HN Explorer

Show HN: A search engine for your personal network of high-quality websites (usegrasp.com)

266 points by vignesh_warar ↗ HN
Hey all,

Last time when we were on HackerNews [1], we received a lot of feedback, and we incorporated most of it.

- We have changed our name from grep.help to usegrasp.com

- A privacy policy page

- Bulk import

- Pricing page

We are happy to introduce a new feature: a personalized answer search engine that provides direct citations to the content on the page.

Demo: https://usegrasp.com/search?q=is+starship+fully+reusable%3F

1 - https://news.ycombinator.com/item?id=35510949

99 comments

[ 4.9 ms ] story [ 160 ms ] thread
avinassh
hey I just want to let you know that you need to update title and meta properties too. Whenever the link is shared on social media, it still says grep

    <!DOCTYPE html><html><head><meta charSet="utf-8"/><title>Grasp | New kind of search engine</title>
    <meta name="title" content="Grep | New kind of search engine"/><meta name="description" content="Search engine for your personalized network of high-quality websites"/>...
dvh
Can this index entirety of stackoverflow and Wikipedia? How does it keep it in sync with recent updates?
vignesh_warar
Yes, the search results are a combination of our own index and a third-party index, but our future goal is to be fully dependent on our own index.
Brajeshwar
May be it is just me and I'm sure you have put in good effort to justify the pricing. However, that $10+ makes me think, “Do I really need this? I can stay without this and am I missing anything, perhaps from gaining something outsized?”

If this was like $4.99 /mo with an annual of $49.99, I might have just done it; even if I may not use immediately but to support someone starting out.

Or alternatively, a $9.99 /mo ($99.99 annual) would still be something within a budget that I'm not over-thinking.

My thoughts.

vignesh_warar
Agreed. There is room for different plans, but we want some search metric to define the pricing.

On a side note, I truly did not want to put up a paywall. But, it is necessary to support our servers.

jvanderbot
I find it hard to believe that I, a casual, occasional googler, would demand $10/m of server time.
ravenstine
Not everyone's search needs are the same. I gladly pay for Kagi every month, despite how at least 80% of people would never consider paying for whatever The Google already gives them. If you're fine with ads and limited control over your search results, then use The Google.

Also, there's more to providing a service than just server time.

bachmeier
> Also, there's more to providing a service than just server time.

Exactly. So what's the point of limiting the number of searches to 800? The target market is those that rely heavily on search engines, and they're telling them they don't want their business.

xpe
Don’t feel bad for seeking revenue from your customers. This is the correct incentive. Find pricing models that work.
johntash
Another thread mentioned something like a 30day unlimited trial. That could be a good way to gather metrics on how many searches/day to expect from users, and should help with pricing.

Which part of the service actually costs you more money? Is it the cpu/bandwidth of returning search results, or is it the indexing that happens before then, or the storage/memory?

Maybe instead of charging per # of queries, you could charge per number of sites the user follows/searches? You could always leave something in the ToS about abuse to have an option to stop users who are making an excessive number of searches.

bachmeier
The pricing page shows me $15/month for 800 searches. There's no way that pricing is going to work. Even setting aside the issue of getting people to pay that much, who is going to track their number of searches to make sure they don't go over the limit? And nobody that does 10 searches a day - which is about the limit if you don't want to think about this - is going to pay for a search engine.
chankstein38
This. I opened the pricing page and saw the 2 tiers and immediately closed it when $15/mo wouldn't get me unlimited. I'm not tracking search counts, I have no idea how frequently or how much I search, and ultimately I don't need an extra decision of "is this worth $0.018 to search for?" when I want to search for something. The concept sounds interesting but I'm not so confident in it being that much better that I'd accept that cost decision every time I search.

