APSE – A Personal Search Engine | Hacker News Reader

APSE – A Personal Search Engine

apse.io

141 pointsthunderbong61 comments

I’ve seen a handful of this kind of “Google, but only for things I’ve seen before” app. I think it’s something the world needs, but there are a lot of different approaches and I don’t think anyone has quite nailed it.

Ultimately the best solutions will likely use many different cataloging strategies depending on the content, and will allow you to tag or otherwise organize important content.

Funny enough if I had such an app I could make a list of 4 or 5 examples, but right now can only find one:

https://github.com/amirgamil/apollo

I remember seeing one posted to HN that used the browser API to essentially dump websites you visit to disk.

There’s also bookmarking and archiving tools like:

https://pinboard.in/

https://unmark.it/

https://www.linkace.org/

https://archivy.github.io/

https://github.com/kanishka-linux/reminiscence

https://perkeep.org/

Personally I use Historio.us, but maybe it is not in the same category you are talking about?

(I found it here on HM and so far - I have used it for a year or so - seems to be a good fit for my needs)

What if I only want a nice front-end to search my chrome browsing history? Is there a good app for that? I know I can visit the chrome history page but looking for a better UI

Apollo struck me as the better of the two similar things posted today, mostly because this APSE thing purports security while operating on somehow securely storing screenshots taken of your desktop while you're working on it (Yeah haha what about all my chat apps)

I used https://getmemex.com/ for a while but it was too buggy to be relied on.

There's https://timesnapper.com/ for Mac and Windows.

I used the Windows version for a while but had to stop using it because it was buggy: it ignored the disk usage limit I'd set and proceeded to fill up my entire boot drive with images.

No. This is a fundamentally different idea. This is not a bookmarking/archiver/knowledge base/search engine. For my personal searchable knowledge base I personally use DevonThink. This tool is crazily different -- it screenshots your screen and makes your entire screen history searchable. That's game-changing because it captures information that you don't consciously choose to save or store. That's a whole class of information that's not available with existing tools.

I think the main issue with 'Searching the personal bookmarks' is the lifetime of the information in those content we've bookmarked, Some like history, art etc. are ageless but for almost everything we search day-to-day on the search are for latest information on something.

I really like this - but I'd rather just pay $200 once than pay a subscription. If an issue is maintaining the models, you should make that a separate line item where you can pay as a subscription for updates or pay for the model on a yearly/monthly basis as necessary (some people might be OK with the level of OCR at any given moment).

What would be really interesting is if you could have some hardware that hooks into the HDMI port and with every flicker of the screen record and OCR everything, and add some diffing and compression and some really cool categorization algorithms and you could have something amazing.

---

Obviously this is an MVP but I imagine the perfected version of this would be able to see the difference between screenshots and use some sort of strongly-connected-component like algorithm and figure out which things are "browser" which things are "browser--hacker-news", and so forth. Create a great UI around this idea and even better UX and you have something interesting.

Hell, miniaturize some camera and hook it onto glasses and you have that one Black Mirror episode.

There was one April fools prank by either torrentfreak or someone similar (maybe tbp team?) Who said how they are building a technology that takes screenshots of everything you do every second and stores on "cloud" so you can rememeber what you did when you did. Maybe that was a jab at patent trolls or something im not sure now. That was a hilarious piece then

Hey Ryan,

This is interesting. I don’t know if enough people will want this to make it a viable business, but I like the idea a lot.

I would suggest this would be an easier sell if your main client was open source. If it were open source, you would need a slightly different economic model, but here are things you could charge for:

* E2E cloud encrypted backups

* Better AI models (or just different features, like e.g. searching for YOLO categories in photos)

* Support

* Group features as a sort of realtime knowledgebase. Note you’d need a bunch of privacy stuff added. But you could charge a lot here if it worked I think.

* Spyware / time tracking software sold to the odesks of the world.

In exchange, you might get quite a lot of people using it. As an alternate, you could make the lowest tier free forever if people allowed anonymous-ish data submission, that would let your free users pay by letting you tune better models, and your paying users could benefit and keep their privacy.

Happy to talk more if you’d like.

Strongly reminds me of Google Desktop [1]. I wonder if there is still alternatives around. Searching my own computer is still such a terrible experience. However, it seems not much business can be made out of this.

[1] https://en.m.wikipedia.org/wiki/Google_Desktop

Google Desktop was just a cross-platform version of the macOS Spotlight feature.

I’d be pretty unwilling to pay a subscription service for this because I know I’d only see the value after having it for quite a while. On the other hand I’d gladly pay $100 upfront if I knew I could have it forever as a desktop app.

This is pretty clever. It's taking periodic screenshots of your screen and running OCR on it, then indexing the results.

Other than feeling a little icky about an app watching my screen -- not to mention picking up literally everything I do -- my main question from the demo video is whether it's able to intuit the URLs when you're looking at Web pages. There's going to be a fair amount of things it can't just link you to, in which case all it can offer is the screenshot it took. (Which is useful to be sure)

I'm curious as to how quickly the storage space usage grows with all those screenshots.

