WEBVTT

00:00:00.820 --> 00:00:01.860
Can you hear me?

00:00:02.300 --> 00:00:02.820
Yeah.

00:00:03.820 --> 00:00:07.880
Hey, how are you, man? You had a whole trip to
Kuala Lumpur, was it?

00:00:04.940 --> 00:00:05.700
Yeah, good.

00:00:07.900 --> 00:00:10.040
Yeah, Kuala Lumpur was good.

00:00:08.240 --> 00:00:08.620
Back?

00:00:10.400 --> 00:00:10.880
It was really good.

00:00:12.260 --> 00:00:12.640
Right.

00:00:12.840 --> 00:00:18.060
There was haze, but it cleared. But it was just
nice to be outside of England.

00:00:19.160 --> 00:00:23.700
You know, some geo... What?

00:00:19.560 --> 00:00:22.260
You were there for the one-on-ones with MK,
right?

00:00:24.660 --> 00:00:30.820
so there's this street basket guy on Instagram
called MK and he plays

00:00:30.820 --> 00:00:36.840
in America and South America and he was on a
tour in Asia and he posts Instagram

00:00:36.840 --> 00:00:42.720
videos of one-on-one basketball where they are
kind of you know punking on each other and

00:00:42.720 --> 00:00:49.100
being super I don't know disrespect yeah very
aggressive and pushing the boundaries and

00:00:44.140 --> 00:00:45.380
aggressive I

00:00:49.100 --> 00:00:50.560
trying to tilt your opponent.

00:00:52.040 --> 00:00:58.100
There was a whole drama happening in Malaysia
when he was there because some of the local

00:00:58.100 --> 00:01:02.400
people felt that he was too disrespectful
against some of the players.

00:01:02.720 --> 00:01:05.720
Some other guy showed up and there were fights
and stuff.

00:01:06.240 --> 00:01:09.940
You were there for the MK. That was last week or
was it two weeks ago?

00:01:09.920 --> 00:01:13.240
Oh yeah, I was there for them. I mean, I have no
interest in basketball.

00:01:14.020 --> 00:01:18.020
I was interested in the food, just catching up
with people.

00:01:14.100 --> 00:01:14.660
Yeah.

00:01:18.720 --> 00:01:21.700
And I observed a couple of things that I found
interesting.

00:01:22.740 --> 00:01:28.800
I was in Malaysia like two years ago, and then
coming this time, I noticed electric

00:01:28.800 --> 00:01:34.220
vehicles. like two years ago I don't remember
any electric vehicles and now two years later

00:01:34.220 --> 00:01:40.560
there's this a ton of them on the road from
brands that I've never heard of like Zeekr

00:01:40.560 --> 00:01:46.700
and Leapmotor and Great Wall and all sorts of
crazy brands I've never heard of.

00:01:41.000 --> 00:01:42.400
BYD

00:01:47.000 --> 00:01:49.360
Do you know BYD? It's Chinese, I think.

00:01:47.380 --> 00:01:53.840
I guess it's the same. Oh yeah there's plenty of
BYD but we have BYD here in

00:01:53.320 --> 00:01:53.740
Okay.

00:01:53.840 --> 00:01:59.200
the UK. We have J.Crew here in the UK. So I
wasn't sort of counting things I knew, I was

00:01:59.200 --> 00:02:05.160
counting things like I've never seen before. And
there's a lot of EV change.

00:02:05.460 --> 00:02:06.860
It must be the same in Vietnam, right?

00:02:08.280 --> 00:02:14.700
Quite a lot of BYD. And of course, we have
VinFast, which is a famous large car
manufacturer

00:02:14.700 --> 00:02:15.480
in Vietnam

00:02:16.500 --> 00:02:22.320
and in Vietnam, the Vingroup, actually Reddit
and Facebook and all there.

00:02:22.660 --> 00:02:24.700
Vingroup, they have a lot of properties.

00:02:25.480 --> 00:02:30.480
It's a big family or it's a big chain of
businesses.

00:02:30.940 --> 00:02:32.440
Alright. There's

00:02:32.060 --> 00:02:39.320
Vingroup also expanded to America. And there's a
lot of chatter about how they

00:02:35.220 --> 00:02:37.220
Oh yeah. ...chatter

00:02:39.320 --> 00:02:44.600
look very successful on paper, but it's actually
like, you know, cousins billing each other

00:02:44.600 --> 00:02:47.560
to, you know, prop up the revenue and things
like that.

00:02:45.160 --> 00:02:48.360
Yeah. ...buffter. There's some huge Vietnamese
companies.

00:02:48.540 --> 00:02:53.240
The IT company always took my eye because it
advertises in the Premier League, is it FTP or

00:02:53.240 --> 00:02:53.860
something like that?

00:02:54.180 --> 00:03:00.220
FPT. Yeah, FPT is big. I think in terms of, so I
used to organize cloud native

00:03:00.220 --> 00:03:06.160
meetups in Ho Chi Minh City, and that was
focused on Kubernetes, you know, back then,

00:03:06.340 --> 00:03:09.600
and then later, I mean, of course, Grafana and
observability.

00:03:10.820 --> 00:03:14.360
that I did in Singapore. And then when I came
back after COVID,

00:03:15.420 --> 00:03:18.940
I created, there was no Ho Chi Minh City
chapter, so I created one.

00:03:19.240 --> 00:03:21.580
And initially there were like 80 people showed
up.

00:03:21.640 --> 00:03:25.000
It was very fun, you know, and we started light
with Kubernetes.

00:03:24.700 --> 00:03:27.100
80? 80? 80?

00:03:25.700 --> 00:03:31.740
At 8-0, yeah, the first meetup was like, you
know, very interesting.

00:03:28.580 --> 00:03:31.080
very

00:03:33.000 --> 00:03:39.360
We had a couple of people that were very
interested to promote this and host

00:03:34.780 --> 00:03:36.180
were

00:03:39.360 --> 00:03:43.180
in their office, which is a digital bank in
Vietnam.

00:03:43.760 --> 00:03:48.460
And they hosted, like, you know, I had only
choice between two offices.

00:03:48.740 --> 00:03:53.420
So it's either Zylk of Vietnam. I think they
helped organize a little bit.

00:03:53.540 --> 00:04:00.300
And then there was Cake Bank. But then FPT and
all, like, they ran the OpenInfra.

00:03:55.780 --> 00:03:58.940
When they were heard of this campaign, they were

00:04:00.620 --> 00:04:08.360
Like, I don't know if you know OpenInfra, but
it's like an AWS for

00:04:08.360 --> 00:04:13.300
your data center type of solution,
openinfra.org.

00:04:11.060 --> 00:04:13.960
Like a Vietnamese AWS or something.

00:04:14.340 --> 00:04:17.960
No, no, OpenInfra Foundation is a large...

00:04:19.260 --> 00:04:25.700
I guess it's like if you want to run EC2 on
VMware or something like that, it's

00:04:25.700 --> 00:04:28.440
basically your cloud. It's OpenStack.

00:04:28.580 --> 00:04:30.480
Sorry, that's the one I was looking for. You
know OpenStack?

00:04:29.640 --> 00:04:31.640
oh yeah I'm just like yeah I've heard of that

00:04:32.220 --> 00:04:37.640
yeah so so OpenStack which is like you know
because we talk about FPT so there was this

00:04:36.420 --> 00:04:37.520
yeah yeah

00:04:37.640 --> 00:04:43.120
OpenInfra Foundation and there's this OpenStack
that was very popular and when I did cloud

00:04:43.120 --> 00:04:48.560
native they were quite aggressive initially in
approaching like uh you know it should be open

00:04:48.560 --> 00:04:54.340
infra why are you doing cloud native and then
after one year um they were there on the first

00:04:54.340 --> 00:05:00.300
meetups but after one year it became like I
tried we had a chat where we were organizing

00:05:00.300 --> 00:05:04.220
together and every time I asked like is there
anyone anyone want to speak like there's a

00:05:04.220 --> 00:05:05.960
Kubernetes 10-year anniversary.

00:05:06.340 --> 00:05:12.200
Well, long story short, I felt that the
OpenInfra Foundation was a bit hostile, and they
were

00:05:12.200 --> 00:05:15.100
mainly backed by FPT and other local data
centers.

00:05:17.780 --> 00:05:21.080
yeah, it just made me step out of the whole
thing.

00:05:18.040 --> 00:05:18.560
Well,

00:05:19.820 --> 00:05:24.360
that whole, oh my gosh, what were you saying
again?

00:05:24.580 --> 00:05:30.940
That Red Hat-driven, what was that initiative
called for the open infrastructure

00:05:30.940 --> 00:05:35.060
again? They had their own operating system and
everything.

00:05:33.480 --> 00:05:34.880
Red Hat opening track?

00:05:37.080 --> 00:05:39.220
Oh, a Flatcar? No.

00:05:39.120 --> 00:05:41.720
No, no. You mentioned it earlier.

00:05:40.700 --> 00:05:41.500
OpenShift, no.

00:05:43.000 --> 00:05:45.300
Governments and IBM are backing it.

00:05:45.660 --> 00:05:46.820
OpenStack.

00:05:47.100 --> 00:05:53.440
Yeah, OpenStack. I mean, there was definitely an
OpenStack cabal.

00:05:49.340 --> 00:05:50.180
Yeah.

00:05:54.360 --> 00:05:57.540
I don't even know if it's still going. I think
it's given up, right?

00:05:57.660 --> 00:05:59.580
It's like just Kubernetes nowadays, right?

00:06:00.080 --> 00:06:07.420
Yeah, so they kind of pivoted. They felt like
that was their playground.

00:06:07.420 --> 00:06:12.540
and some of them I mean some of them from the
north they are quite

00:06:15.580 --> 00:06:19.360
I don't know how to say but they're a bit more I
don't know

00:06:21.680 --> 00:06:23.220
against foreigners I would say

00:06:24.620 --> 00:06:29.160
I really felt that that was the case that's
definitely how they made me feel as I was trying

00:06:28.300 --> 00:06:31.320
Well, it's very typical in Asia.

00:06:29.160 --> 00:06:32.220
to organize events so I said I don't need this

00:06:31.620 --> 00:06:34.320
Actually, that's my first memory of Asia in a
way.

00:06:34.520 --> 00:06:38.700
Like when I was working in Korea, like if it's a
foreign company, it's like,

00:06:39.860 --> 00:06:41.460
you know, you're not gonna fly here, buddy.

00:06:43.280 --> 00:06:48.410
yeah it's sad because it was all in your Ho Chi
Minh City is not like that like in Ho Chi

00:06:43.880 --> 00:06:45.320
It's very protectionist.

00:06:48.410 --> 00:06:54.480
Minh City it's very um open and very active like
when I when I started

00:06:54.480 --> 00:07:00.040
doing like the first time I came into this
meetup scene there were germans and others

00:07:00.040 --> 00:07:03.700
organizing JavaScript meetups, Ruby on Rails
meetups.

00:07:04.560 --> 00:07:08.600
I know in Singapore that was super active, even
much longer than in Ho Chi Minh City.

00:07:09.160 --> 00:07:14.720
Docker kind of really promoted that heavily and
then a lot of companies took note of how

00:07:14.720 --> 00:07:22.620
effective that was, of how big Docker grew by
really pushing out swag

00:07:22.620 --> 00:07:26.480
and support and advertisement on a global level.

00:07:24.940 --> 00:07:26.880
Oh, yeah, we all have a Docker shirt.

00:07:28.300 --> 00:07:31.640
clothes on my back thanks to Docker oh yeah

00:07:30.800 --> 00:07:35.340
oh yeah I'm still wearing like the third three-
year anniversary Docker shirt t-shirt that

00:07:35.340 --> 00:07:41.400
they they sent um that people when I go on calls
go like oh you

00:07:41.400 --> 00:07:45.240
you're wearing a interesting shirt and then they
see it every week and they're like you still

00:07:45.240 --> 00:07:46.700
still wearing that shirt

00:07:48.440 --> 00:07:53.780
yeah no anyway like I think Docker showed a lot
of large enterprises that this really worked

00:07:53.780 --> 00:08:00.000
and then Google they also started but then they
didn't want to have

00:08:00.000 --> 00:08:06.000
another platform on it, right? They had their
own dev communities that they organized

00:08:06.000 --> 00:08:11.860
they kind of split it up it's like Netflix made
movies popular and then you get Disney Plus

00:08:11.860 --> 00:08:18.000
you get Prime and then everything is messed up
because they want to own the platform

00:08:12.400 --> 00:08:14.340
Prime,

00:08:18.000 --> 00:08:21.120
and it's all split and splintered

00:08:19.680 --> 00:08:22.620
a bit of politics.

00:08:23.300 --> 00:08:28.520
I just remembered there was a Vietnamese guy
that I met while I was in AWS re:Invent called

00:08:28.520 --> 00:08:29.760
An Ho Viet.

00:08:31.680 --> 00:08:36.860
The AWS Vietnam user group and his company is
called Awesome.

00:08:31.840 --> 00:08:32.280
Yeah.

00:08:37.220 --> 00:08:38.860
I don't know. Does that ring a bell?

00:08:40.480 --> 00:08:45.040
Yeah, so again, I am an AWS Community Builder.

00:08:45.180 --> 00:08:51.560
I talk at AWS Community Days. I reached out
numerous times in Vietnam, but

00:08:51.560 --> 00:08:54.840
when I reached out, there was this conva, and
then they say, oh, you know what?

