Welcome back to the deep dive.
Download transcript (.srt)
0:00
0:01
0:03
0:06
0:07
0:10
0:12
0:16
0:18
0:19
0:20
0:22
0:23
0:25
0:26
0:28
0:32
0:34
0:38
0:40
0:43
0:44
0:46
0:48
0:50
0:53
0:56
1:04
1:07
1:11
1:12
1:13
1:17
1:19
1:20
1:20
1:25
1:27
1:29
1:30
1:33
1:35
1:38
1:43
1:47
1:49
1:50
1:53
1:54
1:54
1:57
2:00
2:00
2:02
2:04
2:07
2:08
2:09
2:10
2:11
2:12
2:14
2:18
2:19
2:21
2:24
2:24
2:25
2:29
2:31
2:32
2:33
2:36
2:39
2:43
2:46
2:47
2:47
2:50
2:51
2:53
2:55
2:59
3:00
3:02
3:04
3:04
3:06
3:09
3:09
3:11
3:13
3:15
3:17
3:18
3:19
3:23
3:24
3:26
3:28
3:29
3:31
3:33
3:34
3:37
3:39
3:42
3:43
3:46
3:47
3:49
3:51
3:53
3:54
3:57
4:00
4:03
4:05
4:07
4:09
4:10
4:12
4:14
4:17
4:18
4:20
4:21
4:23
4:25
4:27
4:28
4:31
4:31
4:34
4:37
4:39
4:41
4:44
4:45
4:47
4:48
4:50
4:53
4:55
4:57
5:03
5:05
5:07
5:09
5:11
5:12
5:14
5:17
5:19
5:20
5:22
5:24
5:25
5:28
5:30
5:32
5:35
5:36
5:39
5:41
5:42
5:44
5:46
5:48
5:50
5:51
5:54
5:56
5:59
6:02
6:05
6:08
6:11
6:13
6:17
6:20
6:21
6:25
6:26
6:27
6:28
6:31
6:35
6:38
6:39
6:40
6:41
6:42
6:43
6:46
6:47
6:49
6:52
6:55
6:58
7:02
7:02
7:04
7:06
7:07
7:09
7:11
7:13
7:15
7:16
7:18
7:20
7:21
7:24
7:28
7:29
7:31
7:34
7:36
7:36
7:39
7:42
7:45
7:47
7:49
7:51
7:53
7:58
7:59
8:00
8:03
8:05
8:08
8:10
8:12
8:14
8:15
8:18
8:21
8:23
8:26
8:27
8:31
8:35
8:37
8:38
8:39
8:42
8:44
8:47
8:49
8:52
8:53
8:55
8:56
8:59
9:02
9:04
9:06
9:08
9:10
9:11
9:13
9:15
9:16
9:18
9:19
9:21
9:24
9:27
9:28
9:31
9:32
9:36
9:37
9:39
9:42
9:45
9:46
9:48
9:48
9:51
9:53
9:55
9:57
9:59
10:01
10:02
10:04
10:07
10:09
10:11
10:14
10:15
10:16
10:18
10:21
10:22
10:25
10:26
10:28
10:29
10:32
10:35
10:37
10:38
10:40
10:43
10:45
10:48
10:50
10:51
10:55
10:58
11:01
11:03
11:06
11:08
11:09
11:12
11:15
11:16
11:17
11:19
11:21
11:23
11:25
11:29
11:31
11:32
11:33
11:35
11:37
11:39
11:40
11:42
11:43
11:45
11:48
11:50
11:53
11:55
11:56
11:59
12:03
12:07
12:09
12:11
12:14
12:17
12:19
12:20
12:22
12:24
12:27
12:27
12:28
12:30
12:32
12:35
12:37
12:38
12:40
12:41
12:46
12:47
12:49
12:50
12:53
12:55
12:56
12:58
13:01
13:03
13:07
13:08
13:10
13:13
13:15
13:18
13:21
13:24
13:25
13:26
13:29
13:32
13:34
13:35
13:37
13:37
13:41
13:43
13:45
13:46
13:49
13:51
13:54
13:57
14:00
14:03
14:05
14:08
14:13
14:15
14:17
14:19
14:20
14:22
This is where we cut through the noise
and focus on what really matters from all the stuff we read
and research.
Today, we're tackling a very, very modern problem.
Just digital clutter, right?
All the links, notes, PDFs we save.
We're diving into a tool called karakeep.
