ajigarjian
ajigarjian on Hacker News
Agreed, at this moment it’s an anecdotal labeling. Honestly if anyone can point us towards some other great tooling in this space, we’d love to compare and contrast more deeply. Right now we’re benchmarking against data.gov
We only want people to pay for Crystal if they feel they get outsized value out of it; we’d never ask you to pay just to make a few searches.
That being said, each query makes a few LLM calls to make sure we’re returning the most relevant datasets, which costs money. So at some point we would expect power users to contribute, which seems reasonable.
Ultimately our goal is to democratize access to this data. So the cost would never be more than a nominal amount
It’s U.S. data as per the title - but our goal is to integrate datasets from every country to make it comprehensive, no matter what you’re looking for
Hi, yeah we have our mission on the About page. Appreciate you posting the additional context
Posted by
ajigarjian on Sunday April 13, 2025 @02:23PM
from the show-hn dept.
https://askcrystal.info/
Hi everyone! We're relaunching Crystal, which lets you search and analyze 300k+ government datasets using plain English. For example, prompting "Air quality since 2020 in NYC" will find the most relevant datasets for you.We find it's way better than any search tool out there today, like data.gov. We're hoping anyone who uses public data as a resource, like researchers, consultants, journalists,...
Posted by
ajigarjian on Tuesday April 01, 2025 @09:50AM
from the show-hn dept.
https://www.askcrystal.info
Hi everyone! Crystal lets you search and analyze 300k+ government datasets using plain English. For example, prompting "Air quality since 2020 in NYC" will find the most relevant datasets and analyze them, providing a short report with key statistics extracted from the data based on your question.We felt like AI is great at web search today, but not so much for actually combing through...
I also trimmed down the graph rendering so it doesn't take as much time (now that the play pop up isn't in the way)
Thank you for the feedback - you playing and taking the time to write all of this is super helpful and I really appreciate it.
The option to move quickly between puzzles totally makes sense - I added in buttons below the "Graphs title" to do so. Does it work well from your PoV?
I also made it so that the "play" button only pops up on your first graph of each day, afterwards it will not impede you as you binge :)
I fixed the August 16th labeling (regex strikes again) too, nice find.
The date thing is almost certainly due to varying timezones of players - they're all around the world at this point. I'll have to think more deeply about how to fix that, but it is 100% going to happen.
Thanks again for all the feedback and for playing!
Hm, what pops up when you click play (or anywhere outside of the pop up)?
I've been staring at it too long and can't tell if it looks good or bad
Tried my hand at an update on the calendar. Thoughts? The colors were proving tricky
I added in an option via a settings button (up top) to toggle grid lines...hoping that is a good start!
Appreciate the kind words!
Thanks for the feedback, I'll look into fixing this. Are you referencing a graph that had a 10 month period as the x axis?
Still looking for that Goldilocks zone ha
This is a great idea and in the backend it tracks how many tries it took to win, so the data is there. Also, NYT crosswords does this similarly where you can see on the calendar how far into each puzzle you got visually.
Let me take a stab at this...I'll let you know when it's been updated.
Appreciate the feedback, really good insight!
Thanks for enjoying, James!
Posted by
ajigarjian on Sunday September 01, 2024 @10:55AM
from the show-hn dept.
https://www.graphs.world
Hi everyone! Graphs is the intersection of data visualization and the daily web puzzle genre (Wordle, NYT Mini Crossword, Connections, etc.).Every day, you are given a nameless graph and 5 options to select the right dataset it is representing. Upon winning, you can access the underlying dataset and other info + maintain a streak and stats. It's super simple - check out today's chart and you...
Posted by
ajigarjian on Friday July 26, 2024 @11:11PM
from the show-hn dept.
https://www.graphs.world
Hi everyone. I'm making Graphs as a love letter to the crossword, Wordle, Connections, and other "educational" daily games. I started building it a week or two ago and have some more ideas on the roadmap, but I'd love to share it today to see what people think.Every day, you are given a nameless graph and 3 options to select the right data set it is representing. It's super simple - check out...