Tosight.xyz
So, I’ve been messing around with AI and built this tool, ToSight, that tries to make sense of those super long Terms of Service and Privacy Policy things. You know, the stuff nobody really reads? I’m hoping to make it easier to figure out what’s actually in them.
Here’s basically how it works right now:
What it does:
Easy URL thingy: You just paste in a website URL, super simple.
- Grabs the ToS & Privacy Policy: It tries to find those pages and pulls out the text.
- I’m using this thing called Mozilla Readability to do the text grabbing.
Saves it all in Supabase: It dumps the text and the URL into Supabase to keep it stored.
- AI Magic with Deepseek: Then the fun part - it sends the text over to Deepseek to try and figure stuff out. Specifically, it looks for things like:
- What users can do on the site
- What data they grab from you
- Who they share your data with
- How they use your data
- How long they keep your data
- How secure your data is supposed to be
- Your rights as a user (data stuff rights)
It tries to do this for both people who are logged in and those who aren’t, if the policy talks about both.
Puts it all in a database: It takes what Deepseek spits out and saves it all structured in the database.
- Shows it on the website: Then on the website, you can pick a site and see the insights all laid out.
Stuff I’m still working on / Problems I’ve had:
Scraping is kinda janky sometimes: Getting the text from websites isn’t always perfect. Sometimes Mozilla Readability just doesn’t work, which is annoying.
ToS/Privacy Policies are sometimes just…bad: Seriously, some of them are so unclear or boring, it’s hard for anyone (even AI!) to get much out of them.
LLMs mess up sometimes: Deepseek is cool, but it’s not perfect. It can make mistakes, so I know I need to add something to double-check the results.
Wanna try it?
You can check it out here: https://tosight.xyz
Let me know what you think!
Seriously, any feedback or ideas are welcome. Still trying to figure out the best way to do all this. 😊