Arcana/RAG
This service is currently in a beta phase and can change regularly. The same applies to the documentation.
The Arcana service works together with our Chat AI Service. Compatible models will use RAG (Retrieval-Augmented Generation) in the background to process files and use them as the basis and reference for answers, which often results in better and more factually correct responses. In order to use it, you need to activate your account for the Arcana page and set up an Arcana. Once this is done, you can share it with colleagues or the public. Details about this process can be found in the usage examples.
This is very useful for making large manuals or legal text accessible to an audience that is only interested in smaller aspects of the larger document. An example would be a list of study or exam regulations documents. These can be uploaded and made accessible to students using the Chat AI service. Students can ask questions and get responses that contain information from these documents.
For getting, please check out our Support page, there you will also find a short FAQ.
If all you need are some public Arcana links, please check out our publishes list.
We need your help, if you have created an Arcana which you would like to promote to a larger audience, please reach out to us. We can put is here and communicated it to other users.
Why Use RAG?
The RAG service is designed for businesses and applications that require AI-generated responses to be accurate, explainable, and adaptable to real-world knowledge. Whether it is used for customer support, knowledge management, research, technical guidance, or expert systems, RAG ensures that AI remains intelligent, trustworthy, and useful in dynamic environments.
By integrating real-time data retrieval with AI-powered language generation, RAG transforms AI from a static knowledge tool into a dynamic, continuously learning system-ensuring that responses are always up to date, relevant, and reliable.
Key Terms
- RAG: How the RAG works is described in the RAG Service page
- Docling: The program indexing the files is described in the Docling process page
Service Overview
This Service has two sides. The first is the one where you access a specific Arcana using the Chat AI Service. Similarly, you can share this ID or the access link with others so they can use the RAG for directly accessing specific documents.
The second side is the Arcana manager where the RAG service can be set up. There it is possible to create and set up documents for use in the Arcana service in Chat AI. Additionally, the indexed material can be fine-tuned for specific retrievals if the existing indexing is not good or accurate enough.
It is possible to upload PDF, Text, and Markdown files to be used as base material.
Usage examples
We have written a getting started guide as well as a usage guide. Please refer to these sections directly if you need information about the interface.
The getting started guide details the arcana interface. This includes creating an account if you already have an Academiccloud account as well as setting up an Arcana. It also contains some information about how to generate the ID and tokes as well as how to use the access link.
The usage guide focuses on the Chat AI interface and how to access Arcanas there. It also gives some general guidance on how to do prompt engineering with Arcanas.
You need to use the Meta Llama 3.1 8B RAG model for Arcana to work.
It is in the bottom of the model list.