Unlimited would be different. $15/mo to search as much as I want, it's still insanely steep since I can get so much more for so much less on the internet but at least I wouldn't have to accept a per search cost decision.

ysavir
Agreed. Pricing needs improvement. Not just in the bang-for-your-buck sense, but the per-search model. Right now I have the following questions:

1. What counts as a search? If I go to page 2 of search results, does that use up another search?

2. If I have to refine my search to get the results I actually want, can I get a refund for all the searches I made that didn't predict how the query would be interpreted?

3. What happens when I meet the limit? Am I charged per-search? Can I just no longer search?

4. If I meet the limit and then go to DDG, or Google or whatever, are you okay with that? What if the results there are good enough that I start wondering whether or not I want to pay $15/month for a search engine? How much is retaining a paying customer worth to you?

5. If I have to start counting my search numbers, I'm very quickly going to learn to search less. And the better I get at searching less, the less need I'll have for a search engine, let alone a paid service. Are you worried about your pricing model pushing people towards non-search engine solutions of exploring the web and/or finding web pages?

groby_b
As far as I'm concerned: Charge whatever you need to charge to be profitable. This can only work if it's self-sustaining without external influence. A good compromise might be to offer a psychologically attractive price point for less searches, and a slightly higher one for more searches at a round number - 800 is the weirdest number :)
anenefan
Do I have to relax my OS security for it to get data from the local machine or using a new version of linux ... since I got nothing for each query?
vignesh_warar
Could you please share the specific error message you are facing, perhaps from the console?
anenefan
Page loads fine. Hit return, something obvious happens and then it's back to the original state. I've got a lot of nosey web stuff denied though, just days ago nabbed a supposedly AI powered web tracker.

I guess I'll try a live boot as it's working for everyone else.

goy
What do you think about a feature that allows users to follow each other ? The websites followed by one user can be automatically included in the search results of his followers. It will enable people to include in their networks the websites followed, for example, by authorities in their domain, members of their communities, etc
vignesh_warar
Yes, this is going to be our next feature. In fact, this is the exact idea that I prototyped a couple of months ago using Twitter data, but we are completely off from Twitter now [1].

1 - https://twitter.com/Vignesh_warar/status/1573020208289132545

ysavir
That is awesome! I've had the idea for the past few years of making a social search engine, where the only results are pages tagged by people in your "friends" list and several degrees outwards from that.

I never had any interest in actually building the idea, so it's amazing to see something similar hit the market.

groby_b
When you implement that, an important thing to keep in mind is that "people" are usually a mix of interests. (E.g. if you followed me, you'd get a large helping of browser tech, photography, 3D printing, human rights work, and fashion).

It might be a better idea to allow people to curate search lists by interests and share those focused lists.

People are more interesting in the sense of "of your six closest friends, 5 vouched for this site" when a site/search result pops up.

marginalia_nu
Cool. We definitely need fresh takes on search.

I think historically too many have been attempting to copy Google. It was a bad idea when they were great, and it's a worse idea when they're floundering. An imitation very rarely exceeds what it imitates. Is why after untold amounts of Microsoft R&D money Bing is still google-but-kinda-worse.

vignesh_warar
Thanks!

I still remember the day I saw your project on HN [1]. Your work encouraged me to start exploring my search engine ideas. I can't believe you are commenting on my work.

1 - https://news.ycombinator.com/item?id=28550764

marginalia_nu
Great!

Big part of why I've been working so openly on my project is to inspire and to let others see that there's actually still things to be done in this space, and despite what one might assume, impactful things can be accomplished with a relatively modest budget.

stringlytyped
This looks like an amazing tool. I've always thought it would be great if I could curate my own search index. And the way results are summarized with citations is really cool.

However, I am not sure the free plan is generous enough to properly evaluate the search engine and see if I can incorporate it into my workflow. And the pricing feels steep. I would have a look at Kagi's pricing model.

viraptor
> I've always thought it would be great if I could curate my own search index.

