FAQ

The platform is available in English at first. We will gradually add a multi-language mode allowing you to set up the language of your choice.

MLOps is a set of practices that aims to deploy and maintain machine learning models in production in a reliable and efficient manner. The LLMOps is based on the MLOps and adds all the functionalities to specialize, operationalize and manage LLMs.

The platform, currently in version 1.1, makes it possible to set up compute and storage infrastructures as well as to create and deploy Machine Learning and fine-tuning pipelines, and to operationalize and manage LLMs. So it's what we can call an MLOps and LLMOPs platform.

You pay for a subscription that gives you access to all the functionalities of the MLOps platform, present and future. On top of that, you'll be billed each month based on how you use compute and storage resources.

Follow us on Linkedin & Twitter and visit our blog regularly where we regularly publish articles on artificial intelligence.

The Craft AI platform is fully compliant with the European Data Regulation (GDPR) and has numerous functionalities to assist in the management of the life cycle of personal data.

You have access in real time to the daily cost of each of the machines that are used. You can easily change settings, add GPUs, and visualize your infrastructure costs by project.

We work with numerous cloud providers: AWS, GCP, Azure, Scaleway, OVH. We can set up your accommodation at your convenience with each of these actors. We provide data hosting within the European Union.

The Craft AI platform is available in SaaS format and you can also install Craft AI on your own cloud servers.

You can import all types of data into the platform, whether structured or unstructured. You can easily create flows that allow for the regular ingestion of new data.

Rights make it possible to limit access to various projects, production and infrastructure management.

You can create as many projects as you want by subscribing to our platform.

You can develop your code in Python at first, but we also plan to add R, JavaScript, and Julia.

You can deploy your pipelines using an endpoint that allows you to share the results with an external user or by configuring prior execution rules.

The platform is compatible with all open-source Data Science frameworks and libraries and with ML platforms.

A privacy-friendly AI ensures compliance with the GDPR or other similar regulations and uses summary data to ensure total data confidentiality.

Summary data is artificial data that does not contain real events. It is generated from real data so it remains statistically equivalent and therefore allows Machine Learning operations to be carried out.

A frugal AI system minimizes the amount of energy needed to train, deploy, and re-train models.

Fair AI removes biases and discriminations that may be inherently present in data.

A trusted AI is an AI that is more likely to be accepted and used by users. In our opinion, it must be explainable, respectful of privacy, energy-efficient and fair.