Yeah, karakeep.
It used to be called Hoarder.
Some listeners might remember that name.
Right, Hoarders.
And look, this isn't just another bookmark app.
We're looking at something different here.
It's self-hosted, and it actually
uses AI to help organize your own personal digital stuff.
The real insight, I think, that we're exploring
is how these powerful tools like AI organization, which usually
live in the cloud on company servers,
are kind of moving back, back into our own hands.
It's a definite trend.
OK, but before we really unpack karakeep,
we want to give a quick shout out to our supporter
for this deep dive, safeserver.de.
They help with hosting for software exactly like this,
and they support you in your digital transformation journey.
You can find out more at www.saserver.de.
So karakeep, as you said, it was hoarder,
and it's really built for, well, people like us maybe.
The data hoarders.
Guilty as charged.
We save everything, links, notes, images, you name it.
But the trick is finding it again later.
That's the hard part.
Totally.
And karakeep's angle is that it's self-hostable completely,
which means you run it, you control the server,
you control where your data goes.
And that's the hook, isn't it?
Because we've seen tons of read-it-later apps, Pocket,
Instapaper, all those.
But karakeep stands out because of, well,
two things mainly, that AI integration for organizing
things automatically and how it fights back against link rot.
Yeah, link rot is a killer for archives.
So it's more than just saving links.
It's like building an intelligent personal archive
that hopefully lasts.
Exactly.
And for anyone listening who's maybe newer to this,
let's just nail down what self-hostable really means here.
Good idea.
It's not like signing up for Spotify or something.
You're not using someone else's service.
You install karakeep on your own computer or maybe
a server you rent.
So you're in charge.
Totally in charge.
You decide the rules.
You control the data.
You're not depending on some big tech company
to keep your personal knowledge base safe or accessible.
It's about independence.
It's like instead of using the public library
for your most vital notes, you build your own secure vault
for them.
That's a great analogy.
And speaking of custom things, the name Karakeep,
it tells you a lot about the philosophy behind it.
Oh, yeah.
What's the story there?
Well, Karakeep comes from Arabic.
The word Kayasi Keek, Karakeep, it basically
means odds and ends, miscellaneous clutter, stuff
that doesn't look organized, but it has personal value.
Huh.
OK.
So it's not about forcing you into neat little folders.
Not at all.
It acknowledges that, yeah, the stuff
you grab from Reddit or Twitter or Hacker News,
those random notes, those PDFs, it might look like a mess,
but it's your mess, and it's valuable.
Karakeep embraces that clutter.
Or makes it searchable.
Exactly.
It gives you the tools to find things
within that valuable mess using smart tech.
I like that.
It's built for that impulse we all have.
Oh, got to save this for later.
But it turns that impulse into something, well, actually
useful long term, a real knowledge base.
Right.
It elevates the read it later idea.
So what exactly can it hoard?
What does the everything part cover?
It's pretty broad, actually.
It handles your standard web links, obviously,
but also simple notes.
You can just jot things down directly.
Images, PDFs, too.
OK.
And when you save a link, it's not just storing the URL.
It automatically goes out and fetches
the page title, the description, maybe a preview image,
gives you context right away.
Which is way better than just a list of naked URLs.
Much better.
But then you get to the really interesting part,
the intelligence layer, as you called it.
Yeah, the AI stuff.
But that seems like the secret sauce here.
AI tagging, summarization.
So imagine you save a long, complex article.
CareKeep can use AI, potentially external services,
like OpenAI's models, to automatically
suggest relevant tags.
Or even generate a quick summary for you.
OK, that's useful.
But hang on, if it's self-hosted,
sending my saved articles to an external AI,
doesn't that kind of defeat the purpose of privacy?
Ah, excellent point.
And the developers thought of that.
This is key.
CareKeep is specifically designed
to work with local AI models.
It supports a framework called Olama.
Olama, right.
So I can run an AI model right there on my own server.
Exactly.
You can download and run various open source AI models locally
using Olama, and CareKeep talks to that.
So you get the smart tagging and summarization,
but your data never leaves your control.
No third party clouds involved, unless you explicitly
choose that.
That's huge for the self-hosting crowd.
Privacy and power.
It's a major selling point.
And speaking of keeping your data safe and useful,
let's talk to Linkrot again.
Yes, the bane of bookmarks.
Click a link you saved a year ago, and poof, 404 not found.
