It tracks Reddit and LinkedIn hourly, and has an AI filtering feature where you can describe in natural language what type of posts you want or don't want, and it will filter out the rest.
E.g. "Only show me posts where someone is asking for recommendations of a social listening tool" - this is actually the filter I used to find this post
Pretty much all of my users for ForumScout have came from using my own product to find users with buyer intent.
I also track mentions of competitors and have it show me posts where people are complaining about them, so I can offer my tool as a better alternative.
Every day, about 10,000 people announce that they're open to work online.
I built a saas that listens for these 'Open to work' posts on LinkedIn, and matches the poster's profile (work history, qualifications, skills, location, etc.) against job descriptions, in real time.
So it's like: "here's someone who exactly matches your job requirements, and they've JUST announced that they want a job".
It's quite simple and hands off because you literally just need to copy and paste in your job descriptions and wait for matches to come in.
There's a 30-day free trial with no card required if anyone would be happy to test it out and let me know what you think.
It's been in development for about 5-6 months, so it's very new and I'm really looking for people to test it out and give me feedback :)
I personally don't, but I run a tool called ForumScout and you could literally just spin up a listener for whatever news-related keyword you want to track and it will add all mentions to your feed.
And you can add an 'AI filter' where you can explain to the AI what posts you want/don't want and it will check every single post and make sure it matches your criteria before adding it.
So for example if you want to track mentions of "Trump", your AI filter could say "Only show me posts discussing his economic policies. Avoid any mention of his family or personal life", and that's what you'd get.
I guess it would be useful if you want some very specific stuff, but plain old news aggregators work just fine for me :)
X just updated their terms of service as of yesterday (effective on the 15th of November).
Obviously, no one actually has time to read it, so I made a color-coded summary of it that tells you exactly what you’re agreeing to in simple terms without the legalese.
Scalability: This is more of a cost/resource issue than a technical one. Right now, ForumScout checks every hour for new posts. This isn't ideal, but it allows me to offer the product for much cheaper than if we were to check, say, every 5-10 minutes. Scalability is much more manageable with an hourly check, but the goal eventually is to get as close to real time as possible. I plan on offering a higher pricing tier for real-time checks to account for the extra resources.
Filtering: By default, the searches are performed via exact match. From testing, there are no false positives; there's some code that double-checks that each post contains an exact match of the keyword(s). Users can manually turn off exact match if they want broader searches. Sentiment searching is a VERY good idea. I will add it to my list!
Integrations: This is a great point, and something I had not yet considered. Right now, the feed can be exported via CSV, but direct integrations would be much better for UX.
Technical: Honestly, it wasn't too difficult - every hour, the automation will make a call to all the different APIs and search using the parameters provided by the user at setup. Then we format and combine all the results, remove duplicates, perform some more filtering (depending on what settings the user has chosen), and add the results to the user's feed. If emails are turned on, we send an email with the new results. As mentioned, making this real-time will be more difficult, but it would really just mean doing this every few minutes, requiring more resources. I did build a custom scraping function for Reddit comments, because the native Reddit API doesn't allow direct searching of comments like you can on their website/app, but everything else is third-party.
Hey HN, this is something I've been working on over the last 4/5 months.
I got the idea when watching a Black Mirror episode called Joan is Awful.
This episode is about a woman who signs up for a streaming platform called Streamberry, accepts the terms of use, and goes about her life.
When she gets home from work, she notices a new show on Streamberry called 'Joan is Awful'.
Long story short, Streamberry's terms of use included a clause permitting them to digitally clone the user, and use their likeness in any way they want (in this case negatively, ruining Joan's reputation).
I remember finishing the episode with a huuuge drive to start the project because it really clicked how big the issue is - so big, that black mirror made an episode about it.
It effects everyone, everywhere, and we're all helplessly aware of it.
DocDecoder is pretty simple, in that it uses GPT-4 to summarize a provided web page, and then colour codes each term for ease of understanding.
Despite the simplicity, I think it does a really good job of informing you about what you're actually agreeing to when using the services you use.
It tracks Reddit and LinkedIn hourly, and has an AI filtering feature where you can describe in natural language what type of posts you want or don't want, and it will filter out the rest.
E.g. "Only show me posts where someone is asking for recommendations of a social listening tool" - this is actually the filter I used to find this post
Pretty much all of my users for ForumScout have came from using my own product to find users with buyer intent.
I also track mentions of competitors and have it show me posts where people are complaining about them, so I can offer my tool as a better alternative.