00:08:54.960 --> 00:08:59.600
AWS in Vietnam, it's really just selling to
partners

00:09:01.600 --> 00:09:07.600
community things not like in Singapore like yeah
like they're

00:09:03.180 --> 00:09:05.960
Oh, so it's not like developer aimed, yeah.

00:09:06.920 --> 00:09:07.580
Yeah, it's like

00:09:07.600 --> 00:09:13.340
they're not running like uh hey you know we're
hosting this um engineer who works at the

00:09:09.620 --> 00:09:10.000
a...

00:09:13.340 --> 00:09:18.520
telecom company or and and he built something
interesting uh on AgentCore or this is how they

00:09:18.520 --> 00:09:24.640
they they adopt certain patterns cloud patterns
and you know this very cool like

00:09:22.640 --> 00:09:24.780
...and you know, it's very cool...

00:09:24.640 --> 00:09:30.460
you know as as a person as an individual um
maybe I'm at a job and

00:09:31.600 --> 00:09:37.360
to play with something, I just want to learn or
I want to share my own experience playing

00:09:37.360 --> 00:09:37.800
with it,

00:09:39.240 --> 00:09:41.540
that you can go to even at AWS Singapore.

00:09:41.780 --> 00:09:45.760
They organize and they host these type of
meetups, which is really fun to go to.

00:09:44.900 --> 00:09:50.940
Yeah, yeah. I mean, you know I was the AWS user
group lead for a number of years.

00:09:52.200 --> 00:09:54.080
Oh, I didn't... Yeah.

00:09:54.000 --> 00:09:56.760
And I handed it over to Steve.

00:09:57.880 --> 00:09:59.660
Yes, Steve Teo, right?

00:09:58.180 --> 00:10:05.560
and the vibe was always to be developer focused
and to have tech talks I'm

00:10:05.560 --> 00:10:10.840
not even sure so you're saying the Vietnamese
one was a bit more a bit more like noob

00:10:10.840 --> 00:10:13.300
oriented and a bit more a bit more sales
oriented

00:10:12.760 --> 00:10:15.240
It's not noob, it's more like sales.

00:10:15.600 --> 00:10:21.060
It's more like partners of AWS selling,
reselling AWS or some large conferences.

00:10:19.140 --> 00:10:23.820
All right, that's a bit sad.

00:10:22.800 --> 00:10:28.960
I mean, that's my experience. And I've reached
out many times, even with people from

00:10:25.280 --> 00:10:25.860
Yeah, yeah.

00:10:28.960 --> 00:10:35.200
Singapore, like Safras, who was Google Cloud
first and then became AWS community

00:10:35.200 --> 00:10:36.780
lead across the region,

00:10:38.140 --> 00:10:44.280
Safras. He just said that he just left AWS this

00:10:39.620 --> 00:10:40.240
Never heard of it.

00:10:44.280 --> 00:10:50.460
week, I think. But it's very sad because, I
mean, I feel that it's

00:10:49.360 --> 00:10:52.500
I mean, it's not that sad.

00:10:50.460 --> 00:10:51.540
so much fun to share.

00:10:52.620 --> 00:10:58.120
I guess we're just being picky because we're
like elite coders and we want, or coders.

00:10:59.900 --> 00:11:04.340
infrastructure builders and we're looking to,
we're on the frontier, right?

00:11:04.540 --> 00:11:08.880
So we're a little bit picky about what we like
to attend.

00:11:08.380 --> 00:11:13.820
I look at Philippines and Kuala Lumpur because
they're close to Singapore so even when I went

00:11:13.820 --> 00:11:18.880
to community day in Singapore people from the
Philippines community groups were there people

00:11:18.880 --> 00:11:22.660
from the Kuala Lumpur community groups were
there I talked with some guys they said yeah
come

00:11:22.660 --> 00:11:27.880
to Kuala Lumpur they got me on the community day
there as well and I followed them on

00:11:27.880 --> 00:11:32.500
LinkedIn and we're friends and I see the
activity and I see how engaging it is.

00:11:32.640 --> 00:11:35.340
But in Ho Chi Minh City, it's not happening.

00:11:35.480 --> 00:11:39.680
And even when I try to approach people to get
something going, usually the response is like,

00:11:39.760 --> 00:11:45.220
we don't do that here. We don't, like the people
or even the AWS or people in Ho Chi Minh

00:11:43.020 --> 00:11:49.040
are in the air. Actually, when I was in
Malaysia, I talked about

00:11:45.220 --> 00:11:47.300
City are more focused on selling to business.

00:11:49.040 --> 00:11:53.720
community meetups, you know, what's happening
with community meetups post-COVID.

00:11:54.420 --> 00:11:54.860
Hmm.

00:11:56.580 --> 00:11:59.580
the verdict that I got was that most of them are
dead.

00:12:00.820 --> 00:12:05.620
but there are some diamonds in the rough sort of
thing.

00:12:00.940 --> 00:12:02.020
seem to be yeah

00:12:05.780 --> 00:12:10.780
Like the one that is pretty active is the Rust
meetup in Kuala Lumpur.

00:12:10.600 --> 00:12:12.780
oh that's awesome yeah that's a great

00:12:11.980 --> 00:12:14.180
And they meet every couple of weeks.

00:12:14.640 --> 00:12:16.980
I actually know the organizer, Ivan.

00:12:18.280 --> 00:12:24.760
And that sounds good. I mean, I would probably
in desperation call myself a Rust developer

00:12:24.760 --> 00:12:29.720
and attend those meetings because it's got some
cool people there talking about cool things.

00:12:30.300 --> 00:12:35.340
it's it's not about even being a Rust developer
it's it's like you know students who want to

00:12:30.600 --> 00:12:31.560
It's not about

00:12:35.340 --> 00:12:40.300
learn um you know some you know want to attend
and see some interesting projects and and

00:12:36.440 --> 00:12:36.800
Yeah.

00:12:38.880 --> 00:12:41.120
a bit weird.

00:12:40.300 --> 00:12:42.480
that's the same in the community in in kl

00:12:41.920 --> 00:12:48.000
It is a little bit elitist that it's called Rust
because it should be a little bit...

00:12:48.000 --> 00:12:54.060
There was back in the day a very good meetup in
KL called WebCamp KL or something

00:12:54.060 --> 00:12:56.760
like that, hosted at the Mindvalley offices.

00:12:56.760 --> 00:13:01.660
and that was really great because we got all
sorts of people there and it was very it was

00:13:01.660 --> 00:13:06.000
very vibey and it was a nice venue ah so

00:13:06.460 --> 00:13:11.160
I heard Da Nang in Vietnam is more like, like
when I'm in Ho Chi Minh City and there's like a

00:13:11.160 --> 00:13:16.960
hacker hackathon or something, I was at some
hackathon and then people came up came down from

00:13:16.960 --> 00:13:23.180
Da Nang and there were like foreigners and Da
Nang also is famous with like expats and like

00:13:23.180 --> 00:13:28.720
for people doing nomad lifestyle and I heard
several people said come to Da Nang and and I

00:13:28.720 --> 00:13:33.760
think there's like a bit of a hacker culture
there of you know bootstrapping your startup and

00:13:33.300 --> 00:13:34.780
a I think

00:13:33.760 --> 00:13:40.360
and talking and sharing and meeting and and you
know I think it lives in in Da Nang and

00:13:40.360 --> 00:13:45.540
I hope to revive it in Ho Chi Minh City this
time at the time I was working remote for the

00:13:45.540 --> 00:13:55.220
last two years but this time I have an office
and one of no

00:13:46.520 --> 00:13:51.860
you're in you're in some suburb out of Ho Chi
Minh, right?

00:13:52.060 --> 00:13:54.240
For some reason I thought you were in Da Nang
for a while

00:13:55.220 --> 00:14:00.380
I'm in in Ho Chi Minh City and I'm not like but
at least now there I'm working for a local

00:14:00.380 --> 00:14:05.720
company that has an office and that also says
hey we really want to bring back the meetups

00:14:05.720 --> 00:14:09.900
like the CTO specifically told me like he kind
of misses that and he wish that we could do it

00:14:09.900 --> 00:14:14.060
more and I've been here now like four months we
haven't set anything up because I've been

00:14:14.060 --> 00:14:18.800
crazy with building something, but it's
definitely something that I want to start
sharing.

00:14:14.560 --> 00:14:20.880
yeah I think yeah I think meetups are quite hard

00:14:20.880 --> 00:14:27.260
like the thing that annoys me at my employer
They try to do, we do these summits,

00:14:27.580 --> 00:14:32.160
we do these like get-togethers, these squad days
or something like that.

00:14:33.160 --> 00:14:39.460
And I think to myself that like, it just could
be better, you know, like, I

00:14:39.460 --> 00:14:45.860
don't know what I'm expecting. I guess my
reference point is a bit spoiled because in

00:14:45.860 --> 00:14:48.400
Singapore, A, it was easy to get to the venue.

00:14:49.020 --> 00:14:53.780
B, it was a really nice venue. I think in the
end we were using the AWS offices, which is

00:14:53.780 --> 00:14:58.000
like, you know, completely downtown, great bars
downstairs,

00:14:55.400 --> 00:14:56.000
yeah

00:14:59.160 --> 00:15:00.180
great food.

00:14:59.460 --> 00:15:06.180
yeah good to catch up after right and talk about
all kinds takeaways yeah

00:15:01.940 --> 00:15:04.540
Yeah, exactly. Easier to get home.

00:15:05.220 --> 00:15:08.640
Now it's such a schlep anyway.

00:15:09.480 --> 00:15:11.840
Oh, more stories from Malaysia.

00:15:11.840 --> 00:15:18.020
The interesting thing is that my friend in
Malaysia was using GLM 5.3 because

00:15:18.020 --> 00:15:24.260
I think there was some sort of Malaysian
investment to...

00:15:24.260 --> 00:15:25.220
Is it Z.ai?

00:15:25.440 --> 00:15:31.660
yeah I'm using it as well it's it's like I'm
using it for the workloads yeah

00:15:26.280 --> 00:15:30.860
and he seemed pretty happy with 5.3.

00:15:31.560 --> 00:15:32.780
Is there...

00:15:31.660 --> 00:15:37.680
I mean even what's this uh virtual card company

00:15:37.680 --> 00:15:44.220
called that's quite famous that a lot of people
have like you to issue a digital card with a

00:15:41.860 --> 00:15:48.000
mean like Visa

00:15:44.220 --> 00:15:45.180
credit card type.

00:15:47.200 --> 00:15:48.460
Very popular one.

00:15:48.000 --> 00:15:49.880
or MasterCard? What the hell? Amex?

00:15:49.480 --> 00:15:51.580
Yeah, but a startup.

00:15:53.340 --> 00:15:54.040
Revolut?

00:15:54.380 --> 00:15:58.940
Yeah, Revolut. Revolut, they just had a blog
post on their token, how they managed to cut

00:15:58.940 --> 00:16:02.480
token costs. And they are also heavily in on
GLM.

00:16:01.480 --> 00:16:04.660
use. Oh yeah, there was an AI engineer talk
about that.

00:16:05.040 --> 00:16:05.600
Yeah.

00:16:05.560 --> 00:16:11.020
Yeah, so do you have a Z.ai subscription or do
you just dabble with OpenRouter?

00:16:09.400 --> 00:16:13.060
I have DeepSeek and Z.ai, and I have a wallet.

00:16:11.140 --> 00:16:11.520
I have Z.ai

00:16:13.260 --> 00:16:15.080
I don't have a subscription, so I'll just top it
up.

00:16:16.160 --> 00:16:22.260
The reason being, the Anthropic models are
extremely expensive, and the enterprise

00:16:16.620 --> 00:16:17.140
Yeah.

00:16:22.260 --> 00:16:25.740
budget is kind of like cut.

00:16:25.920 --> 00:16:32.160
So the alternative is a LiteLLM proxy provided
by

00:16:32.160 --> 00:16:38.180
a platform team that only makes available

00:16:38.180 --> 00:16:44.220
certain open-weight models. And I think they
were evaluating GLM or DeepSeek Pro for

00:16:44.220 --> 00:16:49.320
Flash or whatever. And at the time when I was
building the platform that I'm building, I just

00:16:49.320 --> 00:16:55.360
said, because it wasn't available yet, so I set
up my Z.ai budget, like my wallet, and

00:16:55.360 --> 00:16:58.980
I run all of my testing on 5.3.

00:16:58.980 --> 00:17:04.540
So, yeah, I mean, it's just, yeah,

00:17:03.920 --> 00:17:05.160
The backup.

00:17:07.480 --> 00:17:13.960
I have a full sandbox environment, you know,
like everything mirrored

00:17:13.960 --> 00:17:20.020
from that is like, you know, that API endpoints
and wallet

00:17:19.900 --> 00:17:22.620
wallet. Ivan said it, but I thought it was quite
interesting.

00:17:20.020 --> 00:17:22.980
and version control.

00:17:22.840 --> 00:17:28.980
He just flips something over in his Claude
config, just changes the

00:17:28.980 --> 00:17:34.400
backend environment. And then he's using, so he
uses Claude code with GLM 5.3.

00:17:34.680 --> 00:17:37.360
I was like, wow, that's interesting.

00:17:35.200 --> 00:17:35.640
Yeah.

00:17:37.360 --> 00:17:41.200
You can just, I had a friend, because with
SpecLedger,

00:17:42.760 --> 00:17:48.860
one of the guys building SpecLedger with me and
the other guy, he was running his Claude Code

00:17:48.860 --> 00:17:52.380
against his own endpoints, which at the time I
think was DeepSeek.

