Hi Folks,
Been delving into the world of GPT and I was really hoping that you guys could point me in the right direction.
I work (in Scotland) with a LOT of disabled, carers and unemployed who are currently subject to a cost of living crisis and repeated errors by the primary benefits agency (the department of work and pensions). Put simply, if people get to me, every other mechanism has failed them.
Suffice to say my workload is overloaded.
I’m looking to try and setup some form of GPT setup to aid me in the preliminaries.
I have a huge (and I mean huge) document library I rely on every day and what I would like to have is fairly simple.
I’d like to be able to access a webpage over my LAN on my ubuntu server. Type in a list of preliminary issues and create a case file for someone. Dump all of their correspondence into a network ingress folder and have the server convert that to text. Then use an offline version (by which I mean not connected to API with openAI) of GPT to read and understand that correspondence, build a timeline of communications, then go through my document library to match the issues in that correspondence to the current regulations for benefits.
In addition I would like it to continually update the document library from specific websites.
The intention here is to have GPT build a case file / document with a full summary of the issues for the individual, timeline, important information and pointers to exactly what provisions I should be concentrating on.
Im not looking for it to give me a perfect case file, just to do a preliminary read so when I then get the file I have a decent framework to build on, and to speed up my dealing with each complaint.
This would apply to both the DWP, other government organisations and large companies screwing some of the most vulnerable.
I lack the necessary knowledge of GPT to do this (I only have basic knowledge) and I am hoping that the community will assist me in building and deploying this out of the goodness of their hearts. It’s a very worthwhile project and I would be very greatful for any assistance people could render.
Martin