content.
- GPT-4o is used for these operations:
- Translation fallback
- Text transformations
- Content generation
- text-embedding-3 is used to generate embeddings to index your content and provide recommendations.
- They are stored and served from Azure AI for low-latency global search.
- Azure AI Vision is used for image descriptions and image tagging.
Learning
We don’t use your data to train models. Our AI models never learn from or store your content. The only automatic processing is the necessary indexing and vectorization that powers search functionality. This is purely for search performance and not part of our AI-driven capabilities.Contact and other resources
If you have further questions about our AI practices, you can contact our security team. Visit the Kontent.ai Trust Center
for comprehensive resources about our AI practices and security compliance. You can also find more information about our responsibility with AI.