HomeVideos

DHH Omarchy Quattro Announcement | TheStandup

Now Playing

DHH Omarchy Quattro Announcement | TheStandup

Transcript

1125 segments

0:00

Today we have on legendary

0:03

uh pusher of operating systems

0:05

specifically making the year of Linux

0:07

2026 actually happening DHH. We have

0:11

also with us trashdev amazing snack

0:14

connoisseur and teach and today we are

0:16

going to be talking something extremely

0:18

special.

0:23

Uh, anyways, sorry. I don't know if you

0:25

if everybody knows this, but it turns

0:28

out we have this right here. We have

0:33

some Omachi news coming in. So, I'm

0:35

looking right here at the board, at the

0:37

conbon board. What are we doing today on

0:38

the stand up? We have to talk about

0:40

Omachi. And I am saying that correct, by

0:42

the way. And so, with DHH on, we figured

0:44

that maybe he could do a little bit of

0:46

setting the stage on why we're talking

0:48

about Omachi specifically, cuz that kind

0:50

of seems silly. It's been it's been out

0:53

for so long at this point. I still have

0:54

my transitions on. It's we we've had

0:56

it's been out for so long. Why uh why

0:59

are we talking about it?

1:00

>> Well, we're going to talk about Amachi

1:01

because it was basically rewritten from

1:04

scratch and the latest version, Umachi

1:06

Quattro, which is the first version

1:08

that's gotten its own moniker, just

1:11

dropped on Friday and is ushering in

1:15

nothing less than the future of the

1:17

personal computer. We are making

1:19

computers malible, personable,

1:23

changeable, and it's all thanks to

1:26

agents solving the biggest problem we've

1:29

had in Linux that they're a pain in the

1:31

ass to set up. They're a pain in the ass

1:33

to configure. They're a pain in the ass

1:35

to diagnose.

1:37

Yet, thanks to the wonder of magical

1:40

premium tokens, that is no longer true.

1:42

We have AI to do everything for us. And

1:46

not only is Amachi Quattro virtually

1:49

fully AI developed, I spent about three

1:52

months together with Ryan Hughes and

1:54

others, prompting our way to this

1:58

incredible distribution that we have

2:00

today. And then the experience of

2:03

actually using it once you set up your

2:05

default agent is just nuts. You can make

2:08

anything happen. We've rewritten the

2:10

whole thing in QuickShell. We've dumped

2:12

about 17 different dependencies. We've

2:15

shrunk the ISO by 30%, we've increased

2:18

the install time by like 40%, it now is

2:22

up and running on a fast computer in

2:23

about 45 seconds. It was already quick.

2:26

>> I lost my world record.

2:28

>> You had you had Omachi 3 world record,

2:30

which was it might have even been Omachi

2:32

2 world record. It was very fast.

2:34

>> True.

2:36

>> Hey, Brian, don't forget scrum plannings

2:39

later.

2:50

No,

2:53

I can't do this.

3:02

[music]

3:05

Product manager here. We've decided to

3:08

put you on call. [laughter]

3:14

I CAN'T I I can't do I can't do on call.

3:18

>> You're on call?

3:21

>> Luck.

3:23

>> What are you saying? With Sentry on call

3:26

is so much fun. It has all of our

3:28

errors, logs, [music]

3:29

spans, metrics, replays, and even

3:32

releases in one spot. It has all the

3:35

context we need [music] to solve any

3:37

bug. Even if half the time Sentry beats

3:39

me to the fix anyway.

3:41

>> Just cuz software development is torture

3:43

doesn't mean on call has to be.

3:45

>> Wa, that's amazing. Does Sentry also

3:47

attend my meetings?

3:48

>> Oh, that's the best part. No.

3:50

>> Sentry [music] can't actually make on

3:52

call fun, but it can make it less scary.

3:54

Everything you need to fix real issues

3:56

in [music] one place.

3:58

>> Buckaroo. Got that 6 p.m. postmortem

4:02

planning. Don't forget

4:05

>> it was already ridiculously fast, but

4:07

we're making it ridiculously leer fast.

4:10

Um, [laughter]

4:11

because we've just optimized everything

4:13

and it's just it's tighter, it's better,

4:15

and it's completely malible. We now have

4:17

a plug-in repository. There's a

4:19

community-led version on

4:20

omachiplugins.com

4:22

that launched six days ago, and it

4:25

almost has a,000 plugins. Now, we're on

4:27

chart at least to get a thousand plugins

4:30

within the first week. People have made

4:33

everything happen. There's a million

4:34

games. There's AirPods controls. There

4:38

is flight radar thingies. There is just

4:41

all sorts of There's even a plugin for

4:44

uh finding back a girlfriend. I just saw

4:46

that. Um Whoa.

4:48

>> That's Hey, that's a good plugin. Okay.

4:50

>> It just crashes or what? [laughter]

4:52

>> It turns off It turns off Arch. It

4:54

installs Windows and you have to go out.

4:57

>> But but as amazing as all this is, this

5:00

launch is just nuts. We're running at

5:01

like 5,000 ISO downloads a day right

5:05

now. Um, even crazier, just dropped, I

5:09

don't know if you caught this, Prime,

5:10

because you've been on your uh you've

5:11

been on your stream, but literally uh an

5:14

hour ago just dropped a huge

5:16

announcement that um I'm incorporating

5:18

the Omicron Foundation with $8 million

5:22

in pledged funding from um just like a

5:25

few people you might have heard of.