(I'm just talking out of my ass but) I'd imagine the screenshots can be compressed pretty efficiently, as the content of any photographic detail like images/picture don't need to be preserved with outstanding quality..

Yeah I bet there's a lot of stuff you can do to greatly reduce space usage without even nerfing resolution or losing color. For example if you are always using the same desktop background you could save that once and then just store a reference to it for all subsequent screenshots where it also appears. Also, window chrome and things like webpages and IDE windows will compress well, since there are so full of long runs of the same background color

I guess the screenshots are not stored, only the text extracted using OCR.

Based on the “How it Works” section of their website, it looks like both the screenshot and the text are saved.

https://apse.io/faq/

"It depends. Screen resolution and multiple monitors affect how much storage an image takes. For a general guideline, if you have a 1080p monitor and run it nonstop for a full 8-hour workday, you could expect in the range of 200-500MB.

Apse is fully configurable. If you find it takes too much storage, it can be set to delete snapshots after a certain number of days. For space savings, by default images are saved at a lower quality. If you would rather keep full-quality snapshots, you can set that in your configuration as well."

>> Other than feeling a little icky about an app watching my screen

I understand that feeling, but it is misguided.

We have become so used to feeling creepy while being watched by others, that we get the same feeling when just watching our self.

There could be many reasons for Apse to fail, but this should not be one.

This reply assumes someone else cannot get ahold of that data.

Agreed. It's the difference between privacy and security. I think the issue here is the former.

>> Other than feeling a little icky about an app watching my screen

Doesn't your OS already know everything that's on your screen (and hard drive, for that matter) at all times?

>Privacy First

Yeah no. If I don't see source code then I'm not bitting.

Do you use Windows, macOS, iOS, Android? Have you seen the source code to any of these? And no, AOSP doesn't count. It's a fraction of the code running on consumer Android devices.

Don't get me wrong, I prefer open source solutions, but I'd more prefer a company focused on building privacy-first products and being able to fund themselves.

> Do you use Windows, macOS, iOS, Android?

Their answer to this might simply be "no". There are people who just wont touch devices that are shipped with code / programs that don't respect the privacy of the user, unless installing free software to override the stock proprietary system is an option.

Have you seen Windows, macOS, iOS, Android advertising their privacy?

https://privacy.microsoft.com/en-us/windows10privacy

https://www.apple.com/newsroom/2021/06/apple-advances-its-pr...

https://developer.android.com/privacy

Yes.

Um, yep, I see macOS and iOS advertising that all the time, not only through their mainstream TV ads, but also in recent years like every time I upgrade one of my family's Apple devices.

Maybe stupid question, but can’t someone do an open source bash loop that will run a screenshot and ocr script every second and kill your business?

I’ve been wanting this for myself for so long; very happy it was built! Would love a one time payment option though.

I'm calling it now: this is going to be HUGE. This is insane: it's unlocking access to a whole class of data that was previously inaccessible. This is not for information that you consciously choose to save, organize, and store: for that there are millions of note-taking apps out there. This app allows you to search information that you did NOT consciously record by capturing your screen history in the background. That set of data: stuff I've seen but did not save is not available anywhere except your browser history, and even that is not full-text searchable. This captures information from ANY app, not just your browser. It allows you to answer the question which no other app can: which app did I see that piece of info in? Right now the only alternative to this kind of app is your brain, and this can rightfully be called a brain extension.

Calling this a personal search engine is a bad idea. It needs a better name. But I know this is going to be one of those paradigm-shifting-innovations because it's allowing searchable access to a whole class of data that was previously inaccessible.

I love the idea of screen-shotting in the background, but I don't need the OCR + Search Engine, because I already use DevonThink, which does OCR + full text search really, really well. If Apse just stores its screenshots to a folder on the hard-drive, that's all I need.

I'm positive there are programs that can take a screenshot on a timer already, or at worst easy ways to achieve that

Hints of "You can already build such a system yourself quite trivially" ...

ShareX + Tesseract seems like an appropriate (and open-source and local) combo

Very interesting! But the price seems steep for what appears to be only an incremental improvement over using OS tools to search computer contents and you email's search feature to find messages, etc. Unifying it may be a nice convenience, but I'm not convinced per the price. Still, there's a 14 day free trial, so I'll try to prove myself wrong.

I'd be very wary of installing an app that took screenshots randomly

I like seeing this, I’ve had multiple discussions about building something similar but always run into the same difficulty with convincing individuals to pay for this. I however see a great b2b idea in this because finding stuff in a company, especially a large one, is a complete pain in the ass and they definitely have the budget and willingness to pay a reasonable subscription price. The difficulty here is Google actually has a product for that in gsuite/workspace but they’ve let it languish on the vine. So the question there is, was it not something they saw being able to make money, or not able to make enough money for a company already operating on hundreds of billions in revenue? I kind of hope someone decides to find out though.

Scary thing is if Google offered this as a free app, most people would happily use it.

Google Desktop.

https://en.m.wikipedia.org/wiki/Google_Desktop

But it was...