It undermines the whole idea of a personal archive, right?
So how does CareKeep tackle that?
Seems like an impossible fight.
Well, it uses some clever archival techniques.
It doesn't just save the link.
It aims for a full page archival,
meaning it uses tools like one called Monolith
to essentially download the entire live web page.
Yeah.
Not just the text, but the images, the formatting,
the CSS, everything.
Wow.
And it bundles all of that into a single, self-contained HTML
file that lives on your server.
So if the original website disappears tomorrow,
you still have the complete readable content saved.
OK, that's not just a bookmark.
That's like taking a perfect permanent snapshot
of the page, a portable copy.
Precisely.
It's robust archiving.
And it does something similar for media, too.
If you save, say, a YouTube link.
Don't tell me it saves the video.
It can.
It uses tools like YTDLP, which many might know,
to automatically download and archive the video file itself.
So my saved stuff is basically immune to the original source
disappearing, link-proof and media-proof, kind of.
That's the goal, to make your archive truly resilient.
OK, so we've got all this stuff saved, links, notes, images,
PDFs, archived pages, even videos.
How on earth do you find anything
in this potentially massive pile of, what was it, Karakib?
Well, yes, your personal Karakib.
Well, search is critical, obviously.
And Karakib uses a modern search engine called MylaSearch.
MylaSearch, I've heard of that.
Supposed to be fast.
Very fast.
And it provides full text search across everything.
Your notes, the original link URLs, descriptions, tags,
and the actual content of those fully archived pages
we just talked about.
Everything is indexed.
Everything.
But they went even further.
They included OCR.
OCR, Optical Character Recognition for Images.
Exactly.
So let's say you saved a screenshot of something
important, or maybe a photo of a whiteboard diagram,
or even a restaurant menu you wanted to remember.
Carekief's OCR will actually scan that image,
find any text within it, and make that text searchable too.
Whoa.
So I could search for meeting notes,
and it might find that whiteboard photo I saved.
That's the idea.
It makes even your visual clutter searchable.
It's a really thoughtful usability feature.
That's actually incredibly useful.
OK, so the back end is powerful.
Archiving is robust.
Search is smart.
What about just using it day to day?
Is it easy to get stuff in?
They seem to have put effort into that too.
There are browser extensions, naturally, for Chrome, Firefox,
make saving links quick.
Standard stuff, but essential.
Right, and native mobile apps iOS and Android,
so you can access and save stuff on the go.
Good.
Plus, for more advanced users, there's a REST API,
support for bulk actions if you're importing lots of stuff,
even SSO single sign-on integration.
And yes, there's a dark mode.
Dark mode, always important.
OK, let's peek under the hood a bit.
For listeners thinking about running this themselves,
the tech stack gives clues about stability, how well it's built.
Sure.
It's definitely a modern stack.
The front end uses Next.js with the app router, which
is pretty current and performant.
For the database side, they use something called Drizzle ORM.
Authentication is handled by Next off.
After the communication between the browser and the server,
they use TRPC.
Whoa, OK, lots of names there.
Drizzle, TRPC, Next off.
For someone maybe just dipping their toes into self-hosting,
does this mean it's super complicated to set up
and run compared to, say, an older PHP app?
That's a fair question.
I mean, yes, the underlying tech is sophisticated.
But the reason developers choose tools like Drizzle or TRPC
is often to make things more reliable and faster
in the long run.
OK.
TRPC, for example, helps prevent certain kinds of bugs
between the front end and back end.
Drizzle offers strong typing for database queries.
The initial setup likely involves
Docker, which is standard for self-hosting these days,
but once it's running.
The goal is stability and speed.
Exactly.
The complexity is there to provide a smoother, faster
experience, especially as your archive grows.
You don't want it bogging down when
you have thousands of items saved.
This stack is built for scale.
Right, complex engine for simple, fast driving.
Makes sense.
So why did the creator build this?
Was it just a technical challenge?
It was partly that, yeah.
The creator is a systems engineer,
so they have the skills.
And they mentioned wanting to keep their web development
skills sharp, but mostly it came from a personal need.
A frustration with existing tools.
Pretty much.
They were already a heavy user of bookmarking and note
taking apps.
They mentioned getting hooked on the idea by Pocket initially.
Like many of us.
But Pocket is proprietary, cloud based.
Once they moved towards self-hosting, that was out.
They apparently liked another app called Memos for quick notes.