5:26

Maybe you heard of Michael Dell. Maybe

5:28

you heard of Toby Lutka. Maybe you heard

5:30

of uh Patrick Collison or or Matthew

5:34

Prince or Brendan Narib or Jason Freed

5:37

or Jack Dorsey. Um a few of the people

5:40

we've signed on for this insane mission

5:43

to actually make Linux on the desktop

5:46

2026 happen and beyond. I mean this is

5:48

also an insane raise to do for this kind

5:51

of mission. Initially, I just asked

5:54

floated the idea with Toby on Sunday for

5:57

this foundation. I like, do you know

5:58

what? I should just I PUT A MILLION

6:00

DOLLARS INTO THIS. And immediately he

6:02

replies, I'll match. And I go like, what

6:04

do you what do you mean? You don't have

6:05

to think about it. Um, you you'll just

6:07

match a million dollars. And then I went

6:09

like, well, maybe there are other people

6:10

who would also just go like, yeah, I'll

6:12

give you a million dollars to go make

6:15

Linux totally awesome. And so, we just

6:18

raised that. We just announced that. And

6:19

then we also announced that we are

6:21

straight away spending a bunch of that

6:24

money on Vaxree becoming the exclusive

6:27

sponsor for him so that he does not have

6:29

to think about any commercial

6:30

development for the next three years

6:33

plus a two-year option. So potentially

6:34

for the next five years [applause] we

6:36

just set with Bax doing hyperland

6:40

development and posting uh waifuss on on

6:43

X. So that and then there's a bunch more

6:47

announcements in the pipeline. I mean,

6:49

um, I'm running this thing as the

6:50

executive, uh, director, but I mean, I'm

6:53

putting my money in because I don't need

6:54

the money. So, all of the money is going

6:56

to go towards developers who are

6:59

integral to the development of Amachi

7:01

and the subsystems and everything else

7:02

that we have to uh promotion of the

7:05

whole thing to the infrastructure and

7:08

this is going to be amazing. So, uh,

7:10

yeah, that's my uh out of breath spiel

7:13

here for just a few crazy things that's

7:14

happened this week. I have to say, wow.

7:16

>> Let me let me just mention one more

7:17

thing to the announcement because this

7:18

is the important thing is here is Linux

7:21

is [ __ ] happening. We are going to

7:22

make this happen. I have not been this

7:24

determined about almost anything else in

7:26

my goddamn life about making Linux

7:29

happen. And this couldn't have happened

7:31

if we were still stuck with the hardware

7:33

we had 3, four years ago. What's making

7:36

it happen now is that we finally have

7:40

MacBook level quality competitive

7:42

hardware. And two reasons for that. One,

7:46

Intel Panthle just absolutely

7:48

mind-blowingly efficient and um

7:52

performant CPU and and Dell revamping

7:56

that XPS lineup and for the first time

7:59

in my career giving me a laptop I would

8:02

pick over any other. I'm not just saying

8:05

that because Michael Dell just gave me a

8:06

million dollars, although I would have

8:07

said it now if I hadn't said it already.

8:09

I've already [laughter] said it. Fair.

8:10

Fair. I've been pinging the XPS for a

8:13

long time because it simply is my uh

8:16

favorite computer. So,

8:18

>> I I feel good about that it's in that

8:20

order. Otherwise, you're like, YOU'RE

8:21

[ __ ] CHILL. But um

8:24

>> now it is my professional duty to talk

8:26

nicely about Dell. So, I'm very happy

8:27

that that lined up with the fact I just

8:29

really like what not only Dell is doing,

8:31

what Intel has been doing. Also, AMD, by

8:33

the way, even though they're a little

8:35

bit behind on efficiency, they're also

8:37

putting out some absolutely cracking

8:39

products. I know I'm uh now have to hawk

8:42

Dell all the time, but um the nice

8:44

people from AMD also just sent me this

8:46

uh AMD 39

8:49

395 plus laptop from HP, which is the

8:52

same chip that's in the framework uh

8:53

desktop. Absolute freaking beast. So, if

8:57

you don't need the uh battery life that

8:58

you get out of that XPS, then that's a

9:01

wonderful option, too. So, all the

9:03

pieces, all the pieces are coming

9:06

together at the same time. Did you know

9:08

we're even getting [ __ ] Fortnite by

9:10

the end of the year? I finally annoyed

9:12

Tim Sweeney.

9:13

>> That's actually crazy. That's actually

9:15

crazy.

9:17

>> After two years of hacking him on

9:19

Twitter like at regular intervals, I

9:22

mean, I'm not going to take credit for

9:23

the whole thing. Maybe Valve also

9:25

deserves a little bit because like

9:26

Steambox and Steam Deck and whatever is

9:28

sending

9:29

>> the millions of dollars go into Proton

9:31

and all that. Yes, they also deserve

9:32

just maybe a touch.

9:34

>> Maybe just a touch. I mean again the

9:36

amazing look at how many different

9:38

factors are coming together and

9:39

conspiring this year 2026 is when it all

9:43

happens

9:45

>> and it's the year of the Linux laptop

9:47

even not even year of the Linux desktop

9:50

>> yes that is actually key because when we

9:52

talk about Linux on the desktop I mean

9:55

>> nobody buys desktops anymore I mean I'm

9:57

talking broadly here if you look at

9:58

Apple sales like it's all MacBooks

10:01

>> the MacBook was really the number one

10:04

target And until you had something that

10:05

was competitive with it, it was going to

10:08

be uphill to sell it, right? Because you

10:10

were going to ask people to do

10:12

compromise. You're going to either tell

10:13

them that, well, the battery life isn't

10:15

great, but you should use Linux anyway.

10:17

Which, by the way, great argument. I've

10:18

been making that argument for two years.