You can... kinda. YaCy is pretty much dying as a project and network. But it still works if you want to have your own independent search. It's great for indexing specific endpoints you care about + N degrees of links. (Like your list of RSS, browser history, etc.)

johntash
YaCy always seemed like it'd be much more popular if it had a UI/UX overhaul. Once I figured out how to have it auto-index feeds from certain sites, recent linkding saved links, etc, it became a lot more useful to me. I also have SearXNG include results from yacy automatically so I basically never interact with yacy directly anymore.
passion__desire
Why isn't there a search engine specifically for e-commerce websites?

Generally what I do is I search for X on google. Google throws up some shitty suggestions. I go on to those websites to checkout their products. Facebook comes to know about my intent of buying X since they have their sdk integration with these websites. When I open instagram, facebook starts suggesting X selling websites to me. I check out those and buy X from there.

I have observed this effect 3 times personally.

Infact such a search engine will optimize towards a better buying experience. Reviews, best things to look for when buying X, price sensitivity, whether X can be delivered to your area, etc.

visarga
Because Google didn't have enough time in 25 years to fix that. Or they make more money on ads than straight referrals.
passion__desire
Is there a way to tell facebook, I am looking for X? Because facebook suggestions are much better than plain google.
hattmall
Search it on marketplace?
passion__desire
Marketplace is for singular items, posted by individuals. I am looking for websites or some kind of yahoo directory structure where I can checkout from the available products.
hiisukun
I haven't used it for a while, but there is a website called 'staticice' that appears to still be an e-commerce price/product search engine. I'm not sure how it works, maybe the sites sign up, or provide prices, or staticice scrapes. It's mostly for gadgets and computer parts but has worked well enough a few times.

Similar to pcpartpicker but much less 'utility', just pure search for model w/ price.

I wonder how one would discover websites like this currently though -- it's certainly not in google's interest to make the top result infobox something like "Didn't find what you are looking for? Try searching using this other site: goodcommercesearch.net" !

marginalia_nu
The market for discovery in general is really under served almost across all areas.

Magazines and to some extent brick and mortar stores used to serve the product discovery market but right now it's so hard to find a product that fits your needs.

So much bait-and-switch nonsense and straight up scams.

Doesn't help that many e-commerce sites are such a pain in the ass to use. 45 second page loads with an additional 15 seconds of random layout shifts is comparatively good. Product listings that show like 6 items per screen, in a random order with useless and truncated descriptions and no useful search function.

I don't understand how they're getting any sales at all.

andrepd
There are price comparison websites, high-quality reviews (rtings, notebookcheck, maybe wirecutter), deals sharing websites (pepper et al).
ricardo81
There sort of is, at least in Europe with Google's CSS providers. IIRC there's a few dozen aggregators and aggregators must have at least 50 merchants in order to be a CSS partner. The products are shown as a carousel and can undercut Google by 20% due to tax/competition reasons. I'm not fluent on the underlying reasons- the results are also on Google shopping.
FinnKuhn
There is for the german speaking countries and it is called Geizhals[1]. It allows A LOT of detailed filters for different categories and shows you websites offering that product. Most times when I need something I just input my criteria for the thing I want to buy and look for the best price to performance ratio of the products that fulfil my criteria.

[1] https://geizhals.de/

neuah
This looks really cool! It could be useful to be able to whitelist some sites without the 4 degree connection. For example, if i wanted to include large networks like reddit, github, stack overflow,etc. in my search results, 4 degrees may start to bring in a lot of junk/undesirable stuff into the index. Also love the idea of being able to follow or search within curated lists made by other users.
vignesh_warar
As of right now, the Grasp Network builder won't consider outgoing links from UGC (User-generated content) websites such as Reddit, Twitter, etc. as it would bring in junk, as you mentioned. I am maintaining a full list of UGC sites.
visarga
For me it does the search but hangs forever on the LLM.
vignesh_warar
LLM part is currently under load. I am currently doing a full rewrite for it, which will improve its speed and reliability.
chaxor
Wow, if 4 degrees out doesn't give 70k nodes it's extended to 7 degrees out? I figured almost any website would yield a near complete graph at just ~6 degrees out.

