- Many friends complain about the overwhelming amount of content online, especially tech news.
- A lot of content takes quite a bit of time to read.
- Finding the information you want often requires browsing different websites and spending a lot of time, especially when it’s content that can’t be quickly accessed through a search.
Our solution involves using AI—not only to aggregate information (which many RSS readers can do)—but to address the issue of content overload after aggregation, where many articles go unread. So, we use AI to filter the information based on natural language processing, pulling a small portion of data that meets specific criteria from many sources. We then create content digests, allowing users to read in less time, with the option to click through to the original source if they want to see the full text.
I know I must have missed something, as the HN community didn’t give much feedback on this solution. Is there a problem with the solution itself, or should I focus more on improving the way I write posts or design the landing page? I’m feeling quite lost right now.
The problem is a fascinating one, and one that I (like most) have struggled with for many years. That said, I won't use an external filtering service (AI-based or not) because then all I'm doing is allowing someone else to shape my informational world.
Instead, what I've done is to be more aggressive and conscious about doing that filtering myself. I severely limit the amount of attention that I pay to things that I can't affect and that I have little inherent interest in. I read my news through RSS, using an RSS aggregator that allows me to choose sources and set filters that determine what specific items what do or do not make it into my aggregated news feed. I've become extremely pleased with using Kagi for searches, in large part because I can tell Kagi what sites I want to hear less from and what sites I want to hear more from, and in part because I can actually find what I'm looking for with it.
That sort of thing. In short, to deal with information overload it's necessary to reduce the amount of information you take in, and the only person I trust to decide what is important for me to see is myself.
Feedback:
1- Show HN posts can be hit or miss traffic-wise, based on the time of day, if everyone was busy, etc. So, don't take small response to mean much. (Some great projects that have gone on to be strong got less initial Show HN feedback that yours.)
2- Personally, I think your landing page could benefit from a video near the top that shows how you use it. The high level concept is pretty clear, so I won't restate what you are doing here. But more visual communication (either detailed series of screenshots, or, more likely, a video) showing "how do I set this up and use the output of the product" might go a long way to letting people know what it is you have in mind to share with them. This would strengthen the "show" part of the Show HN, IMHO.
3- It's not really clear there's any proof that it works, or that the AI doesn't just do a poor job of summarizing/aggregating/etc. (which people are polarized about--some find it amazing, some find it quite poor). Does this tool achieve the stated goal? Even if we're not sure, how could the we find out for themselves? This part seems unclear, reading the page, before trying to Sign Up. So much so, that it seems kind of like a landing page to gauge interest before anything was built. Is that the case? It's honestly hard to tell. Again, to address this, show an end to end example as a 2 to 10 minute video.
4- If the first Show HN was a trial balloon, next time maybe write a Blog Post detailing more about your product, and submit that as an Article. Then keep doing that whenever there is major News to share. .02. Eventually one of your submissions might pop off.
Absorbing higher quality, curated sources will be better for anyone than suckling off the fire hose and wondering why you feel like you've had too much. Take the breathing scratching exercise break and get back to the value add work.
All the classical solutions apply (improve your reading level, take more selective interests and know what to do when enough is enough.)
LLM is a toy. You should use it to spot check your development, your "down stream" usage sounds like you're on to something, are you? I still like to run things by my "go to" from time to time, however I cannot say it has been more than novel.
And I think this crowd sees these AI business models as a few workflow scripts and a well crafted prompt. Few may want to play with someone else's (marketed) attempt unless it really shines value.