Show HN: My first full stack project (plotnotes.ai)
Just finished my second year at college studying Computer Science, and wanted to try and gain experience as a full stack dev this summer, so I made a website that allows users to create stories using openai's API, using embeddings for custom terms users can define, and create short stories and chapters alike. I'm mainly just looking for some feedback on how I can improve it or make it more presentable, so any help/feedback is welcome and appreciated!
28 comments
[ 3.4 ms ] story [ 75.7 ms ] threadThat being said, I caved and set up an account with a temp email and the password 'password' (might wanna check on those password security rules) just to play with it so I could give some constructive feedback. Unfortunately, generating a story isn't working (the spinner is spinning infinitely) and I see a few errors and warnings in the Developer Console. I'm not sure what the Short Stories page is supposed to show, either, but it also appears to be malfunctioning.
I don't mean to eviscerate your project - it is always exciting and scary to share something with the world. This needs a little more work before it's ready for show-and-tell, though, IMO. I'm not even sure what I would use this for. Best of luck.
Like jamisonbryant said, people want some assurance about what they’re getting in exchange for their personal info and a direct line to their email.
Took a quick peek at your repo’s README, under features you have “Next.js”, cutting your teeth on some marketing is worth it! Think about what makes you want to use a tool, It’s probably not the framework it was built with.
The bent for this space tends to be towards start ups, so expect some advice towards how to market/best feature something you made, vs what would look best on a resume.
It should be possible to show the editor without the login-wall, but with the text input being disabled and be replaced by a pre-defined input like "Write me a story about bunny rabbits" (I don't know what your app is supposed to do so this is my understanding after reading the scant descriptions), and tell the user to hit "Generate story!" - but when they click it, the mock-up won't be talking to ChatGPT, instead it will just get a response (or one of several bunny rabbit stories) you've stored as text.
That way you don't need a login-wall, and no one will abuse the project and eat up your ChatGPT tokens (is that the reason for needing a login?).
Visually it looks pretty good. However like what jamisonbryant said, people on this website aren't fans of when they have to sign in to test something. I made a test account to try it out, and unfortunately submitting my prompt resulted in a Error status 504 on the request with message
"An error occurred with your deployment
FUNCTION_INVOCATION_TIMEOUT"
This is probably because you are waiting for the external openAI API's response before responding to the user's request. You might want to return a response immediately and asynchronously process the openAI part of things and notify the user when its ready somehow.
For UI, I might suggest something more “inlined” for the prompt screen. You might find some inspiration from outline writing apps like OmniOutliner [0] or emacs’ org-mode. Those will let you move around heading and groups of rows easily. If you could expand a row into multiple heading rows generated by AI, I could see that being really “in flow” for a writer!
You could imagine someone writing a couple details, hitting some key combo that expands it into a few headings, and then they can move around to the heading they want to expand on and start writing in it. They can recursively expand sentences/rows when ever they need to flesh something out.
It’s hard to get keycombos and the feel right, but I think it would be worth it.
[0] https://www.omnigroup.com/omnioutliner/