00:17:52.140 --> 00:17:53.880
and um

00:17:53.760 --> 00:17:57.100
And because of that, when we build the
SpecLedger CLI

00:17:58.980 --> 00:18:04.480
as like it would onboard you, it would set up
the repo with like some boilerplate prompts and

00:18:04.480 --> 00:18:10.440
like skills for SpecLedger itself so they could
interact with the API and, you know, log in.

00:18:10.580 --> 00:18:16.680
And then it would launch a Claude Code or
whatever, you know, coding harness that

00:18:16.680 --> 00:18:20.900
you wanted to use. But honestly, we only really
try, like we're using Claude Code ourself.

00:18:21.080 --> 00:18:23.360
I mean, he would require that launching.

00:18:23.520 --> 00:18:29.960
So that user config, he would need it to be
supporting search endpoints like the

00:18:29.960 --> 00:18:33.700
environment variables required because for his
setup he needed to launch Claude Code with

00:18:33.700 --> 00:18:39.100
those configuration settings. So he built that
aspect of the launch or user configuration

00:18:39.100 --> 00:18:42.740
because the user configuration was supposed to
be like you can choose Claude Code or you can

00:18:42.740 --> 00:18:49.240
choose Codex or Gemini CLI or whichever harness
that you want and we're just delivering

00:18:49.240 --> 00:18:53.380
the skills and everything so that all of your
alignments, your intent,

00:18:51.620 --> 00:18:57.800
alignment. Well, after seeing Ivan using Claude
Code, I mean, it seems that

00:18:54.860 --> 00:18:55.880
are on the platform.

00:18:57.800 --> 00:19:02.440
Claude Code doesn't discriminate, and it seems
to work really well with other models.

00:19:02.600 --> 00:19:05.900
So why not just, as a harness, settle on Claude
Code?

00:19:04.040 --> 00:19:06.560
I think Anthropic loosened up a lot there.

00:19:07.300 --> 00:19:10.660
But you think Anthropic messed it up?

00:19:09.720 --> 00:19:14.860
Yeah, because originally when OpenClaw came
around, then Anthropic was very, like, they

00:19:10.860 --> 00:19:11.360
Because originally,

00:19:14.860 --> 00:19:20.320
banned OpenClaw the project. They started
locking down user accounts that were detected.

00:19:19.960 --> 00:19:20.600
but they affected

00:19:20.320 --> 00:19:22.720
They searched for Hermes VM strings or
something.

00:19:21.660 --> 00:19:26.100
There must be some limitations that Ivan didn't
show me, like the artifacts must obviously

00:19:26.100 --> 00:19:26.760
not work.

00:19:26.480 --> 00:19:30.820
I think that it changed because they didn't
support all of that.

00:19:31.240 --> 00:19:35.440
And OpenAI went the opposite direction. They
hired the guy from OpenClaw.

00:19:35.640 --> 00:19:41.700
They explicitly said you can use your OAuth
subscription with Codex and you can use it

00:19:41.700 --> 00:19:47.820
with OpenClaw. And with agent client protocol,
you can have your own interface on top

00:19:47.280 --> 00:19:48.680
Cool.

00:19:47.820 --> 00:19:53.920
of Codex to interact with it, one being OpenClaw
and the other one being

00:19:53.920 --> 00:19:59.800
Kiro Crew. the one that I built but I'm pretty
sure Kiro Crew also and now I see Kiro Crew

00:19:59.800 --> 00:20:06.240
mentioning support for Claude Code as a harness
like ACP as a harness which

00:20:06.240 --> 00:20:13.460
I thought would get my account banned and
another thing is you shared I think or I saw
Theo,

00:20:13.460 --> 00:20:19.520
he shows how his platform is able to launch
Gemini or Codex but

00:20:19.520 --> 00:20:23.520
also Claude Code and he also mentions it's on
ACP I didn't check if that's

00:20:24.260 --> 00:20:25.140
to the end, but

00:20:24.860 --> 00:20:29.240
So if you have like $200 plan, you get 4k of API
tokens.

00:20:26.920 --> 00:20:29.200
wish I used T3 Code a bit more.

00:20:29.340 --> 00:20:30.880
I was pretty excited after,

00:20:32.020 --> 00:20:33.620
for some reason I have this sort of

00:20:34.640 --> 00:20:40.100
schadenfreude relationship with Theo. I want him
to be wrong, but he's right a lot of the

00:20:40.100 --> 00:20:40.420
time.

00:20:41.340 --> 00:20:47.520
I mean, I think he changes his opinion enough to
be right

00:20:48.260 --> 00:20:49.340
a lot of the time.

00:20:48.480 --> 00:20:54.520
But I was pretty impressed with T3 Code but I
think that there were some things that I

00:20:54.520 --> 00:21:00.560
didn't like about it like the fact that it was a
big app and it felt a bit overwhelming and

00:21:00.560 --> 00:21:04.140
what was the other thing?

00:21:04.780 --> 00:21:10.820
I'm pretty addicted to computer use think it
supports it and I'm also pretty, it didn't seem

00:21:10.820 --> 00:21:14.220
to globalize my MCPs.

00:21:14.380 --> 00:21:18.900
Like in the video it says like it helps with MCP
management but when I was using it I

00:21:18.900 --> 00:21:25.120
couldn't figure out, you know, like I used my
Fastmail MCP and

00:21:25.120 --> 00:21:29.360
I wanted across all the agents I use but I
couldn't get it working so I was like oh
whatever

00:21:29.360 --> 00:21:31.700
I'll come back to this another time.

00:21:32.560 --> 00:21:35.720
So the thing I picked up when it

00:21:37.580 --> 00:21:40.500
that was the tail end of my holiday was Rex.

00:21:40.840 --> 00:21:41.960
Did you see my Rex video?

00:21:42.700 --> 00:21:48.840
I was just about to say you shared two things
that you said you were excited about and

00:21:48.840 --> 00:21:53.040
I said I think on both that you should have to
show me and I know you send a Rex video

00:21:54.860 --> 00:21:56.140
I haven't had time to look at it.

00:21:56.200 --> 00:22:02.140
I mean, I could maybe show you now, but it would
be a live demo and God knows what would

00:22:01.600 --> 00:22:05.720
But there was another thing that you were
excited about as well, right,

00:22:02.140 --> 00:22:02.360
happen.

00:22:07.160 --> 00:22:07.780
that I said?

00:22:07.580 --> 00:22:08.940
I'm excited about a lot of things.

00:22:10.560 --> 00:22:14.500
We're living in a golden age here. It's just got
to make the most of it.

00:22:14.560 --> 00:22:16.720
We got to acknowledge it. That is a beautiful
time.

00:22:17.420 --> 00:22:21.780
Oh, yeah, T3 Code is pretty good, annoyingly
good.

00:22:22.560 --> 00:22:27.980
And I said, you should show me. You said, Rex,
I'm going to try Rex, or I really like Rex.

00:22:28.100 --> 00:22:29.580
And then I said, OK, show me. I thought,

00:22:30.820 --> 00:22:31.860
yeah, I haven't had time.

00:22:33.280 --> 00:22:39.340
I feel like I'm kind of stuck in the way that
I'm working, although I did make

00:22:39.340 --> 00:22:42.860
a change in the way that I'm working a little
bit.

00:22:42.940 --> 00:22:46.420
But I'm stuck on Claude Code. I don't think it's
optimal what I'm doing.

00:22:44.460 --> 00:22:45.100
but I

00:22:46.620 --> 00:22:51.840
But are you sharing the screen yet?

00:22:46.940 --> 00:22:49.140
Okay, I have to show you.

00:22:51.820 --> 00:22:55.640
I'm here to show you Rex then.

00:22:51.880 --> 00:22:52.380
Not yet, right?

00:22:57.720 --> 00:23:00.500
You see my... Am I sharing the right screen,
hopefully?

00:23:01.120 --> 00:23:03.320
Yeah, it's blindingly white.

00:23:03.580 --> 00:23:08.320
I'm, you know, I'm a little bit blinded right
now.

00:23:04.200 --> 00:23:05.840
Yeah, I... I'm

00:23:09.560 --> 00:23:15.620
So I was just showing a session that I was just
working on on my own personal project.

00:23:16.680 --> 00:23:17.760
So what is Rex?

00:23:16.720 --> 00:23:23.020
What is So Rex is basically tmux

00:23:23.020 --> 00:23:26.400
and Ghostty combined. You use Ghostty, right?

00:23:23.960 --> 00:23:24.480
Okay.

00:23:27.480 --> 00:23:31.560
Yeah, I use cmux, which is also built on top of
a Ghostty library.

00:23:30.080 --> 00:23:36.080
Okay, so the cool thing is that you can have
remote ones, like for example

00:23:36.080 --> 00:23:39.600
I have my T14 running Rex, my Linux machine.

00:23:40.440 --> 00:23:42.720
But as I mentioned I'm really... Is this

00:23:41.340 --> 00:23:43.480
Oh, is it from Superlogical?

00:23:43.740 --> 00:23:44.960
Yeah, it's from Superlogical.

00:23:44.920 --> 00:23:49.740
Okay, so maybe give that background, because
Superlogical is a new company created by

00:23:49.740 --> 00:23:55.820
Hashimoto, the creator of Terraform and
HashiCorp who retired like a year ago and then
he

00:23:50.980 --> 00:23:51.520
Craser,

00:23:55.820 --> 00:24:01.660
built in his free time Ghostty which is a
library that gives you I don't know his focus
was

00:24:01.660 --> 00:24:07.880
it's written in Zig he really he's like one of
the two Zig um projects there was Bun and

00:24:07.880 --> 00:24:12.320
then there's Ghostty and he says Zig allowed him
to build like this super fast terminal

00:24:12.320 --> 00:24:18.800
emulator um and that was all open source and
free and it's built for macOS.

00:24:18.980 --> 00:24:24.540
And Ghostty terminal, the Ghostty terminal
itself is just a demo of what the library can
do,

00:24:24.760 --> 00:24:28.480
right? And then a lot of people built on top of
that, such as cmux.

00:24:27.760 --> 00:24:28.620
So, cmux.

00:24:28.920 --> 00:24:35.020
And then he started a company with several ex-
HashiCorp people after IBM took

00:24:35.020 --> 00:24:38.380
over. And they were very excited, Superlogical.

00:24:38.540 --> 00:24:44.400
One of the first videos they introduced, they
talk about completely changing the protocol for

00:24:44.400 --> 00:24:44.800
tmux

00:24:45.860 --> 00:24:48.300
is kind of very outdated. It runs over SSH.

00:24:48.460 --> 00:24:50.540
It's not performant in the way that it handles
characters.

00:24:50.620 --> 00:24:53.560
So they build this dedicated protocol

00:24:54.620 --> 00:25:00.600
for an advanced multiplexer, terminal
multiplexer client server protocol.

00:25:00.820 --> 00:25:06.860
And if the client supports that protocol, it
basically is a super, super fast alternative to

00:25:06.860 --> 00:25:13.160
tmux. So it's kind of like Ghostty being the
super fast terminal emulator that can render 3D

00:25:13.160 --> 00:25:15.300
character scenes.

00:25:15.860 --> 00:25:22.000
really nicely in the terminal and then Rex being
now the first

00:25:22.000 --> 00:25:28.100
you know it being in beta you can be on waiting
list so I've

00:25:28.100 --> 00:25:31.120
heard about it and you got on it

00:25:29.620 --> 00:25:36.140
got on it I got on by putting my email address
in the wait list I

00:25:36.140 --> 00:25:40.580
also just want to give some context so one of
the open source groups that I've been involved

00:25:40.580 --> 00:25:43.500
in for a long time is called suckless

00:25:45.100 --> 00:25:46.000
the founders

00:25:47.560 --> 00:25:52.640
a good friend of mine, Anselm Garbe, he's, well
we're not, we don't keep him touched so much

00:25:52.640 --> 00:25:55.120
because he's based in Vancouver right now.

00:25:56.420 --> 00:26:02.180
So the time difference is a bit painful. So like
for example in suckless.org, I think I've

00:26:02.180 --> 00:26:03.920
probably mentioned suckless to you before,

00:26:05.000 --> 00:26:07.440
though one of the main tools was a window
manager.

00:26:08.200 --> 00:26:09.120
Mm-hmm.

00:26:08.580 --> 00:26:14.660
Though a number of years ago, So st came about
because the community

00:26:14.660 --> 00:26:20.660
and Anselm recognized that terminals sort of
sucked and

00:26:20.660 --> 00:26:26.860
they needed to be reinvented. So st came about
and it's a very simple terminal.

00:26:27.320 --> 00:26:34.040
The source code is beautiful in the sense that I
think it's like a thousand odd lines as

00:26:34.040 --> 00:26:37.340
we try to do in suckless. Well, 2 ,000 lines
long.

00:26:38.480 --> 00:26:44.700
and I think it changed the game with the
terminals it led to some improvement

00:26:44.700 --> 00:26:50.700
and Hashimoto I believe studied it but

00:26:50.700 --> 00:26:58.020
of course took Ghostty to you know a level of
polish that's unbelievable but

00:26:55.380 --> 00:26:56.780
Hmm.

00:26:58.020 --> 00:27:02.280
that said like when I went when I'm I remember
talking with Anselm.

