Description
Serverless AI hosted in France, grounded in your own content
OVHcloud AI Endpoints is a serverless platform that gives you API access to a selection of pre-trained AI models. The catalog spans language models, including vision, reasoning and coding models, as well as embeddings, translation and audio models. The API is designed to be compatible with the OpenAI API, so existing applications and libraries need few changes.
Models do not know your products, your approved texts or the images you may use. Kontainer DAM & PIM holds exactly that.
When an app or agent that calls AI Endpoints is connected to Kontainer through the Kontainer API or MCP, it can fetch what it needs for each request: product data, metadata and file information, straight from the source.
OVHcloud states that the AI Endpoints infrastructure is located in Gravelines, France, and that data is not stored or shared during or after model use.
Benefits
Why integrate OVHcloud AI Endpoints with Kontainer?
✔ Models hosted in France
The AI Endpoints infrastructure is located in Gravelines, France, which suits organizations that want their AI workloads to stay in Europe.
✔ No infrastructure to run
Because the platform is serverless, you call a model through the API without provisioning or maintaining servers.
✔ Data is not stored
OVHcloud states that data is not stored or shared during or after model use.
✔ Access keys per project or team
Create separate API access keys for different projects or teams, each with its own validity period.
✔ Insight into usage
Model metrics show the number of calls per model and the tokens used, which helps you understand and manage costs.
✔ Kontainer as your Single Source of Truth
Let your AI solution work from the same approved product data and files as the rest of the organization.
Features
How OVHcloud AI Endpoints and Kontainer work together
✔ The Kontainer API
Retrieve and update product data, metadata and file information from apps and functions that call AI Endpoints.
✔ MCP for agents
An agent built on models from AI Endpoints can use Kontainer as a tool through the Model Context Protocol, without custom development for every agent.
✔ Function calling
AI Endpoints lets a model invoke pre-defined functions, so it can trigger a lookup in Kontainer as part of answering a question.
✔ Structured outputs
With JSON schema output, a model can return suggestions in a fixed structure, which makes it straightforward to map them onto Kontainer’s fields.
✔ Embeddings for search and RAG
Embedding models can turn product texts and metadata from Kontainer into vectors for semantic search or a knowledge base that models draw on.
✔ Vision and translation models
Where the chosen model understands images, it can suggest descriptions for assets. Translation models can help adapt product texts to other languages.
✔ Product data and files from PIM and DAM
Use names, descriptions, specifications, categories, variants, images, videos and documents, together with their tags and metadata.
Set-up
Connect OVHcloud AI Endpoints and Kontainer in three steps
- Create an access key at OVHcloud
AI Endpoints access keys are linked to a Public Cloud project. Create one for the solution you are building. - Give your solution access to Kontainer
Create a JSON API token, or set up OAuth 2.0, for the Kontainer user the solution should act as. Access is limited to what that user may do. - Let the model reach Kontainer
Connect through MCP or expose Kontainer lookups as functions, ask for structured output where you want to write data back, and test with real product questions.
The set-up requires some technical or IT knowledge. The MCP guide describes the details.
Use Cases
What can you build with OVHcloud AI Endpoints + Kontainer?
- A semantic product knowledge base
Let staff and customers search by meaning across product texts, specifications and documents. - Structured product enrichment
Generate suggestions in a fixed format and review them before they are written to the right fields. - Alt text and tag suggestions
Use a vision model to propose descriptions and tags for new images, for an employee to approve. - Multilingual product content
Adapt product texts to new markets, starting from localized data in Kontainer. - A product advisor
An assistant for sales, support or your webshop that answers from approved PIM data. - Agents across systems
Combine Kontainer with your ERP, CRM and other systems in the same workflow.
Support
Get started with OVHcloud AI Endpoints + Kontainer
Which data and workflows your AI solution needs decides how to set things up. Contact our support team for help with set-up, customization or best-practice advice.
You can write to us at support@kontainer.com
Read more about the Kontainer API and MCP for Kontainer DAM and PIM.
See OVHcloud’s AI Endpoints page and its documentation on features, capabilities and limitations.
For EU AI Act requirements, see Labelling AI-generated content in Kontainer and the EU AI Act (Regulation (EU) 2024/1689).
FAQ
What is OVHcloud AI Endpoints?
OVHcloud AI Endpoints is a serverless platform from the French cloud provider OVHcloud. It gives API access to pre-trained AI models, including language, vision, embedding, translation and audio models, through an API designed to be compatible with the OpenAI API.
Can OVHcloud AI Endpoints be integrated with Kontainer?
Yes. Kontainer can be connected to solutions built on AI Endpoints through the Kontainer API and MCP, so models and agents can retrieve product data, metadata and file information from Kontainer DAM & PIM.
Can agents on AI Endpoints use Kontainer via MCP or function calling?
Yes. An agent built on models from AI Endpoints can use an MCP client to reach Kontainer as a tool. Alternatively, Kontainer lookups can be exposed as functions, since AI Endpoints supports function calling.
Can AI Endpoints use product data and files from Kontainer?
Yes. Product names, descriptions, specifications, categories and variants from PIM can be used as context, and images, videos and documents in DAM can be found through metadata and tags.
Can I build semantic search or RAG on Kontainer content?
Yes. Embedding models available on AI Endpoints can turn product texts and metadata from Kontainer into vectors for semantic search or a knowledge base, which a model then draws on when it answers.
Where is data processed, and is it stored?
OVHcloud states that the AI Endpoints infrastructure is located in Gravelines, France, and that data is not stored or shared during or after model use. What reaches the model is decided by your solution, and access to Kontainer is limited to what the Kontainer user behind the connection may do.
Does our data need to be trained into an AI model?
Not necessarily. An integration can retrieve the relevant information from Kontainer when it is needed, through MCP or the API. That lets you work with current data without making it part of a model’s training.
Can I use my existing OpenAI-compatible tools?
Yes. AI Endpoints is designed to be compatible with the OpenAI API, and Kontainer is connected through its own API or MCP regardless of which model you choose.
How do we get started?
Start by clarifying which data and workflows the AI solution needs. Then contact Kontainer’s support team at support@kontainer.com for help with set-up, customization and best practices.