Memos, yeah.
That's another popular self-hosted one.
Right.
But they found Memos lacked crucial features
for their way of saving stuff.
Specifically, link previews, seeing
what a link was about instantly.
And importantly, automatic tagging.
Ah, back to the AI tagging.
Exactly.
Without that, their saved links just
became this massive, unmanageable list.
Basically, unusable clutter.
So Carrot Keep was born out of that need
to add intelligence and better archiving to the self-hosted
note-taking idea.
Got it.
That personal story really helps place Carrot Keep
in the competitive landscape.
To really get why someone would pick this,
we should probably compare it directly to some alternatives.
Definitely.
And Carrot Keep really does sit in a specific, interesting
spot.
It's trying to blend the polish you see in some commercial apps
with the core principle of self-hosted independence.
OK, so who are the main competitors or inspirations?
You mentioned Pocket.
Any others on the commercial side?
The creator specifically mentioned MyMind
as a close inspiration.
MyMind is known for its very visual, AI-powered
organization.
Looks great, works smart.
But it's commercial, proprietary, cloud only.
Carrot Keep aims for that same kind of smart visual feel,
but puts you in control of the data and the hosting.
And Pocket, as we said, got the creator hooked.
But again, no self-hosting option.
Right, so what about the open source rivals?
We mentioned Memos.
Yep, Memos is great for notes, but Carrot Keep
adds the archiving, the previews, the AI tags
that Memos lacks.
Then there's Omnivore.
Omnivore, yeah, another read it later open source option.
It is, and it's cool.
But apparently its architecture relies pretty heavily
on Google Cloud infrastructure right now,
which makes tree self-hosting, like completely independent
self-hosting, a bit difficult, or at least not
their main focus.
Whereas for Carrot Keep, self-hosting
is priority number one.
Exactly, it's designed first for self-hosting.
Then you have the older, really established players,
like Wallabag.
Wallabag's been around forever, right?
PHP-based.
Yeah, very mature project.
But maybe the UI feels a bit dated to some.
That was the creator's perspective anyway.
And finally, there are other open source link managers
like Linkwarden or Shiori.
OK, and how do they stack up?
They definitely fulfill the self-hosting need,
but they generally lack that sophisticated AI layer,
the automatic tagging, the summarization, the OCR search
that Carrot Keep is really leaning into.
So Carrot Keep's niche is becoming really clear.
It's for people who want that cutting edge AI organization
plus serious archiving against Linkrot
and are committed to self-hosting.
That's it, precisely.
It's for the power user, maybe, who
sees the value in AI tools but doesn't
want to hand their data over to a big corporation to get it.
OK, perfect.
Let's try and synthesize this.
What are the key takeaways for you, the listener,
considering Carrot Keep?
Well, first, you're looking at a really robust system
for tackling digital clutter.
It's built to be future-proof.
With that strong archival focus.
Right, fighting link rot.
Second, it's open source AGPL 3.0 license.
And despite being relatively new,
it's got serious momentum.
You mentioned the GitHub stats.
Yeah, over 20,000 stars, nearly 1,000 forks,
that's a lot of interest.
It really is.
That suggests an active community,
ongoing development, bug fixes, new features.
It's not likely to just disappear.
That community support is vital for open source projects.
Okay, so, final thoughts.
Something provocative for people to chew on.
I think it comes back to that core idea we started with,
bringing power back to the user.
We're seeing these incredibly advanced capabilities,
AI summarization, classification, deep search,
that used to be exclusive to giant cloud platforms.
And now they're running on our machines.
Exactly.
When you control the hardware your knowledge lives on,
and you control the AI that helps you understand
and organize that knowledge,
well, that fundamentally changes your relationship
with information, doesn't it?
How so?
You shift from just being a consumer reliant on platforms
to being like an independent owner and curator
of your own digital brain,
your own knowledge infrastructure.
Owning your knowledge infrastructure.
That is a powerful thought,
has huge implications for how we manage information,
how we learn, maybe even how we think going forward.
Okay, on that note, just a final reminder
that this deep dive was supported by safeserver.de.
They handle hosting for software like karakeep
and can help with your digital transformation.
Check them out at www.safeserver.de.
And yeah, if this sparked your interest,
definitely explore the world of self-hosting
and these kinds of advanced organization tools.
It's a fascinating space.
Absolutely, thanks for diving depth with us today.
We'll catch you on the next one.