10:20

I totally think uh Linux was worth the

10:23

drop in battery life, but now you don't

10:24

even have to make that argument. Did you

10:26

see the Dell XPS or 14 with the LCD

10:29

display went 42 hours in a battery life

10:33

test? I'm like, "What do you mean 42

10:35

hours? It's just absolutely nuts." Um,

10:37

>> yeah. I to to throw this on there, I

10:39

have the slightly older Dell XPS, the

10:41

one that we got at Omacon. And uh, that

10:44

one I've been able to I went on a flight

10:47

to California to Long Beach just a

10:50

little bit ago. And during that flight

10:52

from Rapid City to Denver, I programmed

10:54

and did stuff. And then from uh, Denver

10:56

to Long Beach, I programmed and did

10:58

stuff. I shut my laptop and then hung

11:01

out on the beach. And then at the end of

11:03

the week, I opened up my laptop. Still

11:04

had half battery left. And this was this

11:06

is just running OMachi. Like there's

11:08

nothing going on. There's nothing fancy

11:10

going on here. This just never has

11:11

happened in my life of So the fun side

11:14

story when I started using Linux on my

11:16

laptop was in 2016. And in 2016, I had

11:20

45 minutes of battery life. I couldn't

11:22

make it through one meeting at Netflix

11:23

every single time. It was this game

11:25

where I could not open it. I had to keep

11:27

it turned off until like midway into the

11:30

meeting so I could open it and then do

11:31

some sort of presentation. I had to come

11:32

up with reasons to stall me presenting

11:34

so it wouldn't die

11:35

>> and then uh fans kick on probably too

11:38

prime and it's LIKE I CAN'T HEAR YOU. I

11:40

CAN'T HEAR THE FAN.

11:43

>> Are you are you are you prompting too?

11:45

Is it still lasting a long time even

11:46

with like running like LLMs?

11:50

>> Oh yeah.

11:52

>> Yeah. Yeah, it's a it's to prog

11:55

typically all my LLM stuff if I'm doing

11:57

anything runs in the cloud. I don't use

11:58

uh I almost never use local LLM stuff

12:01

anymore. But yes, on that flight I

12:03

happen to be like, "Hey,

12:05

>> explain to me stuff about Apollo." And I

12:07

started playing with Apollo and I just

12:09

kept on playing with it a whole bunch.

12:10

So yeah,

12:13

this by the way is also key. We're

12:15

talking about battery life. We're

12:17

talking about all sorts of things coming

12:18

together. The agent thing is really the

12:21

thing that completely rewrite the game

12:24

because on battery it was a game of

12:26

catchup, right? Like we had to get to

12:28

the MacBook level and then now level

12:31

maybe we're a little bit ahead,

12:32

whatever. They still make great chips.

12:34

The M chips are awesome. Um, so it's not

12:37

like, all right, we just trounced them.

12:39

But when it comes to agents working on

12:42

your desktop and turning it into your

12:44

personal favorite computer, Apple's

12:48

nowhere. The Mac OS operating system is

12:51

complete trash when it comes to having

12:54

your dream computer produced by agents

12:58

on your behalf because you can't change

12:59

anything of the core of how the OS works

13:03

because that's Apple's domain. I saw a

13:05

fun tweet the other day where where it

13:07

flagged Codeex app as being malware and

13:09

track Dragon this whole idea that I own

13:13

a computer and for me to get software on

13:15

there in a reasonable way that producer

13:17

first has to go BEG APPLE CAN I PLEASE

13:20

MAKE software and get the sign is just

13:23

mind-blowingly insane. We do not have to

13:26

live like this with agents. We're

13:29

getting access to the malible computer.

13:31

This dream that computers makers and

13:34

programmers have had since the [ __ ]

13:36

dawn of time here. And Mac doesn't have

13:40

an answer. There is no answer. Yeah,

13:42

make some little apps within the

13:44

confines of what Apple allows you to do,

13:47

but who wants to live like that? You

13:49

want to live with freedom.

13:51

>> And can I just jump in here for a

13:53

second? It's not just Apple that's bad.

13:56

It's also just like so much of the

13:59

programs, they don't even have a way for

14:01

you to edit the config other than

14:03

launching the program and click like the

14:05

the culture of developing an app is also

14:08

bad. So, it's not just, you know,

14:09

cultures downstream of the creator, but

14:11

the creator is bad and the culture is

14:13

bad. And so, it's just like both of them

14:15

make it impossible to you. Like, I hate

14:17

setting up an Apple cuz I know I have

14:20

hours of pushing menu buttons. I don't

14:23

want to push menu buttons ever. I want

14:25

to have one conf. I want to download

14:26

JSON and put it right into the system.

14:29

>> This is so crazy.

14:31

>> I had an experiment or experience maybe

14:33

four months ago. I was going to some

14:35

place I thought I had a PC that I was

14:38

going to install a Machi on and like I

14:40

hadn't been there for a while. So, what

14:41

I had instead was a [ __ ] Mac Mini.

14:43

So, I was stuck for three days and I did

14:46

not have a a PC because I had left the

14:48

laptop of home. So, I just had that Mac

14:50

mini and I was like, do you know what?

14:52

>> Let me just give it a go. It can't be

14:54

that bad. And I tried to set it up like

14:56

I would want my ideal computer to do.

14:58

And first of all, it took goddamn

15:00

forever. But exactly as you say, the

15:02

entire operating system is just not

15:03

built for agents. It's not built for

15:06

accessibility of the tooling we want to

15:09

use. Raycast, great example. So,

15:11

Raycast, great launcher. Lots of people

15:13

love it, right? The only way you can get

15:15

your config in and out of that is

15:17

exactly AS YOU SAY. YOU GOT TO GO TO a

15:19

menu, you got to scroll down, there's a

15:21

little export config, and then you get

15:23

this config. And then to get it in

15:25

there, you got to GO BACK AND IMPORT

15:27

CONFIG like with your LITTLE MOUTH LIKE

15:29

A [ __ ] [laughter] caveman. It is

15:32

absolutely mindblowing how bad it is.

15:35

And the only reason it's not like shoot

15:38

yourself bad now is because of homebrew.

15:40

Homero has done amazing things for the

15:42

Mac, but it's still limited in what it

15:45

can do. Like you can't set hot keys.

15:48

There's none of the uh Mac OS hotkey

15:51

edible programmatically. The only way to

15:55

set up um your workspace hotkeys, by the

15:58

way, is going to a menu and manually

16:01

click from row to row like a [ __ ]

16:03

pecking bird. It is just absurd and it

16:07

becomes triply absurd in this age of age

16:10

of agents where they could just do it

16:11

for you. You can't check in your

16:12

conflict. You can't do anything. It is

16:15

actually mind-blowing now having lived

16:16

on Linux for two and a half years and

16:19

have built a mod that like people just

16:21

accept that this is how they live. Now,

16:23

I understand that in the past, even

16:25

quite recent past, you just went like,

16:27

"Okay, yeah, that totally blows, but

16:30

like I get better battery life or it, I

16:33

don't know, build quality or I get

16:34

something else." That's fair, but this

16:36

is why it's a complete game changer.