This is a fantastic idea though. I have been looking for something like this for quite some time to just have basically wikipedia, stackexchange, email, and gitlab available for 'work-mode'. My solution was to make my own search engine with various tools, but this may be easier.

Anyone know of other good solutions in this area for restricting to just {wikipedia, stackexchange, gitlab}?

sundarurfriend
I've recently been getting into Brave Goggles, and they have a `discard` feature so any site you don't explicitly mention get discarded from the results. For eg., see https://github.com/brave/goggles-quickstart/blob/main/goggle...

They use 1000 domains from Hacker News posts and do ranking and all that, but for your purpose, you can just have

    $discard
    $boost,site=wikipedia.org
    $boost,site=stackexchange.com
    $boost,site=gitlab.com
chaxor
Thanks, this is really cool. I found one from gist.github.com/JellyWX/ called jude.goggles that is basically already that.
johntash
I've been using a self-hosted instance of SearXNG (a meta search engine) to exclude a lot of spammy sites from search results. I'm not sure if you can force it to only search a list of domains, but might be worth looking into.
smcleod
The quality of the results seems really poor compared to Kagi. It seems to favour commercial / privately owned product sites over community information etc...
vignesh_warar
Hey, could you please link the search results page? I'd be happy to take a look and remove any commercialized sites from the network. The best part about this idea is that if there's a low-quality site, we can block the previous node, which will block all the other sites it was linking to. If a site is linking to a poor-quality site, there's a chance they might link to other poor-quality websites. You might ask whether this will also remove good sites from your network, but a good site will always find a way to enter the network.

The blocking feature has not been shipped yet.

smcleod
So I search for something I would have expected the top results to be discussions, blog posts etc... (controversial on purpose) https://usegrasp.com/search?q=Why+is+terraform+better+than+C...

Kagi (which I replaced google and ddg with over a year ago and have been really impressed with) provides far more useful / interesting results - https://file.io/umsdbo2JcEfg

Obviously it's all subjective etc... but yeah - hope that helps from my little sample size of 1.

vignesh_warar
Thanks!

I found one annoyingly low-quality site in the Grasp search results, which came from https://www.protocol.com (a news website). By default, outgoing links from news websites are not considered while building the network, as they will bring more junk. I maintain a list of UGC sites and news websites to ignore.

I will add https://www.protocol.com to the network building block list.

BTW, the link you shared https://file.io/umsdbo2JcEfg appears to be deleted.

timwaagh
What i do not understand is how it works. I had not followed anything yet it gave me a result. But the search results seem to be allright. It even had recipes for andijviestamppot.
vignesh_warar
By default, Grasp uses the HackerNews network, which is built from the top sites from HackerNews.

Here is a page about how it works: https://usegrasp.com/how

sundarurfriend
Right now, your homepage makes no mention of us following websites, just a search box. It's also easy to miss or not understand the "The network is built on top websites on Hacker News" message in the results page, so people might not understand that it's just the default and is customizable.

Your HN text above in the post does a good job of clarifying things, but I think some more messaging is needed in the website itself to make its potential clear. (Not many people are going to bother checking "How it works" by themselves.)

I'd suggest in the homepage, below "Grasp is a search engine for your personal network of high-quality websites.", have a "create your personal network" link (that could take you to the sign up page or wherever appropriate). And in the results page for the default network, have an info box at the top mentioning that this is just the default network, and the user can customize it, along with a to the "How it works" page.

darkwater
Oh, TBH I was expecting some kind of opensource, local-first scraper + search engine where you put a list of websites/info source "bubble" and it will scrape them over time and present them to you locally.
cdnsteve
Yes, except we want a shared resource for an index so we aren't all crawling sites in excess.

