garussell / aiartistassist Goto Github PK

This project uses OpenAI to furnish some ideas and an image for getting started on user's next creative project. For example, if you want to learn to paint this app will give you ideas and resources on what to do next as well as provide an image to associate with that particular idea.

Home Page: https://www.loom.com/share/2a21a2152b024118bf7d3df30188410d

Ruby 71.84% JavaScript 1.98% CSS 2.67% HTML 23.51%

monolyth motivational-messages openai-api

aiartistassist's Introduction

Artist Assist

UI Video Walkthrough

Overview

The Artist Prompts API is a Rails application that leverages OpenAI's GPT-3.5 Turbo model to generate prompts and suggestions for artists. It allows artists to receive personalized recommendations for their goals, action steps, and resources based on their input.

Features

Generate Artist Prompts: Utilize the OpenAI GPT-3.5 Turbo model to generate prompts for artists based on their specified goals.
Save Prompts to Database: Save the generated prompts, along with associated information, to the database for future reference.
Handle Errors Gracefully: Implement error handling to gracefully manage scenarios such as artist not found, invalid parameters, and other potential issues.

Getting Started

Prerequisites

Ruby 3.2.2
Rails 7.0.8
OpenAI API Key: Obtain an API key from OpenAI to enable communication with the GPT-3.5 Turbo model, or GPT-4 model.

Installation

Clone the repository:

[email protected]:garussell/artist_assist.git

Install dependencies:

bundle install

Setup database:

rails db:create rails db:migrate

Configure OpenAi API Key:

$ EDITOR="code --wait" rails credentials:edit

open_ai: api_key: your_openai_api_key

Save by closing the credentials.yml file

Usage

Start the Rails server:

rails server

Make API requests to /api/v1/:id/post_idea

Artist ID is part of the URL
The prompt is passed into the body of the request (see endpoints)

Endpoints

POST /api/v1/post_idea

Generate a prompt from OpenAI based on the artist's specified goals:

Request Body:

  {
    "goals": "I want to learn to play the drums"
  }

Response

Status Code: 200

Body:

  {
    "data": {
        "id": null,
        "type": "response",
        "attributes": {
            "id": null,
            "action_steps": "Stuf about drumming:,
            "resources": "Resources with links",
            "goals": "I want to learn to play the drums",
            "image_url": "https://oaidalleapiprodscus.blob.core.windows.net/private/org-DUWm0ko8w9epxLgtftuHr2ZP/user-C3kDGpQ319mqmHmO4pjG7bKq/img-7aiQAzBdz1lPJJN2adBKBMSg.png?st=2023-11-20T19%3A35%3A08Z&se=2023-11-20T21%3A35%3A08Z&sp=r&sv=2021-08-06&sr=b&rscd=inline&rsct=image/png&skoid=6aaadede-4fb3-4698-a8f6-684d7786b067&sktid=a48cca56-e6da-484e-a814-9c849652bcb3&skt=2023-11-20T20%3A28%3A47Z&ske=2023-11-21T20%3A28%3A47Z&sks=b&skv=2021-08-06&sig=Qk6szB/AlQ7eUIZnuO4Fdbh0YVl5%2BFluLWuUGJ87Crs%3D"
        }
    }
}

Status Code: 404 :not_found

If the artist does not exist:

  {
    "error": "Artist not found"
  }

Status Code: 422 :unprocessable_enity

If "goals" were left blank

  {
    "error": "You must provide your creative goals."
  }

Future Iterations / In progress

CRUD functionality for Artist
Authentication for Artist
User Interface

Contributing

If you'd like to contribute to the project, please follow these steps:

Fork the repository.
Create a new branch: git checkout -b feature/your-feature
Makes changes and commit: git commit -m "Add your feature"
Push to the branch: git push origin feature/your-feature
Open a pull request

Created by:

Allen Russell
- GitHub: @garussell
- LinkedIn: @garyallenrusselljr

aiartistassist's People

Contributors

Stargazers

Watchers

aiartistassist's Issues

ActionCable-WebSocket-OpenAi-ChatBot-Service

This is going to happen!

Build the back end first, then separate issues for front-end integration.

(this is also where it starts to make sense to separate the back-end with AI services and deploy them separately (on the cloud), as well as utilize Docker / then build the front-ent using a framework)

AI Prompt Config

The AI prompt has some edge case issues. There is a strange default photo generation which makes me treat it like an error message. It would make more sense to parse out a few different prompts for the user and pass those in as params to limit the "open ended" responses and AI confusion.

"Style" alone might not be the best input prompt for the photo. The goal is to align them with the user's "goals" and "style" together.

Add Artist Files to Artist's show page

there should be a form where artist can add their "goals" and this will hit OpenAI
once the response is saved as an 'artist_file', it will be displayed on the Artist's show page and they will have CRUD

Fix dark light

dark/light is not functioning correctly

redis in production env config

Once deployed, add redis server as an "add-on" and set the server url and secret key in credentials

Refactor: Artisan

'Artist' seems to be specific to 'painting'.

I want to change some names in this project.

Artist table to 'Artisan'

revisit column names
add a reference to the table for 'notes' - has_many

ArtistFiles to ArtisanFiles (or Artison#{...})
Create a new table (see issue for 'notes' table)

Save and fetch new Photo

It seems the photo url provided by OpenAI is temporary, so this is a good place to implement photo storage on the cloud. I also want to add the option to retrieve a new photo.

AWS
ActiveStorage

Parse "action_steps" response

Separate by numbers similar to "resources"

Allow 'style' to be an attribute passed through the arist_files#Update

I want to pass 'style' in as a param with "goals" so the user has more control over the prompt. Style and goals can be left open to the user's interpretation but consider adding helpful direction (possibly chatbot-driven - ActionCable) !!!

Implement caching

Utilize Redis/Sidekiq to cache resources

scalability

-Google Cloud for file storage
-containerization / horizontal scaling using Docker
-CI/CD pipelines implementation
-optimize API call to OpenAI
-deployment to heroku
-load testing with Locust

Formatting Ideas

Artist file can be like a card deck that you can "slide" through
Look at bootstrap options
Make sure class and id tags are clearly named and setup stylesheet

(for this project is probably better to stick with vanilla JS - a refactor for later could be to expose the backend API with GraphQL and use React to manage the front end)

notes table

consider adding a table for notes that can be associated with the artist (artist has many notes) and provide a template (form) for the artist to add notes / journal entries to their profile.

apply consistent styling

Apply styling to navbar and consider footer, and page divider(s)

test-coverage

address coverage issue that happened after adding JS event listener re: password fields.

edit-page-styling

Match to other pages

Feature: options for AI prompt

select different skill level

option to edit any response bullet point
option to 'refresh' same prompt (caching)
option to refresh the photo

** the artist's "style" field might benefit for some edge case issues. If user puts in something random the response will be a bit wonky vs. if they put in "Jazz" and their goal was to "improve jazz drumming" - because of the consistency, the responses are much high quality

Complete crud functionality for artist_files
Artist files include the API call to OpenAI

page refresh after creating new artist file

home page

Right now the root path is a partial that renders login or logout depending on session. That is fine, but it could be enhanced with a few ideas:

Incorporate login to a chatbot (when implemented)
Artwork of photo with a logo and/or description
footer tabs with contact info, links, readme, etc