16:38

Now, now that we've leveled out those

16:40

things, thanks to Intel, thanks to Dell,

16:42

thanks to other makers who finally

16:44

caught up and are competitive on the

16:45

hardware side, we can finally press

16:48

those advantages that Linux just has

16:50

that beats the Mac bloody. And then the

16:54

agents enter the picture and you can

16:55

make anything happen. You can make your

16:56

computers do anything. It is just

16:58

unbelievable that all the stars should

17:01

line up in a perfect constellation for

17:04

this moment right now. And then we

17:06

raised [ __ ] $8 million for the Omic

17:08

Foundation. We're going all in with

17:10

Amachi.

17:12

Oh my god.

17:12

>> And you got And you got Vax. I mean, you

17:14

got Vax going full time like just doing

17:16

what Vax does best. Like that is that's

17:18

some real momentum.

17:20

>> So this is just the deal that got signed

17:21

before the announcement. Oh my god,

17:23

there's [ __ ] coming. You have no idea.

17:26

So much [ __ ] coming. The pipeline, man,

17:28

it is bulging with amazing stuff that's

17:32

going to come out of this.

17:33

>> Yeah. So I have a question. And so the

17:34

$8 million you just got, how is that

17:37

being used? And I don't know if you can

17:38

say this, how is this being used to

17:39

making like Linux the desktop or age of

17:42

Linux being desktop thing? How's that

17:44

where's that money going or what's your

17:46

vision with it?

17:49

>> Three main things. A fund the developers

17:52

who create the key subsystems for Amachi

17:56

people like Backre. So doing deals like

17:58

that such that developers like

18:00

Backstreet don't have to think about

18:01

money. They could just make awesome

18:03

Hyperlink, which by the way, even though

18:04

we're now the exclusive sponsor.

18:06

Hyperlink is still open source, free for

18:09

anyone. Any other DRO can put it in.

18:11

It's not like, oh my god, now it's off

18:12

the market. This isn't proprietary

18:14

software, bro. This is all MIT licensed

18:17

stuff. You go to the Omachi repo. You

18:19

can [ __ ] take anything and build it

18:20

into your Nick Stream or something if

18:23

that's what floats your boat. So, we're

18:25

gonna take a bunch of money. We're gonna

18:26

put it towards that. That's the bulk of

18:28

it. The bulk of it is going to go

18:29

towards funding of that. Now, I mean, I

18:32

raised $8 million like [ __ ] just in

18:35

the last few days, just what? Um, I'm

18:38

gonna make it last, right? So, it's not

18:40

like uh by by next Tuesday, we've blown

18:42

it all on hookers and coke here, right?

18:44

Like, it's got to last in my head. We're

18:47

spending this money over the next five

18:49

years. Like, that's I got a communist

18:51

five-year plan here going. So, there's

18:53

five years and we're going to spend

18:56

>> forwarding Michael Scott right now.

18:59

>> Dude, don't go there. Too soon. Too

19:01

soon.

19:02

>> Too soon for a great leap forward or

19:04

Michael Scott. Which one are we talking

19:05

about? [laughter]

19:08

>> Too soon by Michael Scott. Um

19:10

>> yeah, too soon.

19:11

>> So So it's going to go towards that.

19:13

Then it's going to go through all the

19:14

infrastructure required to build a top

19:17

tier Linux distribution.

19:19

uh package servers, mirrors, agent

19:23

runners, um QA servers, all the

19:27

infrastructure it takes to operate a top

19:29

tier Linux distribution,

19:31

>> we're going to build it. We're going to

19:32

run it ourselves and we're going to take

19:34

over more and more and more of it until

19:36

at some point, we'll see how far it

19:38

goes. I mean, my delusions of grandeur

19:39

grow into the sky. So, um getting to

19:42

eventually core levels of Linux

19:44

distribution, that's the that's the

19:46

plan. And then finally, we're going to

19:48

do something that I know nerds are

19:49

goddamn allergic to. We're going to

19:52

market. We're going to tell people that

19:54

Umachi is the greatest

19:57

>> ever to

19:58

>> let's go

19:59

>> encounter their computer. We're going to

20:01

spend money on um merchandise. We're

20:04

going to spend money on on events. We're

20:07

going to spend money on

20:09

activities that make people aware that

20:12

Linux is now fantastic and that the best

20:16

version of Linux for the age of agents

20:20

is Amachi. So part of this is also like

20:23

both it's both extremely ambitious in

20:26

what we're trying to do yet also kind of

20:27

narrow right because I'm not giving up

20:30

on the keyboard. We're going to be

20:32

keyboard first. I'm not going to go

20:33

like, "Oh, what does it take to get

20:35

grandma to use this machine? Can we make

20:37

a watered down version of uh Ubuntu so

20:40

that we too can distribute a Teemo

20:42

version of Windows?" [ __ ] that.

20:44

>> No, no, no.

20:45

>> The mission remains.

20:47

>> I'm building to be the perfect computer

20:49

for me and then I'm giving you all the

20:51

tools so that you can start from a

20:54

system that's already amazing and then

20:56

make the perfect computer for you.

20:58

Right. So

21:01

um what was the question? Oh, how are we

21:03

going to spend the money? Um those are

21:06

the three main pillars, right? Fund the

21:07

developers, uh fund the infrastructure,

21:10

fund the excitement.

21:12

>> Can I Okay, I have a question. A very

21:14

important question, especially in in our

21:16

day and age, the I'd say the biggest

21:17

risk factor to any of these type of

21:19

projects right now is supply chain

21:21

attacks. I think we see them non-stop.