00:27:02.280 --> 00:27:08.360
Anselm always had in mind that tmux or you know
screen back in the days needed

00:27:08.360 --> 00:27:14.180
to be reinvented with this terminal so that you
would have a so that simple terminal would

00:27:14.180 --> 00:27:20.480
have a back end so that if you killed the window
you could resume it and it would be

00:27:20.480 --> 00:27:26.740
you know more integrated and it just amazes me
that like you know people of course you

00:27:26.740 --> 00:27:32.800
know people come up with ideas and the idea is
not new but credit

00:27:32.800 --> 00:27:39.140
credit to Mitchell Hashimoto he's Superlogical
and

00:27:39.140 --> 00:27:45.920
and Rex just in my limited uses

00:27:45.920 --> 00:27:48.800
of it is just incredible it's just incredible

00:27:48.240 --> 00:27:50.060
it's a lot of money behind them you know

00:27:52.520 --> 00:27:59.280
it's a bit like like uh on my on my what's this
on my oh yeah on marchie on

00:27:52.720 --> 00:27:53.280
it's a

00:27:56.280 --> 00:27:58.280
omachi omakon

00:27:59.280 --> 00:28:05.540
my con uh you know you have Mitchell Hashimoto
Jack Pearkes from a VP of engineering

00:28:05.540 --> 00:28:11.840
and R&D at HashiCorp Alasdair Monk, head of
experience VP of design at Vercel

00:28:11.840 --> 00:28:18.640
senior design leader at HashiCorp and Heroku and
then uh Hector Simpson who

00:28:18.640 --> 00:28:24.700
is agentic experience at Poolside but then And
the funding is Armon Dadgar from

00:28:24.700 --> 00:28:26.840
currently, I think, still at HashiCorp.

00:28:27.080 --> 00:28:32.840
And Guillermo Rauch, who is the Vercel, I think,
CTO.

00:28:30.740 --> 00:28:32.260
Oh yeah,

00:28:33.280 --> 00:28:35.540
the Vex, Vex Vercel, yeah.

00:28:33.540 --> 00:28:37.840
Tobias Lütke, who is the entire.io, like ex-
GitHub CTO.

00:28:38.140 --> 00:28:41.360
There's Dax Raad.

00:28:38.440 --> 00:28:44.020
Patrick Collison, Tobias Lütke, oh yeah, Mario
Red...

00:28:44.020 --> 00:28:44.960
What?

00:28:44.960 --> 00:28:47.060
Zechner, that's the pi.dev.

00:28:47.020 --> 00:28:49.400
How does he fund it? As

00:28:48.600 --> 00:28:49.500
Have money.

00:28:52.320 --> 00:28:58.560
Yeah, like, it just shows you what Silicon
Valley

00:28:58.560 --> 00:29:00.180
can do over Europe.

00:29:01.580 --> 00:29:09.340
I mean, suckless, I think, is quite a German-
European

00:29:09.340 --> 00:29:15.340
type of setup. up and and

00:29:12.100 --> 00:29:13.460
You can just see it from the website.

00:29:15.340 --> 00:29:21.840
you can see that you know exactly it

00:29:17.380 --> 00:29:21.040
it's like a university website from 2001

00:29:21.840 --> 00:29:24.460
can only go so far in a way

00:29:25.780 --> 00:29:32.120
amazingly Superlogical as with the right

00:29:32.120 --> 00:29:36.020
friends but anyway that sort of whatever origin
story aside

00:29:35.320 --> 00:29:40.280
so you use you have Rex you've been using that's
the whole background because it kind of that

00:29:40.280 --> 00:29:46.020
to me is important because it's not some random
lib Ghostty you know build project like cmux

00:29:44.220 --> 00:29:46.060
know, built like C

00:29:46.020 --> 00:29:51.240
no it's from the actual creators of cmux and
there's like the creators of Ghostty and there's

00:29:48.860 --> 00:29:49.480
no, it's going to be like the

00:29:50.580 --> 00:29:56.800
I mean, already I'm addicted to Rex and I'm
thinking to myself, I

00:29:51.240 --> 00:30:00.480
significant money behind it yeah

00:29:56.800 --> 00:29:58.820
wonder what Superlogical are going to charge for
this thing.

00:30:00.440 --> 00:30:00.740
Yeah,

00:30:00.480 --> 00:30:04.840
no they have in the announcement they have
something about free I think they will they will

00:30:03.520 --> 00:30:06.700
-Math.

00:30:04.840 --> 00:30:07.280
their focus will be on the hosted service right
the thing that costs

00:30:06.700 --> 00:30:12.920
And then another thing is that it requires
Tailscale to do the sort of

00:30:12.920 --> 00:30:15.720
connection to remote hosts.

00:30:16.340 --> 00:30:18.460
So I'm thinking aloud that,

00:30:16.340 --> 00:30:16.780
Okay.

00:30:19.780 --> 00:30:25.480
at least my employer doesn't allow Tailscale,
you know, it's not a sanctioned VPN.

00:30:26.520 --> 00:30:31.860
Okay, cool. So what's so good about it?

00:30:27.760 --> 00:30:30.120
So it's probably not going to fly.

00:30:32.220 --> 00:30:35.260
Because to me right now, it looks exactly the
same like cmux.

00:30:32.340 --> 00:30:35.320
Okay, it looks like a terminal.

00:30:35.660 --> 00:30:39.520
The thing that I kind of stumbled upon it,

00:30:41.360 --> 00:30:47.540
but the thing that blew my mind was

00:30:47.540 --> 00:30:53.600
the skills. And I did check with Orca if Orca
could do this, and Orca can

00:30:53.600 --> 00:31:00.360
kind of do this, but it's nowhere near as fast
and fluid as Rex.

00:31:00.740 --> 00:31:07.200
So assuming you have the skill, assuming you're
using Rex and assuming you have the

00:31:07.200 --> 00:31:12.820
skills installed and the hooks and all that sort
of stuff, you can do things like this.

00:31:12.940 --> 00:31:19.560
Like, so I'm working on a thing, I'm working
something on Codex, and then I prompted a

00:31:19.560 --> 00:31:24.880
bit further on here. can you launch a terminal
asking Claude the same question because I

00:31:24.880 --> 00:31:30.900
wasn't I was just thinking to myself I wasn't
too sure the way this was going so

00:31:30.900 --> 00:31:39.700
with with Rex installed it's able to base I'm
sorry yeah

00:31:33.940 --> 00:31:39.100
but this is a Codex session this is a Codex um
session

00:31:39.700 --> 00:31:45.240
inside that Codex skill yes I'll use the Rex
skill to open up a terminal and independently

00:31:44.800 --> 00:31:45.700
Okay.

00:31:45.240 --> 00:31:51.380
investigate the same spike. So then it launched
this tab, prompted it

00:31:49.240 --> 00:31:50.140
Nice.

00:31:51.380 --> 00:31:54.360
and it goes on from there.

00:31:53.240 --> 00:31:56.620
X-high effort. Astra, high.

00:31:57.120 --> 00:32:00.880
Opus 5.5, X-high. What are you doing?

00:32:02.260 --> 00:32:04.160
Am I, shouldn't I be using X-high?

00:32:04.640 --> 00:32:10.860
You're on GPT-6. That's one of Theo's latest
videos about OpenAI.

00:32:10.100 --> 00:32:13.580
Oh dropping it down. Okay so

00:32:12.140 --> 00:32:15.720
Yeah, you rarely run it on high or extra high.

00:32:15.820 --> 00:32:19.000
All right, okay, maybe I should and save someone
some money.

00:32:16.440 --> 00:32:17.240
But OK.

00:32:21.600 --> 00:32:27.960
But you get the point. And then furthermore, I
was using Claude in another session

00:32:27.960 --> 00:32:31.480
and I ran out of usage, like your usage, where
we said a half past four.

00:32:31.620 --> 00:32:37.840
I'm like, oh, damn it. And then all I needed to
do is go into Codex and I say,

00:32:38.260 --> 00:32:44.300
hey, Codex, in my Claude visitor spike, I've run
out of

00:32:44.300 --> 00:32:47.140
usage credits. Can you resume it here?

00:32:47.420 --> 00:32:52.760
So with the Rex skill in there, it's able to
like suck everything down

00:32:53.640 --> 00:32:57.880
from the Rex terminal or from where you last
left off.

00:32:58.220 --> 00:33:04.420
And I know, of course, Codex and Claude are
smart enough to like look at each other's

00:33:04.420 --> 00:33:08.260
chat logs. I mean, that's cool.

00:33:06.160 --> 00:33:08.180
Yeah, they're pretty good at that.

00:33:08.460 --> 00:33:11.580
But it also has this ability. And I thought that
was cool.

00:33:11.640 --> 00:33:15.800
And I found it just a little bit quicker and
faster to work with.

00:33:15.980 --> 00:33:18.640
Like, for example, if I had a new session here,

00:33:20.520 --> 00:33:24.000
and you can choose which like you can say this
mac or you can say remotely

00:33:24.160 --> 00:33:30.800
and I haven't tried remote one, but I could say,
like, you know, spawn three new

00:33:30.800 --> 00:33:32.100
Rex tabs.

00:33:35.400 --> 00:33:41.220
You know, I could say different work trees, and
that would also work.

00:33:41.640 --> 00:33:46.360
That's also a cool thing. Let me just go into
some Git repository.

00:33:47.780 --> 00:33:49.920
Okay, so here, sorry.

00:33:51.340 --> 00:33:56.260
I'm in Codex, I can say, spawn three Rex tabs.

00:33:58.100 --> 00:34:00.480
Does Rex have a voice integration?

00:34:02.060 --> 00:34:08.440
No, I don't think so. And also it seems to lack
image transfer across,

00:34:08.660 --> 00:34:11.400
that's a known issue to remote hosts. So that
sucks.

00:34:10.780 --> 00:34:15.660
OK, I'm going to ask some questions. So cmux
have similar capabilities.

00:34:15.660 --> 00:34:21.420
You can tell the agent. It actually wraps the,
like, remember I was showing you how I have

00:34:21.420 --> 00:34:24.600
these little dashes on the side where I can see
the agent turns.

00:34:24.980 --> 00:34:30.120
And I realized that's a cmux thing, because cmux
sees the agent turns and gives this little

00:34:30.120 --> 00:34:35.500
viewer so you can quickly scroll back to a
previous turn, like the little indicators on the

00:34:35.500 --> 00:34:38.280
left side of my screen, I think, on the last
time I showed.

00:34:38.820 --> 00:34:42.000
So cmux also has a deep integration with the
agent.

00:34:42.160 --> 00:34:46.520
It knows when the agent is waiting for some, you
know, it has raised an issue and it will

00:34:46.520 --> 00:34:49.420
notify you using your MacBook notification as
well.

00:34:47.320 --> 00:34:50.160
Yeah, the notification is also very beautiful
here.

00:34:50.620 --> 00:34:57.040
Yeah. And I don't know how much cmux is doing
for me right now, but I

00:34:57.040 --> 00:35:02.560
showed cmux to a colleague and then they used
cmux and they quickly customized it so that,

00:35:02.560 --> 00:35:06.600
you know, they have a nice view of all their
agents that they were running.

00:35:06.600 --> 00:35:11.380
And then they also started having the agents
interact with, because you have an integrated

00:35:11.380 --> 00:35:16.120
browser in there as well. So he was running an
MCP and then he was running a dashboard.

00:35:14.320 --> 00:35:18.500
Yeah, this one doesn't have the integrated
browser.

00:35:19.260 --> 00:35:25.620
So cmux, he actually built this whole thing
where he was interacting with like a graph, with

00:35:25.620 --> 00:35:31.420
nodes on the graph with an MCP. But he was also
able to ask the agent to look at the browser,

00:35:31.660 --> 00:35:35.340
integrated browser window, and then click on
buttons and interact with it.

00:35:36.600 --> 00:35:39.480
interacting with the agent and saying, like,
okay, I want you to do this and that.

00:35:39.580 --> 00:35:43.420
And it would, you know, it would either use the
MCP or the browser.

00:35:43.640 --> 00:35:47.460
And he was doing all of that on cmux, and then
he decided that he was hitting some limits.

00:35:47.560 --> 00:35:53.720
And then he built his own. So he built his own
app, macOS app, and then he integrated

00:35:53.720 --> 00:35:58.640
it completely with, he calls it, I don't know,
I'm not going to say the name.

00:35:58.460 --> 00:36:01.900
Okay, I mean everyone's...

00:35:59.720 --> 00:36:03.820
But it looks very similar to this and, like,
more...

00:36:02.660 --> 00:36:05.120
Well, I mean this is very clean.

00:36:05.300 --> 00:36:07.440
I mean this is just focusing on terminals.

00:36:06.060 --> 00:36:10.940
Well, this is just a terminal. Like, for people
who live in the terminal, like if you run

00:36:09.720 --> 00:36:12.580
So hopefully you got the point here.

00:36:10.940 --> 00:36:11.260
Vim.

00:36:12.700 --> 00:36:16.380
I basically just talked to it, and then I've now
got one...

00:36:13.200 --> 00:36:13.600
Yeah.

00:36:15.840 --> 00:36:18.100
Well, you don't talk to it because you don't use
a voice.

00:36:16.380 --> 00:36:16.560
But you

00:36:19.280 --> 00:36:24.800
I just chatted to it with text, and now I've got
three different terminals.

00:36:22.040 --> 00:36:22.440
Yeah.

00:36:25.360 --> 00:36:31.380
So once... Yeah, like in my video, the big

00:36:31.380 --> 00:36:36.480
unlock for me with Orca was having separate work
trees, and you just click through them.

00:36:39.240 --> 00:36:44.460
it's just a lot faster just to talk to it and
say like, I want to work tree prompting about

00:36:42.860 --> 00:36:43.300
OK.

00:36:44.460 --> 00:36:50.540
this issue and then you just split and you go,
yeah, it's

00:36:48.820 --> 00:36:51.080
OK, yes, I saw you can have multiple tabs.

