[n8n] Multi-Confidentiality-Class RAG System With Builtin Routing, Authentication, and Authorization

What problem does it solve, and how does it solve it:

While RAGs can provide powerful solutions, they still introduce a significant problem, even with the most secure designs, RAGs, are inherently probabilistic, even with restricted access, an employee who doesn't have clearance for certain sensitive data could unintentionally (or purposefully) expose it through the right prompt.

No matter how careful the design is, AI models, by nature, aren’t foolproof in preventing information from being exposed, using AI or particularly RAGs for anything involving sensitive business data without additional protection is literally gambling with your most valuable assets.

Even the best LLMs and RAGs can't completely eliminate the risk of accidental or intentional data leaks, making it a potential vulnerability in your organization's security.

The solution is very simple, if you don’t want an AI model to spill your data, don’t give it that data to begin with. The RAG will not be able to leak it, because it doesn’t have it

1. First, we classify our data into a discrete set of categories, let’s say A, B and C, where A is the most confidential one, and C is the least confidential.
2. Then we build isolated RAGs, one for each data class
3. We give each employee an access level in the system
4. Every employee prompt will go through a Router, that will route it to the correct RAG, while rejecting the ones with insufficient access level

This product is a highly customizable implementation of the above solution, including a setup guide (within the workflow). Built for enterprises with varying confidentiality needs, this system ensures that sensitive data is handled with the utmost care, while allowing for flexible, scalable document retrieval and generation.

Key Features:

• Fine-grained control over data access. It ensures that sensitive information is retrieved and processed only when appropriate, offering robust privacy controls across various levels.
• Clear isolation of data across RAGs, making sure sensitive data is impossible to be retrieved if the user doesn't have the required permissions
• The routing system automatically directs documents to the appropriate RAG based on their confidentiality level, ensuring seamless interaction with the user
• With authentication and authorization mechanisms, the system ensures that only authorized users can access confidential data.
• Powerful retrieval capabilities from your knowledge base, paired with context-aware generation powered by AI, the system pulls the most relevant documents, ensuring that your answers are both accurate and tailored to your specific queries.
• Scalable and customizable to your organizational needs