21:23

Rust is under a current one right now

21:25

with array ref, I think it was or

21:27

whatever it was. Every single one of

21:28

these is effectively a worm ready to be

21:32

wormed on everybody's machine and being

21:34

a part of a distribution. You download

21:37

stuff and it really does get executed

21:39

and people really will take all of your

21:41

stuff and spread even more worming. So

21:43

with that in mind, what is Omachi's plan

21:46

to be different? Because a AUR it's like

21:49

I I've been scared to update my system

21:52

lately because I know AUR has been also

21:53

having a bunch of problems. And so it's

21:54

like, how am I going to have a system

21:57

that I can trust? And so, how does

21:59

Omachi make a universe that I can trust?

22:03

I don't want one more npm. What's the

22:06

answer? [laughter]

22:08

>> I'll tell you, it's real simple. We're

22:09

going to own the [ __ ] universe. The

22:11

entire Umachi installation and

22:14

distribution is going to be sovereign.

22:16

Now, we're leaning at the moment on the

22:19

AUR for a handful of optional

22:22

installations. There's nothing on Amachi

22:23

itself that comes from the AUR.

22:25

Everything is compiled on our built

22:27

machines and our built servers and then

22:28

go into our package repository. Like I

22:31

totally get only been around for a year.

22:33

So if you are basing your current

22:35

opinion on what Amachi is on that first

22:37

version that shipped as a git checkout

22:40

in your home directory and ran straight

22:43

off the AUR, you are about 17 light

22:47

years out of date with where we are

22:49

today. We run our own build service. We

22:51

run our own package repository. We run

22:53

our own mirrors. All of it lives behind

22:56

Cloudflare. This is why we don't have to

22:57

do vintage stuff like torrent and

23:00

mirrors and [ __ ] We just use modern

23:02

tools like CDN's to ensure that I've

23:06

heard of those to download anywhere.

23:08

>> Heard of those.

23:09

>> Um, we're going to get rid of the AUR

23:12

entirely for anything um presents for

23:15

you as an option. I mean, we're almost

23:17

already there. There's a handful of

23:19

things left and they're almost all from

23:21

trusted vendors. It was like, oh, you

23:22

can install Brave the browser. We get

23:24

that from the AU. We don't have to.

23:26

Everything that I've been packaging

23:28

recently like T3 code, right? I just

23:30

packaged that for Theo yesterday. And I

23:35

didn't go to the AUR. I did the original

23:37

package. Like I'll go to where T3 is

23:40

distributed. I'll get it from that. I

23:42

will package it up. I will host it. I

23:44

will control the update just so that I

23:45

can ensure that it's happening in a

23:47

secure way. So the AUR is going to

23:49

basically be reduced to an optional

23:52

thing that people can use if they choose

23:54

to, but we're going to keep

23:56

deemphasizing it because we're going to

23:57

keep getting more and more and more

23:59

coverage so that there's less and less

24:01

reason for you to get to the AUR. Now, I

24:04

actually think the AUR is amazing. I'm

24:06

not saying this to slack on the AUR in

24:08

any way, shape, or form. I do think that

24:10

it is from a vintage of the internet

24:12

where we were more hopeful and happy

24:14

people who can trust each other slightly

24:15

more. But there's 117,000 packages in

24:18

the AUR. We're not hosting that

24:20

tomorrow, right? But as with everything,

24:22

it's a power law. If you look at the top

24:25

whatever 2% of packages, 1% of packages

24:28

on the AUR is probably 80% of the

24:30

downloads, right? So we just have to

24:31

cherrypick off that 1% and then we will

24:34

shrink the scope that is available for

24:37

these kinds of attacks and then we will

24:38

build out our own repository. We'll

24:40

build our own build server. We'll build

24:42

out everything. I mean we're going to

24:44

run our [ __ ] fork at the kernel for

24:46

the things that we need. We already do.

24:48

So the reason in part that your Dell XPS

24:52

laptop runs so well on Amachi is that we

24:54

run a forked version of the kernel that

24:56

has patches for that. This is what

24:58

Spencer Bull, who's by the way on the

25:00

newlyannounced UMachi core team. He's a

25:03

uh

25:05

>> um he's he's a distinguished engineer I

25:07

believe it is at Dell who's been on the

25:10

Umachi train almost since day one. He he

25:12

spoke at Omicon all in amazing research.

25:15

He's been doing the liaison for Intel

25:19

patches that we needed for the XPS to

25:21

run at its absolute peak efficiency to

25:23

go into our fork at the kernel. We're

25:25

going to do that for the whole thing.

25:27

I'm I'm talking to everyone right now,

25:29

including Linux kernel developers to

25:31

come help us out to secure things. Um,

25:34

one of the next steps we got to do is

25:36

this amazing plug-in repository, right?

25:38

Like that's the same problem as the AUR.

25:40

If you're downloading code from the

25:41

internet, you run it, it's code from the

25:43

internet, it could do stuff to your

25:44

computer. We're going to fix it. We're

25:46

going to fix everything. Um, the plugins

25:49

are going to get hosted on our setup.

25:51

They're going to run through our

25:52

agentpowered security vetting, which by

25:55

the way, all of this security drama

25:58

right now, a lot of it is agent

26:00

accelerated, right? Like people are

26:01

finding exploits with agents. Yeah. Do

26:03

you know what that also means? These

26:04

agents are incredible defenders.

26:06

>> And that defense really goes in in in

26:10

depth. So if we arm our side with the

26:12

best, smartest, latest agents and we

26:15

basically have unlimited token budgets.