00:36:50.540 --> 00:36:58.000
not hierarchical, but it's the way it's
organized in tabs, I feel is quite good

00:36:51.240 --> 00:36:51.980
You showed it twice.

00:36:56.460 --> 00:37:01.900
I think what would be interesting for me,
because this doesn't look interesting to me at
all,

00:36:58.000 --> 00:36:58.660
for my brain.

00:37:03.820 --> 00:37:09.120
this and more. But what would be more
interesting is if you can do this across remote
hosts,

00:37:09.540 --> 00:37:15.020
if that's the interactivity with multiple nodes,
like background agents running.

00:37:13.600 --> 00:37:18.260
Oh, yeah, yeah. I'm pretty sure you can do the
same thing with remote learning.

00:37:16.740 --> 00:37:20.960
Yeah. So that's something we should explore for
the next time, if you can have the same like

00:37:18.260 --> 00:37:18.420
That's something we

00:37:20.320 --> 00:37:21.340
Yeah, yeah. Like with

00:37:20.960 --> 00:37:24.160
with Orca, where you have the app. And I mean, I
don't use Orca.

00:37:25.900 --> 00:37:28.460
It's definitely like Rex is on my radar.

00:37:28.640 --> 00:37:32.860
I signed up on the wait list. so I'll have
ability to play with it.

00:37:32.380 --> 00:37:32.780
Yeah.

00:37:33.900 --> 00:37:36.120
Asking an agent to spin up other terminals.

00:37:37.200 --> 00:37:38.420
I think it's interesting,

00:37:40.540 --> 00:37:42.780
Yeah, I mean it's - I just hold on.

00:37:40.800 --> 00:37:47.420
I just want to say one thing. Hold on. I just
want to say something that changed

00:37:47.420 --> 00:37:51.600
the way I work in the last week, and I think for
the better.

00:37:53.620 --> 00:37:58.120
It's Claude Code's ability, and I don't think it
goes across to Codex or others.

00:37:58.120 --> 00:38:03.320
ability to to have a messaging channel between
agents right it's been there for a long time I

00:38:03.320 --> 00:38:08.380
was aware of it but never used it um the reason
was I wanted to control things I wanted to

00:38:08.380 --> 00:38:13.720
keep track of exactly what prompt goes to which
agent and I was having separate like I had I

00:38:11.220 --> 00:38:11.920
And I was

00:38:13.720 --> 00:38:18.260
started with one um you know planning agent that
took care of like user alignment and one

00:38:18.260 --> 00:38:23.000
building agent but then I was hitting deadlines
and I wanted to have parallel work so I

00:38:23.000 --> 00:38:27.240
started having like four agents, but I found
that more than four is hard to manage.

00:38:27.460 --> 00:38:33.680
And I would have the planning agent identify
work items for each other agent,

00:38:33.780 --> 00:38:38.840
and then I would be the messenger between, and I
would read the report, and I would forward

00:38:38.840 --> 00:38:43.640
it to keep control. But I was running out of
time again, so I was tired of it.

00:38:43.720 --> 00:38:48.940
And I asked the orchestrator, the planning
agent, I call him orchestrator, and I had
renamed

00:38:48.940 --> 00:38:53.200
some of the agent sessions. So in Claude, you
can slash rename session, and I had called him

00:38:53.200 --> 00:38:59.280
agent one, two of the project inside my Claude
Code so they were like you know they were

00:38:59.280 --> 00:39:03.600
long-lived sessions and if you have Claude Code
by default at some point it enables idle

00:39:03.600 --> 00:39:09.620
compact so if you before the cache expires which
is like I think an hour that the cache is

00:39:09.620 --> 00:39:15.840
valid the prompt cache um the Claude Code will
auto compact so I have a Claude Code running

00:39:15.840 --> 00:39:22.300
on my laptop for a week and it's auto compacted
like 29 times and

00:39:22.300 --> 00:39:28.600
and and so I have one agent which has some
context of like its name um its

00:39:28.600 --> 00:39:33.760
role what it has worked on within the project
and it has auto compacted so it lost some

00:39:33.760 --> 00:39:38.940
details but I always ask it to focus on the way
of working which is like it keeps a long long

00:39:38.940 --> 00:39:44.260
term goal and it runs dynamic workflows for
large pieces of work and it does small
background

00:39:44.260 --> 00:39:50.600
agents for fix-ups. So what I wanted to get to
was that in the last week, and

00:39:50.600 --> 00:39:53.520
this is also probably why I exhausted my ultra
subscription,

00:39:54.660 --> 00:39:57.400
I basically removed myself as the messenger.

00:39:57.840 --> 00:40:02.680
I asked each agent to introduce themselves to
the orchestrator, which was a named session,

00:40:02.720 --> 00:40:05.440
and they found each other. I said, hey, can you
introduce yourself to the orchestrator?

00:40:05.580 --> 00:40:08.100
And it was like, oh, I see the orchestration and
session number, blah, blah, blah.

00:40:08.320 --> 00:40:12.320
And I'm sending the message on the channel and I
went to the orchestrator and it showed,

00:40:12.760 --> 00:40:17.540
received message from agent one. itself and this
is what it's been working on because the

00:40:17.540 --> 00:40:23.240
biggest problem was I lost track of which prompt
I had to send to which agent and also I

00:40:23.240 --> 00:40:26.540
didn't always inform the orchestrator that the
prompt had been sent so the orchestrator was

00:40:26.540 --> 00:40:30.440
like agent two is still waiting for you to send
the prompt but in reality agent two already

00:40:30.440 --> 00:40:36.440
finished the work so from then on I told each
agent to always report back their status to the

00:40:36.440 --> 00:40:41.400
orchestrator and vice versa the orchestrator
receives their reports and would delegate it.

00:40:39.380 --> 00:40:43.880
So there's... This orchestrator is an agent in
Claude, right?

00:40:43.460 --> 00:40:47.680
It's all Claude Code. So every single one is a
single Claude Code session.

00:40:44.000 --> 00:40:44.240
It's

00:40:48.000 --> 00:40:51.160
Yeah. Well, this is where...

00:40:51.160 --> 00:40:55.480
Why I feel Rex is so much better, because with
Rex...

00:40:55.020 --> 00:40:55.980
How is it better?

00:40:56.820 --> 00:41:03.040
Because you can, it gives you like a, like you
could have Codex, you could

00:41:03.040 --> 00:41:05.840
have Claude, you can have Claude.

00:41:06.440 --> 00:41:13.200
The cool thing is that when you use these Rex
skills, you can communicate with the Rex

00:41:13.200 --> 00:41:15.660
CLI or API, I don't know what it's called.

00:41:16.100 --> 00:41:17.380
So you can ask Claude.

00:41:16.320 --> 00:41:17.460
And you can say like, you know, like.

00:41:18.980 --> 00:41:24.700
So that's what you showed earlier. You said
like, hey, Claude, my Codex is not out of usage.

00:41:24.920 --> 00:41:25.340
Yeah.

00:41:24.960 --> 00:41:28.480
But it went through the message. It didn't send
a message to Codex, right?

00:41:28.520 --> 00:41:30.800
They're not communicating with each other. You
are the messenger.

00:41:29.660 --> 00:41:33.260
Yeah, but it can work that way.

00:41:33.380 --> 00:41:39.420
The messaging can work that way. Yeah, so you
could be like, yeah, you can say, yeah,

00:41:33.960 --> 00:41:34.500
Can it?

00:41:35.720 --> 00:41:38.180
Can they talk to each other? Can Claude talk to
Codex?

00:41:39.500 --> 00:41:45.620
exactly. And you can say like, you can do, I
mean, I haven't really

00:41:42.660 --> 00:41:48.620
really I would like to see that I doubt it okay

00:41:45.620 --> 00:41:51.840
done it in anger, but you can do things like,
you know, I'm in this thing

00:41:51.840 --> 00:41:55.460
here, like Claude Z, communicate with Claude A,
with Codex A.

00:41:56.160 --> 00:42:01.800
about your work. And so everything's sort of
managed by Rex in a way.

00:42:01.940 --> 00:42:07.220
I mean, Rex isn't the orchestrator, but it gives
you that messaging platform, which I think

00:42:06.560 --> 00:42:08.480
Does it give you a messaging channel?

00:42:07.220 --> 00:42:07.900
is really powerful.

00:42:08.980 --> 00:42:13.700
It does a messaging channel through its API,
which is basically just text.

00:42:14.240 --> 00:42:20.260
Does it? Okay, go into Claude and have it
introduce itself to Codex

00:42:15.940 --> 00:42:16.980
Yeah, that's...

00:42:20.260 --> 00:42:21.320
on your session right there.

00:42:23.080 --> 00:42:25.600
Like one of the new ones I started.

00:42:25.940 --> 00:42:32.040
Oh, yeah. Maybe spawn another Codex session in
Claude 1.

00:42:25.960 --> 00:42:26.920
What do you want me to do?

00:42:32.040 --> 00:42:38.420
say maybe slash rename call it agent1

00:42:38.420 --> 00:42:44.520
com demo communication demo okay agent1

00:42:38.980 --> 00:42:40.340
Okay. Comdemo.

00:42:41.940 --> 00:42:42.540
Yeah.

00:42:44.520 --> 00:42:48.720
and then maybe Claude 2 call it agent2 slash
rename agent2

00:42:49.280 --> 00:42:52.500
Well, I mean each tab has its own name.

00:42:52.680 --> 00:42:54.980
I mean, we might be confusing it.

00:42:55.340 --> 00:42:57.820
I mean, I should probably be renaming it here.

00:42:58.580 --> 00:43:00.240
okay OK.

00:43:00.580 --> 00:43:03.300
I mean, I don't think it matters at the end of
the day.

00:43:02.980 --> 00:43:05.620
OK, it doesn't matter. I know that it works if
you give them.

00:43:03.300 --> 00:43:03.540
Okay,

00:43:05.700 --> 00:43:09.820
I mean, even if you don't give them a name, they
will find each other if you have some chat,

00:43:09.780 --> 00:43:12.300
Yeah, but I think it's more...

00:43:09.880 --> 00:43:13.880
because you don't have anything in there. And
now launch a Codex one, a separate new Codex

00:43:13.880 --> 00:43:15.100
one, and call it agent3.

00:43:15.640 --> 00:43:17.340
Okay, let's just get rid of this one.

00:43:20.760 --> 00:43:22.860
Oops, you ran a prompt rename agent.

00:43:23.160 --> 00:43:23.920
You didn't.

00:43:24.760 --> 00:43:26.980
Okay, so this one's called agent one.

00:43:27.220 --> 00:43:30.800
This one's called agent two. and this one...

00:43:29.840 --> 00:43:32.840
This one's not. It's messed up. Agent 2 is
messed up.

00:43:33.560 --> 00:43:34.700
Look, go back to Agent 2.

00:43:37.460 --> 00:43:38.840
What's wrong?

00:43:38.460 --> 00:43:40.460
Yeah, you didn't do the slash.

00:43:41.160 --> 00:43:45.480
Oh sorry, it seemed to understand me.

00:43:42.560 --> 00:43:43.040
Oh.

00:43:46.760 --> 00:43:50.720
Okay, now it works. Okay. And then this one is
Agent 3.

00:43:49.080 --> 00:43:50.000
Okay,

00:43:52.660 --> 00:43:54.980
You're already in code. Yeah.

00:43:53.140 --> 00:43:55.200
Does rename work here too? Agent 3.

00:43:55.740 --> 00:43:59.500
Okay, so tell Agent 3, introduce yourself to
Agent 2.

00:43:56.660 --> 00:43:58.640
So like, my name is Bob.

00:44:00.980 --> 00:44:01.580
Introduce...

00:44:02.560 --> 00:44:03.660
Why is your name Bob?

00:44:04.720 --> 00:44:07.300
Well, I mean, okay.

00:44:06.620 --> 00:44:09.560
You don't need that. You just say introduce
yourself to agent two.

00:44:09.900 --> 00:44:11.400
And let's see if Codex finds agent two.

00:44:09.920 --> 00:44:12.140
I just wanted to pass something.

00:44:13.220 --> 00:44:19.340
Okay, introduce yourself to

00:44:15.040 --> 00:44:15.480
Yep.

00:44:19.340 --> 00:44:23.680
Agent 2, which is actually Claude 2 in this.

00:44:20.600 --> 00:44:25.160
Yeah, and nothing else. Like, no, no, no, no,
no, no.

00:44:28.300 --> 00:44:30.980
So according to you, Rex skills.

00:44:31.560 --> 00:44:34.440
No, no, now it's using Orca. I got to prompt it
a bit better.

00:44:34.920 --> 00:44:41.220
I've got the Orca stuff installed right now.
Wait, let me

00:44:36.920 --> 00:44:37.720
OK.

00:44:41.220 --> 00:44:46.200
just say in Rex. I need to remove the Orca hooks
because it's messing up.

00:44:48.100 --> 00:44:48.900
OK.

00:44:54.040 --> 00:44:56.520
It looks like it might be able to find it.

00:45:03.300 --> 00:45:07.800
While this is going, go to agent 2 Claude and
say introduce yourself to agent 1.

00:45:10.520 --> 00:45:16.680
And don't say no, the other Claude Code,

00:45:10.880 --> 00:45:13.020
Oh, you mean agent 3?

00:45:16.420 --> 00:45:17.200
Oh, okay.

00:45:16.780 --> 00:45:20.740
because it's going to work within a second,
unless your skills are messing it up.

00:45:20.600 --> 00:45:22.340
Oh, it came through.

00:45:23.720 --> 00:45:24.220
Oh.