Eg, here's the latest central or distributed DB, the site in your personal list is on it so we don't need to send the crawler there because it has already been indexed and is crawled by someone else.

aflag
I think the top element with the LLM response is a bit distracting, since it takes so much longer to load than everything else. I think it'd be nicer if it was on a side panel or, at least, that the block allocated to it was fixed and the results didn't keep jumping around as the output for that is generated. Also, it seems odd that sometimes the citation numbers don't start at 1 (as you can see if you search for "vscodium marketplace")

Finally, just brutally honest feedback. I found the concept interesting and I would actually take it for a spin. However, you can only make a few queries before it asks you to sign up. I didn't find it that interesting to actually sign up just yet. So, I'll probably never truly try it out.

vignesh_warar
Thanks for your feedback. The LLM part is a bit buggy since I rushed its release. Currently, I am rewriting the entire LLM section to ensure that it is both fast and reliable.

May I ask how many queries you would like before signing up?

nicenewtemp84
Instead of just asking one person for their answer, you could randomize how many quarries people are allowed, and then see which one converts best.
aflag
Yeah, I should add a disclaimer that I'm just one user who's never been a product manager or anything of the sort. All my opinions may be completely rubbish. But I share them anyways because they may help someone.
aflag
I'm not sure if the number of queries is easy to measure. My experience is as follows:

I found out about it and decided to make some queries. I was trying to understand what it's good at and what it is not. I think I tried the same query a few times and small variations of it. I was trying to figure out when citations would start at 1 and when it wouldn't when it stopped working. I had already made a more generic query and I was trying to think some queries that would be probably answered by hackernews stuff. I would probably play around with a little more today, so even for that exploration the number of queries were too little. It even surprises me the limit is 20, I thought I made less than that.

Anyway, after I knew what it was good at and what it was not. I was planning on just keeping it in the back of my mind throughout the next days, specially when I go back to work (I'm on holidays now) and I thought I'd try it out whenever I had a natural query that I wanted to search for. That would be the true test in my mind.

In my mind, an unlimited 30-day free trial would probably be what makes most sense for that sort of thing. I do realise you'll probably want the user to sign in to offer that. Which, I might, reluctantly, do if the site actually offered me that option, but I didn't get the impression that I'd get anything like that. Signing up is a bit annoying because then I'll have to sign in from work as well, which I never really like to do, but I don't have a better way to offer a 30-day trial anyway. Alternatively, if you limited to 5 queries a day instead of a month it would already be a better proposition, because at least I'd be able to try it again tomorrow or when I get back to work. As it currently stands, I can only try it again in a month. I'll surely have forgotten about this by then.

Edit: actually just realised that not even $15/month gives you unlimited queries. 26 per day seems a bit on the short side of things. For the same reasons above, I think refreshing the queries daily or at least weekly makes more sense. I'd hate to run out of queries at the end of the month. Or maybe just have a price per query and charge based on that while allowing the user to set a limit.

xpe
This search engine charges beyond the free tier. I’m glad.

Why? An observation. As a consumer, if I pay nothing, then I’m likely to undervalue a service. When I undervalue it, I’m likely to use it less mindfully. Ergo, for more frivolous and unimportant things. This is exactly what some advertisers want: your attention without conscious intention. This is where emotions overrule rationality and open the doors to unnecessary spending.

xpe
Let’s help suggest pricing models that would work… not simply the minimum we as individuals would pay, but something to help increase innovation and competition in this space. As a former creator of a search engine, I’m glad to see this.
xpe
To think about pricing, it helps to think about:

1. How much signal do you get from pricing? For example, how much customer commitment do certain price points bring? How much does real world usage help? Strike a balance.

2. You want to attract early adopters. What pricing models are worth trying? How important is offsetting costs right now? Is traction and adoption more important, and if so, how much more? What metrics can help measure how to balance these goals?

3. How can you handle the scenario where you are lucky enough to get a lot of interest? How do last long enough to test your business without going bust? Your pricing model should be driven by these scenarios and your risk preference.

4. Leave yourself ways to adjust pricing without pissing people off. So if your initial pricing is tentative, be clear on that. Or let people lock in a monthly rate now in case it goes up later.

