A front end to the @Boxer app - multi-user access to a GPT prompt, backed by a store of AI documents.
@Boxer is an AI-enabled Learning Management System (LMS). The objective is to be able to build a curriculum of content by processing open-source documents from the web (YouTube videos, GitHub repositories, and plan HTML text) and loading AI generated summaries into a document store. A simple front end then enables students to navigate through the content, can answer questions based on the embedded content, and recommend next steps once the student is familiar with a certain level of content.
The specific domain is to teach participants how to build AI applications using modern Large Language Model (LLM) technology, and the current approaches to this - Retrieval Assisted Generation (RAG), and multi-step workflows using the LLM to generate summaries and process questions.
The benefits of this approach are:
- It is simple to maintain content, given that the field is moving so rapidly. Traditional approaches of generating bespoke new content are often obsolete by the time they are ready.
- Participants get a flavour of what is possible with modern AI by using the tools.
This website is written in HTML5 and Bootstrap (https://getbootstrap.com/). Azure functions are used at runtime, using node.js (https://nodejs.org/en/download/). As far as possible, these are very 'thin' functions - its quite hard to debug. All complex logic we try to move to the front end where it can more easily be unit tested, debugged etc.
GNU AFFERO GENERAL PUBLIC LICENSE.
This is intentionally a restrictive licence. The source is effectively available for non-commercial use (subject to the licence terms as listed, which enable use for learning, self study etc). Commercial use either must abide by the licence terms, which are strong, or a separate licence that enables more normal commercial use & distribution is available from Braid. Contact us for more details mailto:[email protected].