00:45:27.020 --> 00:45:30.200
Introduce yourself to...

00:45:29.300 --> 00:45:34.100
OK, so they introduced themselves as Codex and
Claude Code, and they responded.

00:45:31.980 --> 00:45:35.920
Yeah, and you can see that the way it does it
usually is by this sort of...

00:45:36.960 --> 00:45:40.260
this API that it does, this block capture thing.

00:45:40.100 --> 00:45:44.400
Interesting. So that works. OK, so that is nice.

00:45:44.880 --> 00:45:45.760
It's very nice.

00:45:44.900 --> 00:45:46.640
That is what you just said it does.

00:45:47.120 --> 00:45:50.140
Yeah, I wasn't lying, but I accept the Orca
thing is messing up.

00:45:51.180 --> 00:45:51.580
hmm

00:45:51.220 --> 00:45:59.040
but you're right, it did communicate a lot
quicker, I guess through the

00:45:59.040 --> 00:46:03.080
Claude harness, but the real power is that...

00:46:01.580 --> 00:46:05.340
it went fast there right it didn't do anything
it didn't have to do because because Claude

00:46:03.080 --> 00:46:09.120
Yeah exactly, this seemed to be faster, I mean

00:46:05.340 --> 00:46:08.800
Code has a channel that they can talk to each
other directly

00:46:09.120 --> 00:46:13.940
I didn't measure it, I didn't measure it but
it's cool right?

00:46:14.100 --> 00:46:16.940
I mean this is why I love Rex, just because

00:46:18.200 --> 00:46:24.180
It does enable it in a very clean, sort of like
non-confusing way.

00:46:18.900 --> 00:46:21.860
It does enable it. It's independent of the
harness.

00:46:24.380 --> 00:46:27.840
I don't have to rely on Claude to do all the
magic.

00:46:28.260 --> 00:46:33.060
It's just text. It's like as universal as Unix
pipes or something.

00:46:28.380 --> 00:46:29.000
OK.

00:46:34.000 --> 00:46:36.380
Yeah. OK, that's very interesting.

00:46:36.580 --> 00:46:37.380
I mean, that's interesting.

00:46:39.000 --> 00:46:39.620
Yeah.

00:46:40.840 --> 00:46:45.780
For me, this pattern is interesting because once
you've done that,

00:46:47.100 --> 00:46:53.120
they get into the habits. So with the
orchestrator, every time it would just

00:46:53.120 --> 00:46:55.260
feed back to me. Agent 1 is waiting.

00:46:55.460 --> 00:46:58.740
Agent 1 has completed this and it needs an
answer on these questions.

00:46:59.400 --> 00:47:05.440
Agent 1 has ran the AWS validation loop and
identified a couple of issues with our previous

00:47:05.440 --> 00:47:10.100
decisions. And it kind of gives me this one
dashboard.

00:47:10.140 --> 00:47:12.200
But it's not a dashboard. It's still text.

00:47:12.200 --> 00:47:18.440
and that's where I feel like I'm way behind in
actually having a nice

00:47:18.440 --> 00:47:21.280
interface. I think Terminal isn't it.

00:47:21.500 --> 00:47:24.640
I really need something else. Terminal is okay.

00:47:25.240 --> 00:47:29.060
There's a couple of startups now focused on
building, you know,

00:47:29.820 --> 00:47:36.080
Terminals are sweet. I do realize it does fall
down when you're working with web stuff

00:47:30.540 --> 00:47:31.140
the...

00:47:36.080 --> 00:47:40.180
and perhaps when you're working with the
computer use thing.

00:47:40.740 --> 00:47:44.940
Did I mention that the Rex iOS client is just
amazing?

00:47:46.660 --> 00:47:49.000
There's an iOS client. What does the iOS client
do?

00:47:49.160 --> 00:47:51.500
So this was a terminal. This is not the iOS
client.

00:47:52.900 --> 00:47:56.180
I'm sorry, I have to connect to bloody my VPN.

00:47:57.460 --> 00:48:00.480
Oh, you mean like iPhone? Also, you have a Rex
app.

00:47:57.640 --> 00:47:58.240
Oh, you

00:47:59.320 --> 00:48:04.760
Yeah, like for example, I can jump into that,
you know, that Claude instance that we were

00:48:04.760 --> 00:48:05.380
just chatting about.

00:48:05.220 --> 00:48:07.520
Okay. So that's interesting.

00:48:07.780 --> 00:48:10.840
I mean, like the Claude app is doing that as
well, right?

00:48:11.000 --> 00:48:14.180
Yeah, but the polish on this thing is insane.

00:48:14.380 --> 00:48:16.700
It's the best terminal emulator I've ever used
on iOS.

00:48:17.120 --> 00:48:18.320
Oh, that's nice. Yeah.

00:48:19.860 --> 00:48:20.280
Interesting.

00:48:20.600 --> 00:48:26.460
And it's even got like really nice sort of like
things where you can just like swipe left and

00:48:26.460 --> 00:48:26.740
right.

00:48:26.820 --> 00:48:29.580
Wow. OK. That's something that Claude, OK.

00:48:28.620 --> 00:48:31.060
It's just so smooth. It's crazy.

00:48:31.980 --> 00:48:36.660
Yeah, because the Claude app with, you know,
every time I open it, I have to go to the Codes

00:48:36.660 --> 00:48:42.900
tab, and I have to find the sessions. And if you
name your sessions, then the Claude app

00:48:42.900 --> 00:48:48.380
and you get nice notifications of orchestrators
waiting and agent one has completed this and

00:48:48.380 --> 00:48:54.460
so on but it's still like a clunky chat app when

00:48:54.460 --> 00:48:58.840
you're coming from the terminal where you have a
lot of commands and you have a nice you know

00:48:58.840 --> 00:49:04.600
you have a lot of capabilities but then you're
in this app on the phone and it's just like a

00:49:04.600 --> 00:49:10.840
chat window still I wouldn't want to become like
one of those cursor ide guys stuck to my ide

00:49:10.840 --> 00:49:16.960
or stuck to my terminal and not aware of a
better way like I think there's

00:49:16.960 --> 00:49:21.180
quite a few startups or companies that are
building a more intuitive way of working with

00:49:21.180 --> 00:49:27.260
agents I think kiro crew is one of those um you
know visualizations with a gateway um I

00:49:27.260 --> 00:49:31.480
think the Claude app itself is also trying to be
like you know another way of working with

00:49:31.480 --> 00:49:35.920
agents I I'm not used to that one yet I I don't
like the Claude app it feels it feels

00:49:38.220 --> 00:49:42.480
I like the lag. When I click on something, the
terminal is super snappy and fast.

00:49:42.660 --> 00:49:47.300
So I'm still stuck in cmux. I use the integrated
browser sometimes.

00:49:47.320 --> 00:49:48.140
It's pretty nice.

00:49:48.380 --> 00:49:52.500
The one problem I have is my MCP configuration.

00:49:52.780 --> 00:49:58.600
I really hate when I'm moving between computers,
setting up my MCPs all the time.

00:49:59.300 --> 00:50:05.420
So this is, I'm looking for a solution where my
MCPs are always

00:50:05.420 --> 00:50:07.020
like available.

00:50:08.660 --> 00:50:12.280
And I haven't found that yet. So there's still,

00:50:13.300 --> 00:50:16.580
that's probably the biggest gap for me in my AI
workflow.

00:50:17.640 --> 00:50:23.640
For me, the most annoying thing is that even if
I tell the orchestrator that I authorize you

00:50:23.640 --> 00:50:29.200
to do this, and then the permit, because when
you talk about MCP and the reason that it's

00:50:29.200 --> 00:50:33.200
difficult is sometimes because, you know, it's
credentials, it's sensitive.

00:50:33.200 --> 00:50:39.260
so with me the permission like the security
classifier across

00:50:34.500 --> 00:50:35.560
mission

00:50:39.260 --> 00:50:44.460
the agent is sometimes being very annoying like
agent one says that the classifier blocked

00:50:44.460 --> 00:50:44.720
his

00:50:45.420 --> 00:50:51.440
know. Oh, there's that even. I mean, to be
honest, I'm, on a personal note, I

00:50:45.780 --> 00:50:47.300
ability to do this or that

00:50:51.440 --> 00:50:56.180
don't do this in my employer. I always run my
things, basically, Skip Dangerous or YOLO.

00:50:56.560 --> 00:50:57.500
I can't be bothered.

00:50:56.760 --> 00:51:02.460
I always use the classifier and I think because
I'm doing it for work

00:51:03.200 --> 00:51:08.880
the work laptop has pushed, like I know Aikido
endpoint security, and then the work computer

00:51:04.800 --> 00:51:06.040
Yeah, that makes sense.

00:51:08.880 --> 00:51:13.820
has also pushed a whole bunch of rules that are
like, every single time that it runs a

00:51:13.820 --> 00:51:17.700
command on AWS and an account ID comes out, it
just flips out.

00:51:17.840 --> 00:51:19.840
It's like, big oops, gotta flag this.

00:51:18.880 --> 00:51:19.780
Oh yeah.

00:51:20.220 --> 00:51:26.380
Account code in logs. It's like, I think AWS
itself tells them, says publicly, account IDs

00:51:26.380 --> 00:51:29.900
are not like secrets. They're public
information.

00:51:29.900 --> 00:51:34.220
I remember googling it in the past, not recently

00:51:34.180 --> 00:51:36.840
What I was doing yesterday was so boring.

00:51:38.720 --> 00:51:41.460
Cloudflare doesn't, I'm so used to AWS.

00:51:41.600 --> 00:51:47.820
You know, AWS, I generally use, oh my God, the
OIDC thing.

00:51:49.780 --> 00:51:54.940
you trust the environment and then you can
deploy your Terraform or whatever into that

00:51:50.000 --> 00:51:50.660
Mm-hmm.

00:51:54.940 --> 00:51:59.040
environment using that simple trust association.

00:51:55.740 --> 00:51:56.440
Yeah.

00:52:00.280 --> 00:52:02.720
Yesterday I was messing around with Cloudflare.

00:52:02.860 --> 00:52:09.040
Cloudflare doesn't seem to have OIDC and the
problem that we had is that in in GitHub there

00:52:09.040 --> 00:52:12.340
was some Cloudflare API token set.

00:52:13.760 --> 00:52:19.120
You know, A, I can't read it back out so that's
you know problem A, I can't read

00:52:19.780 --> 00:52:25.800
out so I don't even know what token is set and
even if I could read out the token

00:52:25.800 --> 00:52:32.300
I don't think I could associate it to like if I
log into the Cloudflare dashboard there's

00:52:32.300 --> 00:52:38.800
a ton of tokens there's tokens attached to
different users and it's like in

00:52:33.560 --> 00:52:38.680
Oh, my God, the Cloudflare interfaces
administration panels is horrible.

00:52:38.800 --> 00:52:43.460
fairness Cloudflare does have this concept I
think they call them account keys now so it's

00:52:43.460 --> 00:52:49.480
not associated with users joy but then in the
account keys yeah

00:52:49.480 --> 00:52:55.520
you create them you know ask you to set an
expiration but like you don't want to set an

00:52:55.520 --> 00:53:00.360
expiration when you put it in bloody GitHub
Actions and then you set the key inside GitHub

00:53:00.360 --> 00:53:06.440
Actions and again you have the same problem like
you don't know what the rotation is or

00:53:06.440 --> 00:53:12.760
anything like this I mean basically API keys is
totally caveman right and that's what a

00:53:12.760 --> 00:53:19.640
lot of vendors I think Okta is another one that
really annoys me which

00:53:19.640 --> 00:53:23.940
doesn't have a very clean relationship so the
minute you you're dealing with API keys you're

00:53:23.940 --> 00:53:28.240
just setting up yourself a disaster right it
makes all the security stuff that you want to do

00:53:28.240 --> 00:53:29.580
so much harder

00:53:33.040 --> 00:53:34.800
okay you just tuned out there

00:53:33.680 --> 00:53:40.000
That's true. I mean, overall, it was a big rant
about Cloudflare

00:53:40.000 --> 00:53:46.680
being not very user-friendly, but at the same
time, Cloudflare has an amazing,

00:53:46.880 --> 00:53:52.920
I think, new paradigm in the way that you build
deploy or run in fresh like

00:53:52.920 --> 00:53:57.780
applications on the internet which I think is
something like we talked about last time I feel

00:53:57.780 --> 00:54:03.260
like I'm really missing out by not exploring
their like platforms more I also heard the same

00:54:03.260 --> 00:54:08.540
about Vercel like people telling me that Vercel
is their north star and how they built their

00:54:08.540 --> 00:54:14.620
platform on uh you know how for users to to to
use it and I feel like I'm missing out I feel

00:54:14.620 --> 00:54:18.360
like I'm missing out on on Vercel and Cloudflare
and and in that in that

00:54:20.000 --> 00:54:20.600
happening now,

00:54:21.660 --> 00:54:26.100
it seems to be that all of these SaaS, like
Vercel, which was a CDN, and then became like a

00:54:26.100 --> 00:54:30.120
next deployment mechanism, then I think they
acquired Supabase as well.

00:54:30.320 --> 00:54:33.640
And then with V0, they began with agentic
website builder.

00:54:33.640 --> 00:54:37.460
And now they are launching CI and continuous
integration.

00:54:37.460 --> 00:54:39.780
And I think they have version control as well.

00:54:39.960 --> 00:54:43.760
And then you have the same thing for all of
those other continuous integration providers,

00:54:43.960 --> 00:54:49.520
like depot.dev. They had Cloud Session Store
last year.