26:18

Um note on that announcement coming

26:20

later for for endless tokens for

26:21

Amarachi. Then we will be able to apply

26:24

all the best intelligence from this new

26:26

age towards securing Omachi to the

26:29

maximum degree possible. Now we're going

26:31

to be liable to all the same things

26:33

everyone is liable to. We're going to be

26:34

liable to the throat they exploit in the

26:37

colonel and whatever. But but because we

26:40

are a rolling dro off Arch with a one

26:45

month lag so we don't screw people's

26:48

installations in the middle of the night

26:49

on a [ __ ] Tuesday. That's how we

26:51

started. I woke up with Ryan Hughes at

26:54

3:00 a.m. more than one because some

26:56

package got pushed out and people

26:58

updated their installation and bam, it

27:00

was broken. We ended all that nonsense

27:02

by just being one month behind on

27:04

default. But then as soon as there's

27:06

anything a CVE or whatever, we can just

27:08

update right there. We can ship the

27:10

latest version, the latest patches, the

27:12

latest everything literally next moment.

27:15

We can have agent

27:16

>> monitoring all of it, monitoring the

27:18

CVE, opening the PRs, pinging people. We

27:22

can afford to have uh on call staff. We

27:25

[ __ ] raised $8 million. There's

27:27

nothing we can do.

27:29

>> You can't um you can't buy cursor with

27:33

that kind of money. Uh, so you can't do

27:34

that. Just throwing that out there. It's

27:36

gonna take a little bit more than eight

27:37

million, but eight million is insane.

27:39

>> $20 a month. Brian,

27:40

>> I I do I [laughter] think it's very

27:42

cheap comparatively. But

27:44

>> I do think there's something to be said

27:46

about this this insane world that we

27:48

live in. Um, I forgot who I think it was

27:51

Sam Lambert of Planet Scale said like

27:53

they run stuff with their databases and

27:55

they have all this uptime and all this

27:57

performance and they're a team of 30

27:59

engineers or some small sing double

28:01

digit engineers. But yeah,

28:03

>> and I forget the exact number, but

28:06

Amazon and their one database has a

28:08

thousand effective people working on

28:10

this their database offering. And you

28:12

just have to think the same thing with

28:14

Windows or with Mac like how many

28:17

thousands of engineers, product

28:18

developers, all these people

28:22

putting all this effort into an

28:24

operating system and then you have Vax

28:26

just like one dude who's making an

28:28

experience that's just dramatically

28:30

better. Like how does this even h like

28:32

how are we even in this world in which

28:34

one person is defeating a thousand? This

28:36

can't like something has to be

28:38

fundamentally broken and it's not just

28:40

the power of AI because he's been doing

28:42

this before the power of AI came out. So

28:44

it's something it's something more than

28:46

just that. And so that's why I'm very

28:47

excited about Omachi because I want you

28:49

guys to somehow capture the eight people

28:52

who can defeat a trillion dollar

28:54

organization and that's all you need.

28:56

We're doing it. We are absolutely doing

28:58

it. And it's exactly as you say, all the

29:00

great leaps forward. I'm not quoting

29:03

[ __ ] Mau here.

29:04

>> All the [laughter] like, whoa, we're

29:06

back there again already. DHH, that was

29:08

fast.

29:09

>> Hold on. Hold on.

29:10

>> Independence. [laughter]

29:11

>> All the great leaps forward. They did

29:13

not come out of a goddamn committee.

29:15

They did not come out of a thousand

29:17

people replicating the Orc chart. They

29:20

came out of cracked Polish Linux kids

29:23

like Backstreet who in their goddamn

29:25

bedroom at I don't know when did they

29:27

start Hyperland? Three years ago. So he

29:28

was uh 19, 18. Just go like, "Hey, do

29:32

you know what? I [ __ ] know better."

29:33

Like these Mac people who've been

29:35

working on Mac OS and have $3 trillion

29:38

in market cap behind him. I can beat

29:40

them. And then he [ __ ] does, right?

29:42

like world is moved forward by

29:44

delusional unreasonable people who with

29:48

enough determination become right.

29:51

>> That's the group I'm tapping into. Um

29:55

that's what we're doing.

29:56

>> Can I ask another real question?

29:58

>> Of course. Trash.

29:59

>> So

30:01

>> you guys have been prompting this new

30:02

version Quattro. Like what's the like

30:05

what's the QA process here? Like are

30:08

obviously you're dog feeding your own

30:09

OS, but do you have like a big group of

30:12

like alpha users? Do you have a QA

30:14

department? Like what what's that look

30:15

like? Like how can we trust you that

30:17

this is actually not slop, right?

30:19

>> Yeah. So the first step is you have a

30:21

goddamn obsessed maniac, me, who has uh

30:24

at this point, you mean 17 laptops and

30:26

400 PCs connected to everything. Um, I

30:31

have installed Omachi Quattro and I kid

30:33

you not, probably about 500 times, maybe

30:36

more because just running through the

30:39

installation, making sure everything is

30:40

tight, making sure it all works on all

30:42

these different permutations. That's

30:44

just me. Like that's the first layer of

30:46

defense here. I am not going to ship

30:48

anything unless it is goddamn delicious

30:50

and awesome and works. Then second tier

30:53

of uh defense here is we have an

30:55

incredible community already. I think we

30:57

got like 40,000 people on the Discord

30:59

now or something like that. Um, we have

31:01

multiple channels. So, we have the dev

31:04

channel, that's people running straight

31:06

off Git. Um, praise be those brave

31:08

souls. Then we have the um edge channel

31:12

which gets the latest packages and that

31:14

system has gotten really nice. One of

31:16

the key by the way, just quick aside,

31:18

when I started Amachi, it was a a get

31:20

checkout in your home directory, right?

31:23

That works great for me. a little

31:25

brittle if the user starts meddling and

31:28

why wouldn't they? It's in their goddamn

31:30

home directory. Right? So, we had

31:31

problems at time with overwriting things

31:33

and we were sharing configs and it was a

31:36

little delicate. Let's just say that.

31:38

Right now, we've switched to a

31:40

completely packaged driven system. Um is

31:43

a system package that installed straight

31:45

into all the other system packages and

31:47

therefore its own configuration files

31:49

and so on. They're not in user space.

