I have been working on a project since the beginning of this year. The working name is 'Virtual Helper' and it is an idea I had come up with when working for a company asking to build an internal system for them. As many may or may not know, this comes with a HUGE amount of questions and challenges. At this point in time, ChatGPT was really just starting to take off as something people used and many open source LLMs were misusing licenses so those were never options. Also around this time, Microsoft was working on this type of platform on Azure and Google Cloud was not far behind. However, those options required, and still require, for you to have your company information "shared" so their models could "learn" more. Not a very secure solution in my opinion and Samsung saw this first hand when new product data was leaked on ChatGPT by programmers.

I attempted to explain all of this to the CEO of the company I was working for (glazed eyes almost immediately) and explained it would be a big project that he ultimately did not want to fund. However, the idea of building a "business ready" prompt engine stuck with me and I started building like any idea-fueled person would.

It is now May 2nd of 2024 and I am still developing this project. However, just within a few months' time, the models being used like Dolly v2 are way better! The accessibility on cloud servers has changed as well. The downfall to all the change is that my project also has to adapt to change when prompt engines are really just in their infancy at this current stage.

I have a "launch date" for October of 2024. If you are reading this solution before then and want to get involved, let me know and would be glad to have you on the team. If you are reading this after October 2024, go check it out on virtualhelper.ai. Reach out to me directly and I will get you a demo to it.

Comments