Custom GPT Setup using OpenAI
Budget: $250 – $750 AUD
# Overview:
I am looking for a freelancer who can help me with a custom GPT setup using OpenAI. I have specific tasks in mind that require basic customization, such as fine-tuning existing models.
# Objective:
I'm working on creating lore (context, nuance and information on locations, characters, items, relationships, etc) for a high fantasy game I'm creating and I'm looking to have two custom GPT's setup to help ideate and track lore.
The two GPT's should be distinct from one another, the first ("Archive GPT") will help me ideate and "save" lore, the second ("Historian GPT") will help me to query and fact check saved lore.
The goal is for the lore that is saved to persist across sessions, and be saved into a database.
---
"Archive GPT", and should act as the creator or archiver, in which I want it work how chat GPT normally does and prompt it, have it give me back responses that help me ideate and create lore. However it's responses are "saved" unless I tell it to "save" them.
Once I confirm I like a bit of lore I want to be able to respond to the Archive GPT confirming that I'd like whatever I give it to be "saved".
The second custom GPT, "Historian GPT", will respond to queries with definitive answers. Historian GPT should not make up information it does not know. It can draw conclusions and likeness between things it knows (and be clear that it has done so), but it should not invent or create new lore.
Ideally Historian GPT will be largely using the "saved" info that is fed into "Archive GPT" as its underlying source of truth.
The hope is I can work with "Archive GPT" to form up lore and "save" as necessary, and then query "Historian GPT" on things that have been saved to date. The data saved by Archive GPT should persist across sessions.
# Ideal Tech & Stack
I'm no expert on AI, hence the job posting, open to recommendations here, however, to share some preferences:
- If there is a need for any form of API, ideally it would leverage JavaScript/Node/Typescript
- I'd love the underlying data to be saved into a PostgreSQL database (assuming we want pgvector here as well)
- Ideally we would be able to host most of the setup via Supabase (https://supabase.com/). This includes any API's spun up as Supabase edge functions (Deno), and using the underlying Supabase backed PG DB.
- Custom GPT will be created using Open AI's Custom GPT Builder (where necessary)
If you have expertise in this area and can deliver quality results please reach out.
I am looking for a freelancer who can help me with a custom GPT setup using OpenAI. I have specific tasks in mind that require basic customization, such as fine-tuning existing models.
# Objective:
I'm working on creating lore (context, nuance and information on locations, characters, items, relationships, etc) for a high fantasy game I'm creating and I'm looking to have two custom GPT's setup to help ideate and track lore.
The two GPT's should be distinct from one another, the first ("Archive GPT") will help me ideate and "save" lore, the second ("Historian GPT") will help me to query and fact check saved lore.
The goal is for the lore that is saved to persist across sessions, and be saved into a database.
---
"Archive GPT", and should act as the creator or archiver, in which I want it work how chat GPT normally does and prompt it, have it give me back responses that help me ideate and create lore. However it's responses are "saved" unless I tell it to "save" them.
Once I confirm I like a bit of lore I want to be able to respond to the Archive GPT confirming that I'd like whatever I give it to be "saved".
The second custom GPT, "Historian GPT", will respond to queries with definitive answers. Historian GPT should not make up information it does not know. It can draw conclusions and likeness between things it knows (and be clear that it has done so), but it should not invent or create new lore.
Ideally Historian GPT will be largely using the "saved" info that is fed into "Archive GPT" as its underlying source of truth.
The hope is I can work with "Archive GPT" to form up lore and "save" as necessary, and then query "Historian GPT" on things that have been saved to date. The data saved by Archive GPT should persist across sessions.
# Ideal Tech & Stack
I'm no expert on AI, hence the job posting, open to recommendations here, however, to share some preferences:
- If there is a need for any form of API, ideally it would leverage JavaScript/Node/Typescript
- I'd love the underlying data to be saved into a PostgreSQL database (assuming we want pgvector here as well)
- Ideally we would be able to host most of the setup via Supabase (https://supabase.com/). This includes any API's spun up as Supabase edge functions (Deno), and using the underlying Supabase backed PG DB.
- Custom GPT will be created using Open AI's Custom GPT Builder (where necessary)
If you have expertise in this area and can deliver quality results please reach out.