31:51

We're giving you access to can extend it

31:55

but we're not stepping on each other's

31:56

toes anymore that erases like 70% of the

32:00

bug reports that we had for former

32:01

versions of quattro just or for just

32:04

separating those two things but then we

32:05

have the channel system so we have the

32:07

edge which is sort of people want to

32:09

live on the edge they get the latest

32:11

packages and they get I think we're

32:14

pushing edge packages every six hours by

32:16

default sometimes I push it more often

32:18

when I'm testing things but otherwise

32:19

you get whatever git is six hours behind

32:22

it's great if you want to run on the

32:23

latest. There was actually a little

32:25

issue uh this morning on some of this. I

32:27

fixed that. That's what Edge is for.

32:29

Then we have RC. So when we get closer

32:32

to a new release, I go through a release

32:34

candidate process where we push it out

32:36

to RC and like that has a whole

32:38

dedicated uh group on Discord like

32:40

omachi- release candidate. And we have a

32:42

large number of people really stress it

32:44

out and can test it on like a million

32:46

other machines than what I have. Like

32:48

the 17 I have does not cover the range

32:50

of uh Linux computers available. And

32:52

then when all that is great, then we

32:55

have stable. And by the way, a reminder,

32:57

stable is by default one month lagging

33:00

the latest arch packages. And we do that

33:03

just to like you don't get that limoni

33:05

update in the middle of the night

33:06

because someone just pushed something

33:08

out. And we got bitten by that in the

33:09

early days of uh March as well. So we

33:12

really just like speedrun this process

33:15

of what does it take to build a

33:17

resilient yet up-to-date system. It's

33:20

very easy to build a resilient system if

33:22

you use packages that are four years

33:24

old. That's auntu. So if you want that

33:27

that's available great people have

33:29

solved that problem for people want

33:31

that. That's not what Amachi is. I want

33:33

the [ __ ] latest version of Neoim

33:35

right when it drops. I want the latest

33:37

version of uh codeex. I want the latest

33:40

version of my key developing tools.

33:43

Especially now when like the AIA are

33:45

updating 17 times a day. I mean I think

33:47

Dak bragged that it literally was 17

33:50

times on a given day that open CODE

33:52

UPDATED RIGHT I WANT THAT RIGHT NOW

33:54

WHICH by the way also super cool so we

33:58

have all this set up for all these

33:59

packages right like that's using Pac-Man

34:01

we have our own mirrors we have our own

34:02

package repository but it's still

34:03

Pac-Man that does it and like you don't

34:05

want to churn through that all the time

34:07

so you have your system packages that's

34:09

itself and all the other core util like

34:11

uh QT and the core library hyperland

34:13

itself and quick draw lives over here in

34:16

system packages

34:17

Then you have your tools like open code

34:20

that change 17 times a day. It can't

34:22

live over here. If you shove open code

34:24

into your repository that also keeps

34:26

your system packages, you now have to

34:28

update all your system packages when you

34:29

update open code. No,

34:31

>> no. Brittle. No good. So we have a

34:34

two-tier system that runs at separate

34:36

speed. Pac-Man is running system

34:38

packages at a reasonable pace. You got

34:41

Edge, you got RC, you got stable. Okay.

34:44

Then we have Mi and Mi is goddamn

34:48

amazing. Mi allows you to install any of

34:51

the tooling you need for Ruby, Go,

34:54

Python,

34:56

all the main development environments,

34:58

run them in different versions for

35:00

different projects and then it allows us

35:02

to install all the high churn developer

35:04

tools like Codex, like plot, like open

35:07

code, like um the H CLI, the basecam

35:11

CLI. All of these things can run on an

35:13

update chain that is super fast and low

35:16

criticality and low uh what impact area,

35:19

right? Something goes wrong on mis with

35:21

it like all right, so Codex has an

35:23

issue, they'll pick out or push out

35:24

something quickly. It doesn't [ __ ] up

35:26

your system. And then finally, I mean,

35:29

there's a lot here. We're really in

35:30

depth with it. Finally, we have a

35:33

snapshot system. Whenever you go to

35:35

update Omachi, we take a snapshot. We're

35:37

using BTRFS as our default file system

35:40

so we can make cheap snapshots. We make

35:42

a snapshot before you update. Then when

35:44

you update, let's just say something

35:45

crazy happened. You were unstable. We

35:48

blew our minds and it messed up, right?

35:50

Your computer's [ __ ] Your Arch is now

35:52

broken. Nothing works. What do you do?

35:55

No biggie. You just reboot. You get a

35:57

list of your owner, your earliest

35:59

snapshots. You pick the one from one

36:01

back. You go into that and you live

36:03

happily ever after. It is at this point

36:05

quite difficult to brick a new

36:08

installation. And let's just say

36:10

>> that's really nice.

36:11

>> You manage. YOU MANAGE TO [ __ ] BRICK

36:13

IT. You send your agents a muck and they

36:16

just rip out everything and they rewrite

36:18

all your

36:19

>> be me.

36:21

>> Totally screwed, right? Totally goddamn

36:23

screwed. Here's chef's kiss at the

36:26

bottom layer of all these snapshots that

36:28

happen intermediately. We keep five and

36:30

so on. You just [ __ ] it all up, right?

36:33

We we have factory reset now. Factory

36:35

goddamn reset.

36:37

>> You can factory reset.

36:38

>> That's a revolutionary idea.

36:40

>> Entire goddamn thing

36:42

>> to the moment you installed it. And I

36:43

know it worked when you installed it

36:44

because that was off an ISO that I bet

36:46

it right. So you bring it back to ISO

36:49

level. Not only will that now restore

36:52

your computer to its original install

36:53

set, you can [ __ ] gift it. We have

36:56

full like Mac levels of like I want to

36:59

hand off my computer thing. Um, I've not

37:02

on a lot of dros. I mean, I'm sure

37:03

someone has done it, right? Like that's

37:04

not a standard feature that just works

37:06

well out of the box with encryption

37:08

where you can trust that your data is

37:09

not getting handed over to whoever you

37:11

found on [ __ ] Craigslist who's going

37:13

to take over your machine, right? You

37:15

want that stuff to feel secure and we

37:17

now have factory reset built in. So,

37:20

there's about 17 layers of that

37:22

resilience stack and um we're going to

37:25

keep adding more to it until it's so

37:26

goddamn bulletproof that you need to be

37:28

a terrorist to blow it up. [laughter]

37:32

>> I'll say I've used uh these before and

37:35

it's and it's really nice. Like it's a

37:36

tool that actually we're using in a

37:38

bunch of repos before and uh I like it a

37:41

lot for Yeah.

