https://github.com/maggieappleton/invenntion
https://github.com/maggieappleton/invenntion
Last synced: over 1 year ago
JSON representation
- Host: GitHub
- URL: https://github.com/maggieappleton/invenntion
- Owner: MaggieAppleton
- Created: 2023-01-01T12:17:39.000Z (over 3 years ago)
- Default Branch: master
- Last Pushed: 2023-01-09T07:58:08.000Z (over 3 years ago)
- Last Synced: 2025-02-28T10:53:23.857Z (over 1 year ago)
- Language: TypeScript
- Size: 3.01 MB
- Stars: 17
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# How to Run
This project uses `yarn`, and runs on node 18 (check .nvmrc for the exact version). Run it locally with `yarn dev`.
# Design Brief
## Concept
A venn diagram that combines two concepts or ideas for you. Using a generative language model, we give the user three++ possible ways to combine their initial concepts.
Uses OpenAI's GPT-3 to generate ideas.
Users can save generated results as text or PNGs.
### Initial concept sketch
### [Wireframes](/design-process/wireframes.md)
## [Visual Design Mockups](/design-process/visual-designs.md)
## WIP Proof of Concept
Basic functionality working on the prototype
## Requirements
### Essential
Users should be able to:
- enter their OpenAI API key to make requests
- make sure that API key isn't stored or exposed to 3rd parties
- enter two words or sentences. Max length 100 characters.
- click a button/area that triggers the combination action: sends a request to OpenAI
- use text-davinici-003
- return three options, each less than 180 characters
- see a loading animation while the request is resolving
- see the returned results
- be able to swipe/scroll/click through the three returned results
- be able to reset the whole venn to be blank
- be able to edit the existing inputs and re-run the generation
### Nice to have
- be able to generate more than three results
- this could be re-running the whole generation, or clicking somewhere to generate a fourth or fifth result in addition to the first three
- copy generated results to the clipboard
- save a PNG of the final venn digram with inputs and generated results
- shared a generated PNG and link to Twitter
- animated rainbow gradient on venn circles – maybe a requirement 🌈
## Todos
- [x] Wireframe main user flow
- [x] Wireframe onboarding sequence
- [x] Final designs that meet all the requirements
- [ ] Write an example dataset of prompts and good answers to fine-tune on
- [x] Good enough prompt engineering
- [ ] Research how we can accept OpenAI keys and use them for API requests without compromising user security