Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

flowery.app

Post reply on HN
Print view View original

21–26 of 26 posts

Re: Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

#21

Post by the_florist »

BoggleFiend wrote: post #6

Congrats on adding a project supposedly using an LLM to you resume, but it doesn't seem to do anything useful. "fox" and "beta" both led to a small shake, which as a user I would think indicates they are not words in the English language. "Flowery" returned a definition of: "full of, re­sem­bling, or smelling of flow­ers—e.g., a flow­ery meadow, flow­ery wall­pa­per" This app appears to be a limited but somewhat arts…

The dictionary preview is limited to the nth letter of the alphabet, where n is the day of the month, as hinted by the search box’s placeholder.

Try looking up any word that starts with S, assuming it’s June 19th in your time zone.

Re: Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

#22

Post by infecto »

the_florist wrote: post #19
infecto wrote: post #16

Earlier quoted context omitted.

Absolutely no idea how to use the app either. It wanted me to type an s word? So I typed store and hit enter, it gave me the definition of store. Very very confusing workflow. Also absolutely clueless how this relates to a LLM.

You look up a word in the dictionary and tap the heart next to its definition (as the tooltip suggests) to save it. The Hearts page ( https://flowery.app/hearts ) is your personal vocabulary list, and the Cards page ( https://flowery.app/cards ) generates cloze deletion flashcards for your hearts, which is where the LLM steps in.

Makes sense but I think the main issue with the workflow is that the user is having to type words out? As a vocab builder it would be nice to do more suggestive words or something that does not force me to type in words.

Re: Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

#23

Post by the_florist »

infecto wrote: post #22
the_florist wrote: post #19

Earlier quoted context omitted.

You look up a word in the dictionary and tap the heart next to its definition (as the tooltip suggests) to save it. The Hearts page ( https://flowery.app/hearts ) is your personal vocabulary list, and the Cards page ( https://flowery.app/cards ) generates cloze deletion flashcards for your hearts, which is where the LLM steps in.

Makes sense but I think the main issue with the workflow is that the user is having to type words out? As a vocab builder it would be nice to do more suggestive words or something that does not force me to type in words.

You can quickly populate the search box by tapping any word (or highlighting on desktop) in the hybrid dictionary/thesaurus.

As I mentioned in another comment though, the ideal workflow would be an integrated ebook reader that lets you tap words without context switching.

Re: Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

#24

Post by watchblob »

the_florist wrote: post #17
watchblob wrote: post #9

Overall, I like the idea a lot: it's Anki specifically for building your vocabulary. That said, I struggled a lot with the UX. I think you need to make it more obvious that this is a preview and today you can only use words starting with S. Changing the initial text in the search to 'Try searching for a word beginning with S...' would go a long way. I'd also explain the product to people a bit more when on the landin…

Thanks for the constructive criticism! There’s definitely room for improvement. The design was intended to be minimalistic, but perhaps overly so. The preview could indeed be clarified: The letter corresponds to the day of the month, so the A headwords are unlocked on the 1st. Today’s the 19th, hence S. The cards in the login/signup sheet are notifications, but I agree that the (non-dismissible) signup notification s…

Nice! Good luck anyway and congrats getting this out of the door. I know how difficult it can be to get to this stage. With some UX refinement then I think you'll have a nice proposition. As you said, pricing can evolve.

Re: Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

#25

Post by watchblob »

the_florist wrote: post #18
watchblob wrote: post #10

Earlier quoted context omitted.

I've just tried it on mobile. What's the reason for building a custom keyboard?

I decided to roll my own virtual keyboard after a protracted battle with inconsistencies and quirks across mobile browsers. The glaring difference is that the native keyboard slides the viewport upward on iOS (pushing the fixed header off screen), whereas it shrinks it on Android (causing relayout that repositions any UI element fixed to the bottom). Also, notice how the search box is anchored to (and animates with)…

Personally, I think this should have been a nice to have (if at all). There are other areas of the product which could have been refined which would have had a bigger impact (intro and sign up journey).

Re: Show HN: Flowery – Vocabulary builder powered by LLM and spaced repetition

#26

Post by closetkantian »

the_florist wrote: post #20
closetkantian wrote: post #13

I like the typeface and concept but I found the site somewhat confusing to use. Also, I feel like it should recommend some difficult vocab words for me to check out, based on say a quiz to get a baseline sense of my current vocabulary.

The timeless elegance of Caslon. Features to discover vocabulary are on the drawing board. In the short term, a “word-a-day” feature will recommend five words per week, but the time-tested method for a rich vocabulary is reading, so the long-term ambition is a built-in ebook reader, where saving words would be frictionless and context-aware. In the meantime, you can browse the index, which is biased toward uncommon w…

Okay, thanks a bunch. I work a lot with students studying for the SAT and GRE, and something like this could be very useful for them.
Post reply on HN