00:54:49.880 --> 00:54:55.380
They launched their own CI so that you don't
have to worry about Microsoft starting to charge

00:54:55.380 --> 00:54:58.760
for orchestrating your GitHub runners.

00:54:55.580 --> 00:54:57.460
for orchestrating

00:54:58.980 --> 00:55:05.840
Because they started as a GitHub runner, like
hosting provider, booting up your depot.dev.

00:55:03.360 --> 00:55:04.340
Which company is this? Sorry.

00:55:06.000 --> 00:55:09.420
They sponsored the open source project,
depot.dev.

00:55:08.180 --> 00:55:10.540
Oh, you just... Okay.

00:55:09.800 --> 00:55:12.280
They got a big VC investment as well.

00:55:12.280 --> 00:55:15.360
They completely changed their polish.

00:55:15.620 --> 00:55:20.320
But now, it's just one sample because you have
others like RunsOn and other CI.

00:55:15.940 --> 00:55:16.900
Well...

00:55:20.580 --> 00:55:25.800
Basically, what they did for their whole life,
for their whole existence, is providing

00:55:25.800 --> 00:55:28.960
isolated compute to run CI jobs. Guess what?

00:55:28.040 --> 00:55:29.160
guess what

00:55:30.180 --> 00:55:36.200
What is the main requirement today is to provide
isolated environments to

00:55:36.200 --> 00:55:39.000
run LLMs, sandboxes.

00:55:37.800 --> 00:55:40.980
Did you know just that Dagger went 1.0?

00:55:41.380 --> 00:55:47.400
yeah so dagger went 1.0 and I think they also
launched a cloud for it like a CI engine

00:55:47.400 --> 00:55:53.520
right they are a CI but yeah again they're the
same you need to build all

00:55:53.520 --> 00:55:58.140
of these CI I I'd completely missed that like
all of these startups that were building CI

00:55:58.140 --> 00:56:17.060
runners are now getting raising six million
dollars I

00:56:01.660 --> 00:56:07.240
But surely it's going to have the same problem
that I was just moaning about, the secrets in

00:56:07.240 --> 00:56:13.420
your actions and in GitHub where it doesn't

00:56:13.420 --> 00:56:16.260
solve that security aspect with API tokens does
it?

00:56:17.060 --> 00:56:21.240
think the biggest issue with all of these new CI
systems is that I don't know any enterprise

00:56:17.260 --> 00:56:18.080
I think the biggest issue

00:56:21.240 --> 00:56:26.080
that's going to be fast to move towards adopting
any of those I'm looking at them I was like,

00:56:25.560 --> 00:56:27.320
Yeah, exactly. They talk

00:56:26.180 --> 00:56:30.540
oh, my God. They talk about, you know,
milliseconds for scheduling job runs.

00:56:30.680 --> 00:56:35.000
They talk about, you know, all of these
capabilities, their version control, you know,
when

00:56:35.000 --> 00:56:40.560
GitHub is having outages left and right,
Bitbucket is having outages, and then they come
in

00:56:36.780 --> 00:56:38.860
right. Yeah. Bitbucket based on...

00:56:38.860 --> 00:56:40.460
I mean, to be honest, like...

00:56:40.560 --> 00:56:43.380
and they're like, you know, we are built for
agents.

00:56:42.020 --> 00:56:47.520
...my current client is a big company, it's a
big enterprise, and they're using Cloudflare

00:56:47.520 --> 00:56:48.720
for zero trust.

00:56:50.120 --> 00:56:56.200
And sometimes I think to myself, it's not an
enterprise solution, because...

00:56:54.960 --> 00:57:02.700
Oh, my God, it's horrible. I was on a project
that was looking at Cloudflare to set up a

00:57:02.700 --> 00:57:07.900
CDN connection to their AWS-hosted Kubernetes
cluster environment.

00:57:08.040 --> 00:57:15.120
And we had Argo CD for GitOps, and we wanted to
have ability to...

00:57:12.360 --> 00:57:15.440
ability to...

00:57:15.120 --> 00:57:16.160
Oh, they wanted...

00:57:17.540 --> 00:57:22.260
This was a highly regulated company and they
wanted from the Cloudflare CDN that they were

00:57:22.260 --> 00:57:23.620
paying a lot of money for.

00:57:25.000 --> 00:57:30.160
They wanted to have a private connection to the
load balancers in front of their Kubernetes

00:57:30.160 --> 00:57:32.440
clusters. And it wasn't possible.

00:57:32.840 --> 00:57:38.860
Cloudflare doesn't do that. You can do direct
connect, but it would take three or four months

00:57:35.280 --> 00:57:35.920
you got

00:57:38.860 --> 00:57:39.980
to set up.

00:57:41.320 --> 00:57:46.920
the only thing they would do is like the zero
trust with the Cloudflare tunnels, which is not

00:57:46.920 --> 00:57:51.600
a sustainable solution for like all of your
customer facing traffic.

00:57:51.740 --> 00:57:56.300
And this was a betting app. So the amount of
transactions they were getting is huge.

00:57:56.460 --> 00:57:58.660
And that was just ridiculous.

00:57:59.720 --> 00:58:04.360
I was like, why? I really wanted to kick
Cloudflare out of the project because I was
like,

00:58:04.360 --> 00:58:10.680
Yeah, it's so this is the problem like you
mentioned all these like newfangled solutions

00:58:04.660 --> 00:58:07.060
you just use Cloudflare and everything works out
of the box.

00:58:10.680 --> 00:58:17.580
But like the trick with enterprises is just
solid integrations and

00:58:17.580 --> 00:58:20.680
dependability that's hard to...

00:58:18.740 --> 00:58:21.500
Yeah, but Cloudflare is not a newfangled
solution, right?

00:58:21.620 --> 00:58:27.620
Cloudflare to me is like Akamai. They're this
huge CDN company that started as providing...

00:58:26.540 --> 00:58:32.740
I don't know, they have a sort of a nimble vibe
to me.

00:58:33.000 --> 00:58:37.040
But like, but yeah, when I use it, I think to
myself, oh my god, they need to fix this.

00:58:37.280 --> 00:58:40.080
They need that. They need OIDC.

00:58:38.360 --> 00:58:44.440
I feel like they have two sides. They have the
moneymaker, the big contracts that have

00:58:40.680 --> 00:58:41.500
They're not AWS.

00:58:44.440 --> 00:58:50.660
the firewall and, you know, the huge CDN network
and the huge core of, you know, data centers

00:58:50.660 --> 00:58:55.160
that they built out, which is highly, how you
say, inflexible.

00:58:55.460 --> 00:59:01.040
And when you interact with the dashboards or the
interfaces, it's horrible.

00:59:01.320 --> 00:59:06.260
It's so outdated. It's not UX friendly. and then
they have this whole new side of them which

00:59:03.320 --> 00:59:04.940
You're talking about Cloudflare?

00:59:06.260 --> 00:59:08.200
is like Durable Objects like the complete

00:59:07.680 --> 00:59:08.560
Yeah.

00:59:09.620 --> 00:59:14.960
twist on how you deploy into the cloud at each
location no it's horrible

00:59:10.680 --> 00:59:13.260
I think the Terraform provider is pretty good.

00:59:15.000 --> 00:59:15.680
Really?

00:59:16.320 --> 00:59:21.580
oh my god it's so bad because they just released
I think a minor version and they just break

00:59:17.880 --> 00:59:18.840
It's bad.

00:59:21.580 --> 00:59:26.740
things there's a lot of problems with upgrading
with this we are using Cloudflare and I just

00:59:26.640 --> 00:59:29.020
They are moving quickly, I must say.

00:59:26.740 --> 00:59:28.060
did version bumps and

00:59:29.820 --> 00:59:35.920
it was horrible but yeah I mean they do have a
Terraform provider which is

00:59:35.920 --> 00:59:39.200
already better than some like I don't think
Akamai has it

00:59:37.020 --> 00:59:43.620
Yeah, I mean, I Yeah, I want them to I want them
to win but yeah, they definitely

00:59:43.620 --> 00:59:45.420
not nearly as mature as AWS

00:59:46.620 --> 00:59:51.800
In fact, I'm wondering to myself. Why doesn't
AWS do a zero trust product?

00:59:51.840 --> 00:59:55.400
They probably do but it's But it's probably hard
to use or something

00:59:52.220 --> 00:59:54.260
they do have yeah absolutely

00:59:55.320 --> 01:00:00.700
and it's expensive it's extremely expensive I'm
building like this

00:59:56.700 --> 00:59:57.860
And it's inexpensive

01:00:01.640 --> 01:00:09.180
egress proxy for agentic so for

01:00:03.260 --> 01:00:08.060
I know Tailscale is more expensive than
Cloudflare, amazingly, anyway.

01:00:09.180 --> 01:00:11.040
AWS I'm building this

01:00:12.400 --> 01:00:18.520
agent sandbox environment so it's durable
workflows so I'm not actually having long

01:00:18.520 --> 01:00:21.300
lived agents and I'm giving short lived

01:00:23.200 --> 01:00:23.840
bindings

01:00:25.240 --> 01:00:31.380
when you talk about agents they have like they
either have a long lived identity and they

01:00:31.380 --> 01:00:35.980
are acting on a certain incoming session and
then they have a certain identity that they act

01:00:35.980 --> 01:00:40.760
on behalf of. And this whole identity management
of agents is really complicated, but it does

01:00:40.760 --> 01:00:44.660
matter a lot if you're trying to let an agent
interact with external systems.

01:00:44.840 --> 01:00:49.400
So for this thing that I'm building, I need to
have native egress capabilities.

01:00:49.940 --> 01:00:53.960
Like you can't do get clone on an MCP tool,
right?

01:00:54.160 --> 01:00:56.480
So we need to have native protocol support.

01:00:56.640 --> 01:01:02.940
And for that, I need to support the egress proxy
pattern that Claude and Anthropic

01:01:02.940 --> 01:01:06.000
offers. And they have it really nicely
documented.

01:01:06.260 --> 01:01:10.380
So I'm building this fancy egress proxy for
credential injection.

01:01:10.740 --> 01:01:17.640
And you can either generate your root
certificate authority and then define

01:01:17.640 --> 01:01:23.860
intermediate short-lived authorities and then
manage the, you know, because

01:01:23.860 --> 01:01:26.500
you need to TLS terminate connections.

01:01:26.500 --> 01:01:29.000
So you need to inject credentials

01:01:30.820 --> 01:01:35.060
like the network traffic. And you can do it very
cheap.

01:01:35.500 --> 01:01:41.360
But, of course, if you want to do it proper, you
don't want to manage a rule, certificate,

01:01:41.600 --> 01:01:44.000
authority, private key, right?

01:01:44.240 --> 01:01:47.600
That's a very sensitive piece of infrastructure.

01:01:48.000 --> 01:01:54.280
So you're looking at AWS Private CA, the ability
to run a private key infrastructure

01:01:54.280 --> 01:01:57.680
on your internal VPC only to do TLS termination.

01:01:55.420 --> 01:01:56.820
Yeah.

01:01:57.680 --> 01:02:01.660
The cost is close to $500 a month.

01:02:01.820 --> 01:02:08.820
And I'm building this on my mirror environment,
which I actually can't

01:02:08.780 --> 01:02:15.660
Why don't you just use Tailscale or something
and

01:02:08.820 --> 01:02:12.400
fork out $400 to build this.

01:02:15.660 --> 01:02:16.360
just...

01:02:16.440 --> 01:02:22.580
How does that fit into as an egress proxy for
sandbox and for agents

01:02:22.580 --> 01:02:24.180
that are living in a sandbox?

01:02:24.940 --> 01:02:29.620
Well, you can set up your own VLAN with
Tailscale.

01:02:29.260 --> 01:02:30.440
with Bitbucket?

01:02:29.900 --> 01:02:33.980
With You can set it up with anything, I'd
assume.

01:02:33.360 --> 01:02:34.000
Really?

01:02:34.060 --> 01:02:34.640
Really? Yeah.

01:02:35.680 --> 01:02:38.220
How would you set up an agent to have,

01:02:39.860 --> 01:02:46.020
without giving the agent an actual real key, a
real credential, how would Tailscale

01:02:46.020 --> 01:02:49.820
solve that problem to then connect to Bitbucket
and do a Git clone?

01:02:50.000 --> 01:02:55.600
I don't get that. Because the Git, a new
Bitbucket interaction needs an authentication,

01:02:55.600 --> 01:02:56.080
right?

01:02:57.680 --> 01:03:00.480
will solve my Bitbucket authentication problem.

01:03:02.700 --> 01:03:08.880
It sounds like Tailscale is just like a virtual
private network solution so

01:03:08.880 --> 01:03:13.940
that you can put machines on the same network
without encapsulating on top of the public

01:03:13.940 --> 01:03:16.140
internet. Am I getting it wrong?

01:03:15.280 --> 01:03:17.920
I have to get back to you about that one.

01:03:17.920 --> 01:03:20.680
But like, I was going to extol...

01:03:20.060 --> 01:03:21.800
Classic consultant response.

01:03:23.120 --> 01:03:25.240
I was going to extol the tailscar. Because like,

01:03:26.360 --> 01:03:32.420
let's agree that security in the age of AI is
going to get wild,

01:03:32.520 --> 01:03:35.540
right? We need to heighten that security.

01:03:35.900 --> 01:03:39.240
And on the topic of VPNs, not many people...

01:03:40.080 --> 01:03:45.540
Okay, no public facing API uses VPNs, right?