(Wait for it)

Discontinued.

Something about not being aligned with their cloud strategy IIRC.

It didn't OCR, but indexed emails, documents, visited pages by content, etc. (And enriched your normal Google search results - if it had permission).

Was quite awesome.

It had a I-remember-something-else-that-day mode of search where you could find an email or doc from that day and then look for other items indexed on the same day/hour.

First, kudos to the launch, the product seems pretty interesting and compelling to a lot of users here!

In a sense, that's the source of this mini-rant - not the product, but the feeling that something like this is compelling.

Instead of reducing complexity, we simply are trying to manage it, navigate it, index it. "Keep generating more noise, we can always find the signal!"

It just feels like yet another step in the wrong direction for us as a species.

For those really motivated to avoid google and improve search results, there is also DEVONagent Pro. (Mac only) https://www.devontechnologies.com/apps/devonagent It is desktop & macOS only, but does filter out quite some noise.

Founder here, I am happy to talk about it, my email is ryan at the submitted domain.

Your bio says you run 3 separate projects.

How much time do you spend working on each?

Neat. So much potential, scary potential with this.

Hey, this is a cool concept with the screenshots and shit. Not something I’d use but definitely a unique take.

This is a great alternative to me having to archive everything I do. I often struggle to recall that one article that had the data on the thing or that one message someone sent me at some point. It would be great if every app (like my browser) auto archived all that I see (removing the need for OCR) in some standard format, but short of that this is a good alternative idea. I do wonder how they avoid archiving secrets that users may want to manage in a sensitive way though.

This is just a simple bash script and Apache Tika (https://tika.apache.org/). You could script this together in minutes. Try https://github.com/flameshot-org/flameshot and feed the results through Tika to OCR the results.

You may want to update your logo:

The importance of Ansible trademarks

The Ansible trademark and the “A” logo in a shaded circle are our most valuable assets. The value of these trademarks encompass the Ansible Brand. Effective trademark use is more than just a name, it defines the level of quality the customer will receive and it ties a product or service to a corporate image. A trademark may serve as the basis for many of our everyday decisions and choices. The Ansible Brand is about how we treat customers and each other. In order to continue to build a stronger more valuable Brand we must use it in a clear and consistent manner.

The mark consists of the letter “A” in a shaded circle. As of 5/11/15, this was a pending trademark (registration in process).

https://docs.ansible.com/ansible/latest/dev_guide/style_guid...

The slanted bar in the "A" is slanted the other way, the letter "A" is not alone but includes the full name "APSE", the rest of the typeface is pretty different (the "E" for example), and it's inside a rectangle, not a circle.

IANAL but this seems like a pretty weak resemblance.

Edit: I hope you aren't an official representative of Ansible.

Not a rep

> The Ansible trademark and the “A” logo in a shaded circle are our most valuable assets. The value of these trademarks encompass the Ansible Brand.

Ironically not used a single time on the front page of ansible.com as far as I can see on my iPhone XR.

(I also couldn't find a single case of A in shaded circle on the APSE page but I admit I could be missing something.)

Their logo

And where does that logo show up on ansible.com? Confirming what the parent comment mentions: I don’t see this logo anywhere (also on mobile).

For the OPs website, the top left (on mobile) uses the Ansible A, mirrored. Not a big deal just trying to make sure they’re aware that it’s trademarked

I honestly don't think there's enough similarity there to be considered any sort of trademark infringement. I think if Ansible tried to test that in court they'd likely get shut down pretty quickly. Just for starters, even if you mirror the A like you say, they're not the same font when overlaid for direct comparison. There's also the difference between an A in a black circle and four letters (APSE) side by side in a black rectangle. No judge in their right mind would see any sort of trademark infringement between those two logos.

Trademark law is less about these sorts of technicalities and more about the general impression in the mind of the consumer. Would someone confuse the two logos?

When seeing just the "A" as used in Apse.io's favicon, I wouldnt be able to tell if it was Ansible or Apse without double checking which way round each was, and which font each was.

Most of the "Kelvin Clein" fakes that you see are not actually skirting around the trademark, and rely on never actually being tested in court. At most it pushes enforcement from the front lines of "this is a fake" to the courts with "prove this is a fake".

For instance, in the UK, an off brand clone of Hendricks Gin has been sued for trademark infringement and this article gives a good breakdown of the legal issues[0]. Examples of the different designs here[1].

Its not a given that Apse is infringing, or that Red Hat would sue over it. But given how much easier it is to change these things when the project is young, I'd suggest the author change the logo. Apart from anything else, the similarity to Ansible works against Apse's own notability.

[0] https://www.lexology.com/library/detail.aspx?g=dcae4558-12e3... [1] https://ichef.bbci.co.uk/news/2048/cpsprodpb/1580C/productio...

Ansible makes a big deal out of being a single letter in a circle, the other is four letters in a rectangle.

Not in a circle which seems to be important to Ansible.

Edit: Ansible is very clair about the trademark as quoted above:

> and the “A” logo in a shaded circle

(emphasis mine)

Reply on news.ycombinator.com