NOTE: This article was originally published on Arturo Rodriguez Medium blog – see here.
Discover Knodules, the knowledge containers of the Secure Intelligent Knowledge Engine (SIKE), designed to revolutionise information management through modular organisation, sophisticated access controls, and innovative knowledge monetisation. This article examines the functionalities of Knodules, including their ability to enforce compliance, enable precise searches with advanced tagging, and manage data securely and efficiently.
Learn about SIKE’s user-friendly interface, which facilitates effective file management and robust access controls, allowing administrators to monitor resource utilisation and maintain operational efficiency. Explore how Knodules are transforming the landscape of information management.
What are Knodules?
Knodules are the backbone of the Secure Intelligent Knowledge Engine (SIKE), which organises all its information into modular units. These modules, or Knodules, are designed with sophisticated access controls, ensuring that only users with appropriate permissions can view the content. Additionally, SIKE meticulously tracks the costs associated with accessing each Knodule, paving the way for innovative knowledge-based services.
Key Features of Knodules:
- Data Encapsulation: Knodules help in securing data by encapsulating it within well-defined modules.
- Knowledge Monetisation: They provide a framework for monetising information by controlling access and usage.
- Access Restriction: Access to information is strictly regulated, enhancing security and compliance.
Accessing Information in Knodules
Information within a Knodule is retrievable through a Retrieval-Augmented-Generation (RAG) pattern. Here’s how it works in practice:
Example: Enforcing Compliance with Corporate Policies Using PromptFlows
To ensure content generation complies with company policies stored within a Knodule, you can use the following settings in PromptFlows:
- context: $personal$
- thread: 1
- model: gpt-4
- prompt: compliance
rag:
- module: policies
description: |
1. "policies" is a RAG tool.
2. Use the "policies" RAG tool to search for any topic or context related to policies.
3. If there are more than one subject, create several queries, one query per subject.
4. For example if the query is about A and/or B then first ask about A then B.
temperature: 0
instructions: |
rewrite the following text such that it complies with the corporate policies
---
$input$
Advanced Search Capabilities
Each piece of data within a Knodule is tagged with custom categories, allowing for precise and targeted searches. This categorisation aids in deploying advanced vector search techniques within the information stored in Knodules.
Developers can specify desired tags to refine searches and tailor outputs towards “external_communications”, as demonstrated in the following configuration for a PromptFlows query:
rag:
- module: policies
description: |
1. "policies" is a RAG tool.
2. Use the "policies" RAG tool to search for any topic or context related to policies.
3. If there are more than one subject, create several queries, one query per subject.
4. For example if the query is about A and/or B then first ask about A then B.
tags: [external_communications]
SIKE UX
While SIKE provides robust Web API access to all functionalities associated with Knodules, this section highlights how the SIKE user interface facilitates these interactions:
Files
Here’s how users can manage files within the SIKE knowledge base:
- Adding Files: Easily upload files to integrate them into the designated Knodules for structured access and organisation.
- Deleting Files: Remove files as needed to keep the knowledge base current and relevant.
Access & Token Spend Tracking
The user interface enables Knodule administrators to set specific access permissions:
- Member Permissions: Grant read-only access, allowing users to view but not modify the content within Knodules.
- Admin Permissions: Allow designated administrators the capability to add or delete files, providing them with full control over the content.
Administrators can monitor and manage the token usage of each user, which is calculated based on the number of tokens utilised when executing prompts using the Retrieval-Augmented-Generation (RAG) feature on a Knodule. This functionality helps in overseeing resource utilisation and maintaining cost efficiency within the SIKE platform.
Conclusion
In conclusion, Knodules represent a significant advancement in the realm of information management, encapsulated within the Secure Intelligent Knowledge Engine (SIKE). By leveraging modular design, sophisticated access controls, and innovative monetisation strategies, Knodules not only enhance data security and compliance but also introduce a new paradigm in knowledge management. As we continue to explore and refine the capabilities of SIKE, the potential for further innovation remains vast, promising to transform how organisations manage, access, and benefit from their data in an increasingly digital world.
For more information on SIKE, please check out the details here (or email us at we@notcentralised.com).