5. Create an internal quantitative model that predicts your expenses across some likely future scenarios. Tie your pricing model to some multiple of that. This can double as smart business planning to think about risk and what it takes to reach your goals.

6. Consider adjusting pricing based on how intensively someone uses your service, not simply based on search quantity, but your end-to-end cost. Recall that Twitter’s infrastructure costs are dramatically driven by a relatively small number of users with high fan out. What aspects of your offering are the most expensive? How can you mitigate these costs? How can you map these pricing differences to features that customers care about?

(Last edits: 12:31 pm EDT)

P.S. I created a search engine that never took off about 10 years ago. These questions would have helped me.

kjreact
I’m too used to getting search for free, so asking me to pay for this service is something that I’m very reluctant to do even though I understand that there’s no such thing as a free lunch.

What I’m willing to contribute is my computing resources. If a search service wanted to use my machine for web crawling that would be something I’m willing to trade for an improved search experience.

I’m not sure how feasible this option is because it doesn’t pay salaries or cover server costs, but it does help alleviate some of the computing costs, I’d assume.

abhayhegde
Nice site! Gave me useful suggestions which Google could not come up with.

A suggestion is to change the logo. Currently, it does not make a lasting impression since it is in cursive. Seems like a fashion brand rather than a tech product. I understand there is no need to stereotype (probably this is the one to break them), but even then fashionable alternatives could be tried out.

airstrike
Minor nit but I can't help notice it: the logo doesn't really match the product. Typography choices evoke certain feelings and attributes. Your choice to me says "women's* beauty / personal hygiene", not "find whatever you want at your fingertips"

Short of hiring a designer (and also for fun), may I suggest watching the Helvetica film? https://vimeo.com/570441741

marginalia_nu
Make sure to avoid The Helvetica Scenario though.
leroy-is-here
I looked at the logo to see what kind of non-sense you were talking about, but you are absolutely correct. That logo really does belong on a shampoo bottle.
alexvoda
And people wonder why all logos today are boring.

Of course they are boring if reification has become the norm.

Maybe just because something is cursive and italicized does not mean that it is automatically feminine.

leroy-is-here
I sign my name in cursive and do not think of it as feminine.
airstrike
> "all logos today are boring"

That's an overstatement. Many logos are boring, but certainly not all

I'm not suggesting a boring logo. I'm suggesting a good one that matches the feeling of "grasping"

elicash
I think you're referring largely to women's beauty/fashion products, not mens; I haven't noticed this type of logo with the products I use or on clothes I buy. And to whatever degree the logo might be feminine, I certainly don't think that's a bad thing.

People should push the envelope and not just do the same thing as everybody else! (Obviously do so with intent, but I wouldn't assume there wasn't any.)

vignesh_warar
I laughed out loud after reading your comment. I did not put a lot of thought into designing the logo. Thanks for the video.
gumballindie
A cool idea, but what I don't understand is why are many alternative search engines so poorly designed? I may be too used to google but I think the text should be a bit more readable. But I really like this type of search engines - always wanted search capability among my bookmarks and related websites. Perhaps even a bit of ai fine tuning around them.
ThinkBeat
You can do this with Kagi and "Lenses". Lenses lets you define a set of websites you want results from.
killthebuddha
ChatGPT plugins aren't out of beta yet, but this is 100% going to be eaten by one.

Edit: I guess that's not a super productive comment. I commend the authors for building and shipping something useful, I would personally use something like this.

savrajsingh
That’s what they said about Dropbox and google drive — amazingly both still exist.
vignesh_warar
I am not sure. The thing I want to build is far different from ChatGPT plugins. Could you please be more specific? Why do you think ChatGPT plugins would be direct competitor?
killthebuddha
A ChatGPT plugin that can answer search queries over your favorite sites would be basically trivial to build and would probably work very, very well. Not only that, but it would be immediately composable with any other ChatGPT plugin. All that said, I'm all for building alternatives to ChatGPT. I think we're heading to a future where the ChatGPT plugin store is the iOS app store of LLM apps.