37:42

>> Quick call out. JDX who runs uh

37:46

>> we're sponsoring him too.

37:49

We're doing all of it.

37:50

>> Great.

37:50

>> We're doing all of it. Everything that

37:52

sort of fits into this and and what I

37:54

mean

37:54

>> Linux is I think on my installation I

37:56

think a,087 packages. We're not doing

37:59

a,08 yet I should say. Right. By the

38:02

time my budget is 800 million, we're

38:03

doing [ __ ] everyone, right? But I'm

38:05

starting with eight.

38:06

>> So with eight, we gota we got TO MAKE IT

38:08

LAST a little. We can't do everything.

38:11

We're going to do the key components and

38:13

uh it's going to be great.

38:15

I find me to be like a nice middle

38:18

ground that isn't it isn't quite like I

38:21

feel like I have to have nyx and then I

38:22

spend a lot of time like arguing with

38:24

people on the internet about which

38:26

companies are allowed to use my software

38:28

but instead I just like I just have my

38:31

team all use the same version of

38:32

whatever executables we have and then

38:34

it's like pretty good for most things

38:36

you know.

38:36

>> Can can I say something?

38:38

>> Yeah, of course. Uh,

38:40

>> this is your show to say things here.

38:43

[laughter]

38:44

>> Nyx is actually part of the reason why

38:48

we're doing an institution. That is not

38:50

just like a one-off thing here. I'm

38:52

sorry to say, but Linux is [ __ ]

38:55

infiltrated with goddamn maniacs and

39:00

just clowns. just clowns who want to

39:05

control what you think and who you can

39:07

associate with and whose opinions are

39:09

allowed to use what software. It's

39:12

goddamn insufferable and Nyx is perhaps

39:14

worst of all. Now, there's amazing

39:17

technology in Nyx. Amazing technology,

39:19

right? I figured some [ __ ] out here. But

39:22

the goddamn cancer that is around both

39:26

Nyx and so many of the other

39:28

institutions in the Linux world is one

39:31

of the reasons where like we need a

39:32

counterweight. We need a place over here

39:34

that's not just a full of [ __ ]

39:37

goblins who are crazy. We need just some

39:40

people who are like, do you know what?

39:41

We make fun computers. Do you like

39:44

computers? Then come have FUN WITH US. I

39:46

DON'T GIVE A FLYING [ __ ] what you think

39:48

about other topics or what your politics

39:51

are or I don't know whatever you

39:53

support. You're welcome. You're welcome.

39:56

You can come over here if you just want

39:57

to have fun with computers and find a

40:00

competent group of people who are

40:01

responsible for distribution. We're here

40:03

for you and you are invited. I don't we

40:07

got to get away from this nonsense that

40:09

like all this [ __ ] is teamsbased and

40:11

it's teamsbased along these fault lines

40:13

of culture war [ __ ] Microsoft

40:15

Teams.

40:16

>> Microsoft Teams every time. [laughter]

40:20

>> I'm not doing goddamn land

40:22

acknowledgements here before I talk.

40:24

We're not doing code of conduct [ __ ]

40:27

persecution [ __ ] We're just running a

40:31

Linux distribution for people who want

40:33

to have fun with computers and share

40:35

that hobby with others. And that's a

40:37

really broad tent. I think there's a ton

40:39

of people who would look to run Linux if

40:42

they felt like there was a home for them

40:44

that wasn't full with crazy clowns.

40:49

>> My wife would like Linux. Uh she doesn't

40:51

even know that there's people arguing

40:53

about this on the internet. So

40:55

[laughter]

40:55

>> bless her soul. Bless her soul. Please

40:58

keep her in uh blissful ignorance on

41:01

that point because I'm not actually

41:02

think

41:04

it is a serious problem not just for

41:06

normies so to speak but also for

41:08

developers. There are plenty of people

41:10

who look at the madness

41:13

>> of what's going on in some of these

41:14

Linux distribution and just go like I

41:16

guess Microsoft and Windows is not so

41:18

bad. And that's that's really a missed

41:21

opportunity here. We can do better. We

41:24

can just collect a group of same people

41:26

who want to make good [ __ ] and then let

41:28

everyone use it regardless of what

41:29

they're going to use it for. Um, I hope

41:33

you use it for nice thing, but like if

41:35

I'm selling forks here, I'm not going to

41:37

try to peep in your bedroom window and

41:40

see if you're going to stab yourself in

41:41

the leg with it. And [laughter] I think

41:43

that that's just not like we've ended

41:45

the exchange here. Hey guys, if you like

41:46

this episode, you can watch the rest of

41:48

it on the Spotify. And don't forget to

41:51

like and subscribe. Woo!

41:53

See you later.

41:54

>> Boot up today.

41:57

Voting [music]

41:59

errors [singing] on my screen.

42:02

Terminal coffee [laughter] and

42:06

living the dream.

Interactive Summary

The video features a discussion about Omachi Quattro, a major rewrite of the Omachi Linux distribution. The team highlights improvements in speed, the integration of AI agents to simplify setup and configuration, and the establishment of the Omicron Foundation with $8 million in funding to support development, including key contributors like Vaxry (Hyprland). They emphasize a vision of making Linux more user-friendly, stable, and sovereign, while also discussing the hardware landscape, the importance of avoiding dependency on traditional, potentially insecure distribution methods like the AUR, and maintaining a welcoming, non-political community culture.

Suggested questions

3 ready-made prompts