01:03:45.540 --> 01:03:50.500
but we're probably going to get to a point with
where where everyone has to run something

01:03:50.500 --> 01:03:57.220
like Tailscale in the sense that every
everyone's device has some sort of posture and

01:03:57.220 --> 01:04:03.240
then and then or some layer so when you connect
to that to the public API well

01:04:03.240 --> 01:04:08.520
it won't be public anymore it'll be protected by
something like Tailscale it'll have because

01:04:08.520 --> 01:04:13.600
like when you get when Tailscales connect to
another Tailscale there's there's like a concept

01:04:13.600 --> 01:04:19.780
of identity right with packets that you send and
then that will be the

01:04:19.780 --> 01:04:25.420
gateway to use something because I'm trying to
make this argument that like having an

01:04:25.420 --> 01:04:30.800
internet public API is probably not going to fly
in the future especially if it's a very

01:04:30.800 --> 01:04:31.660
sensitive API

01:04:31.540 --> 01:04:37.580
So, talk about the network separates the

01:04:37.580 --> 01:04:43.800
machines on it, or the principles, you know, the
identities, you're talking more about like,

01:04:41.800 --> 01:04:43.040
Yeah. You're

01:04:43.960 --> 01:04:46.820
it's not like the classic VPN, like WireGuard.

01:04:47.320 --> 01:04:50.200
It's more about zero trust type of setup.

01:04:50.580 --> 01:04:52.760
Yeah, these, I guess so.

01:04:53.460 --> 01:04:59.520
Like, I'm not too sure how I can show this to
you, but Tailscale has this like identity

01:04:59.520 --> 01:05:03.380
sort of part to it.

01:05:04.420 --> 01:05:10.580
So, for example, I could share my Tailscale
network

01:05:10.580 --> 01:05:16.660
with you, but I would probably control things
like, I can only talk

01:05:12.100 --> 01:05:13.120
-hmm.

01:05:16.660 --> 01:05:18.760
to my Raspberry Pi and you can't.

01:05:19.560 --> 01:05:22.980
Things like that. There's a whole lot of routing
and things like that.

01:05:24.090 --> 01:05:29.280
So that's where you say the BitBucket, if you're
able to.

01:05:24.320 --> 01:05:24.840
So,

01:05:30.000 --> 01:05:36.060
I mean, this product, like Zero Trust, I rolled
out Teleport a few times, and it has

01:05:36.060 --> 01:05:40.320
this idea of, it's the same that you don't trust
a network.

01:05:40.000 --> 01:05:41.400
Yeah.

01:05:41.540 --> 01:05:45.720
Every packet carries authentication.

01:05:46.040 --> 01:05:48.100
Well, the packet, it's more like in the protocol
level.

01:05:48.720 --> 01:05:54.040
The authentication is sometimes injected by the,

01:05:54.780 --> 01:06:00.800
like in Teleport, you can have an OIDC provider
it can, with SAML,

01:06:01.160 --> 01:06:07.300
push through your identity. So you log in once,
and then your identity is propagated towards

01:06:07.300 --> 01:06:11.740
any service behind. But you have to set up a
SAML integration or a...

01:06:11.240 --> 01:06:16.100
Yeah, yeah, yeah. There's some sort of, what do
you call that?

01:06:17.480 --> 01:06:22.820
SAML is the most annoying one because it's XML,
but there's the other one, I think it's JWT.

01:06:19.300 --> 01:06:20.640
But... The most annoying

01:06:22.960 --> 01:06:25.220
It's OIDC, sorry. It's OIDC, yeah.

01:06:26.600 --> 01:06:28.140
yeah Yeah.

01:06:27.880 --> 01:06:34.260
so what you would need then if you want to have
an agent running and cloning Bitbucket

01:06:34.260 --> 01:06:35.680
you would need to have some type of

01:06:37.580 --> 01:06:44.000
integration with Bitbucket OIDC so if the packet
comes in it automatically verifies

01:06:44.000 --> 01:06:46.440
the signatures and

01:06:47.840 --> 01:06:53.180
it knows that it was signed by an authority that
it trusts the identity provider and then it

01:06:53.180 --> 01:06:56.700
allows the activity I'm not sure you can do that
with Bitbucket.

01:06:57.940 --> 01:07:04.260
But yeah, zero trust in an agent age, in the age
of agents is probably...

01:06:58.000 --> 01:06:59.340
But that zero

01:07:04.840 --> 01:07:08.520
I mean, it was already a pattern that everyone
should adopt, but you know how many companies

01:07:08.520 --> 01:07:11.280
are actually sharing VPN profiles.

01:07:12.040 --> 01:07:16.660
Even when somebody left the company, they can
still log into the VPN and sharing long-lived

01:07:16.660 --> 01:07:19.640
credentials. That's still such a...

01:07:17.860 --> 01:07:18.380
That's

01:07:21.340 --> 01:07:24.740
I just want to get away from...

01:07:21.960 --> 01:07:22.880
It's nothing new.

01:07:24.740 --> 01:07:28.160
I spent hours messing with API keys yesterday.

01:07:28.500 --> 01:07:35.100
My biggest security win, my immediate security
win is just to have better trust

01:07:29.120 --> 01:07:30.680
Thank you.

01:07:35.100 --> 01:07:41.140
relationships. I don't know how it can work in
terms of zero trust and Tailscale exactly.

01:07:41.160 --> 01:07:44.680
I've got to investigate. But I don't want to
mess around with API keys.

01:07:44.800 --> 01:07:48.340
I want to say, you know, my role in Cloudflare,

01:07:51.360 --> 01:07:57.460
log in with my Tailscale and then that identity
is trusted in that

01:07:57.460 --> 01:08:00.680
context. So I don't have to mess around with API
keys.

01:07:58.660 --> 01:08:05.020
So, but that's the... In security, I think they
call it the zero secret,

01:08:05.180 --> 01:08:08.920
like the very first secret we need to establish
to establish the trust.

01:08:09.940 --> 01:08:13.900
And then there's also, it's very different when
you have a machine identity versus if it's a

01:08:13.640 --> 01:08:18.380
you

01:08:13.900 --> 01:08:14.660
user identity.

01:08:15.820 --> 01:08:21.840
Whereas, because AgentCore has this identity
component and they talk about

01:08:21.840 --> 01:08:28.040
like, you have different identity types When an
agent is acting on behalf of a user,

01:08:28.320 --> 01:08:32.820
then it usually means there's an authentication
flow that the agent gets requested for

01:08:32.820 --> 01:08:36.960
additional approvals before the agent can do
those things.

01:08:37.220 --> 01:08:41.020
So there's this back and forth, like, hey, the
agent wants to do these additional things, and

01:08:41.020 --> 01:08:46.320
there's this additional protocol on handling
that.

01:08:46.420 --> 01:08:48.480
Whereas if you're working with long-lived...

01:08:48.620 --> 01:08:49.600
Can I give an example?

01:08:51.480 --> 01:08:53.780
Yeah, when the agent is trying to do things

01:08:55.420 --> 01:09:00.680
have the ability to do, but the agent needs to
get the permission to do it.

01:09:00.780 --> 01:09:06.460
I know that in AgentCore gateway and AgentCore
identity, there's this inbound and outbound.

01:09:06.640 --> 01:09:11.580
And when a user flow is involved, it usually
means a back and forth because it means the user

01:09:11.580 --> 01:09:13.960
actually have to go and click yes, go ahead and
do that.

01:09:14.460 --> 01:09:19.800
Maybe there's a security classifier on the user
client side that can automatically approve

01:09:19.800 --> 01:09:22.340
based on the local activity.

01:09:22.340 --> 01:09:27.420
but there's this additional verification, which
you can't do when you talk with machine

01:09:27.420 --> 01:09:29.900
identities because there's no user to click
approve.

01:09:29.900 --> 01:09:35.340
When you talk with machine identities, they must
have a scope issue.

01:09:31.260 --> 01:09:37.440
Yeah, yeah. You must have I'm just, yeah, I'm
thinking that you can't do

01:09:37.440 --> 01:09:39.060
approvals in that context.

01:09:39.360 --> 01:09:42.980
Yeah, so the platform I'm building, it requires
that.

01:09:43.100 --> 01:09:49.300
And also the dream of everything being on short-
lived tokens and not

01:09:49.300 --> 01:09:53.940
ever having an API key or a long-lived secret
That is nothing new.

01:09:54.160 --> 01:10:00.420
That is the battle that I've been fighting in so
many organizations that

01:09:58.680 --> 01:09:59.880
in a new organization

01:10:00.420 --> 01:10:03.860
in an ideal scenario, you never have a secret
store.

01:10:03.900 --> 01:10:09.960
you don't Your comparison side, I keep on
forgetting the

01:10:04.000 --> 01:10:05.400
You don't need a password manager.

01:10:06.480 --> 01:10:08.700
But you always have third-party tenants.

01:10:09.960 --> 01:10:11.760
name, it's Tarkos.

01:10:10.720 --> 01:10:11.720
FACOS.guru.

01:10:13.180 --> 01:10:18.280
need like an infrastructure one where it names
and shames people who don't but

01:10:16.260 --> 01:10:20.140
It's there. Oh, also, there's an...

01:10:19.460 --> 01:10:25.480
that basically that the names shame people that
rely on rely on the API key

01:10:25.480 --> 01:10:27.020
for the integrations

01:10:27.280 --> 01:10:33.480
It's called the SSO paywall. There's a website
famous for name and shaming organization or

01:10:33.480 --> 01:10:39.360
product that puts SSO or short-lived tokens
behind a paywall, which is like...

01:10:39.360 --> 01:10:41.740
It should be basic capability of your product.

01:10:41.900 --> 01:10:44.600
You wouldn't need like an enterprise cost. Yeah,
SSO.tax.

01:10:45.220 --> 01:10:46.780
Yeah, that's the...

01:10:47.180 --> 01:10:49.920
Where's Cloudflare? Where's really Cloudflare?

01:10:51.640 --> 01:10:54.580
Somebody managed... Somebody's buddies' buddies
and took them out.

01:10:54.860 --> 01:10:58.600
Oh, you don't... Semporo Cloud is on it.

01:10:55.300 --> 01:11:02.160
Well, I mean SSO is probably not what I was...

01:11:02.300 --> 01:11:03.520
SSO is the first step.

01:11:04.740 --> 01:11:06.740
You should at least have SSO.

01:11:06.060 --> 01:11:10.380
Okay, we need an API key one though.

01:11:11.180 --> 01:11:16.220
That's an idea. Name and shame integrations that
use API keys.

01:11:16.500 --> 01:11:18.840
Hold on. Maybe I just need to ask Gemini.

01:11:17.860 --> 01:11:24.180
so chain Chainguard they're very public about
fighting long-lived

01:11:24.180 --> 01:11:30.820
secrets as well and they have several blog
articles about how some of their products allow

01:11:30.820 --> 01:11:36.880
you to like proxy to external systems that do
not have that I

01:11:36.880 --> 01:11:42.900
think one of them was um OIDC I forgot now but
but uh let me just

01:11:44.600 --> 01:11:48.760
Chainguard. OIDC short-lived credentials.

01:11:50.920 --> 01:11:56.440
Did we talk about the whole Chainguard, like,
this disclosure

01:11:52.560 --> 01:11:54.660
And what...

01:11:57.300 --> 01:11:59.460
failure? They're on a...

01:11:59.460 --> 01:12:00.980
Ah, it's octo-sts.

01:12:01.720 --> 01:12:04.780
No, this is not...

01:12:03.840 --> 01:12:06.180
Unwalt? No.

01:12:07.240 --> 01:12:07.900
SSO.tax.

01:12:17.080 --> 01:12:18.420
CVE or

01:12:19.000 --> 01:12:19.560
Yeah.

01:12:20.140 --> 01:12:20.620
RCE

01:12:21.600 --> 01:12:22.160
Yeah.

01:12:23.380 --> 01:12:28.700
so the have you followed that story about how
they started like it's a known issue with the

01:12:28.700 --> 01:12:30.180
memory management of the PHP

01:12:30.820 --> 01:12:37.180
To be honest, I ignore the PHP because I don't
use PHP and PHP has a billion CVEs,

01:12:37.240 --> 01:12:37.700
but yeah.

01:12:39.460 --> 01:12:41.960
it's not that it has a billion CVEs it's just
that

01:12:43.840 --> 01:12:48.420
So many people use it that it gets abused. And
then there's so many projects that are

01:12:46.080 --> 01:12:46.720
Yeah.

01:12:48.420 --> 01:12:52.200
portraying to be professional but actually have
leaks left and right.

01:12:52.280 --> 01:12:54.860
Nothing wrong with the PHP itself. It's just
that it's so popular.

01:12:54.120 --> 01:12:58.180
Well, I think PHP is kind of silly to run.

01:12:55.040 --> 01:12:58.500
And I don't like PHP.

01:12:58.960 --> 01:13:05.020
But it's silly to run because it's got like an
enormous surface

01:13:05.020 --> 01:13:08.020
area, right? Like if you get in there, you're
away.

01:13:08.460 --> 01:13:14.500
But so the whole thing was the project
maintainers had this

01:13:14.500 --> 01:13:19.380
long-standing open issue about memory management
that they were aware about some problems

01:13:19.380 --> 01:13:25.960
with it. And then with the advent of AI, some
research labs identified an

01:13:25.960 --> 01:13:28.700
RCE and I think raised it to them in September.

01:13:29.500 --> 01:13:34.460
pending fixes that they hadn't you know decided
yet how they were gonna like implement it so

01:13:34.460 --> 01:13:38.560
when this RCE came around they say yeah we just
you know had to make the decision of the of

01:13:38.560 --> 01:13:42.580
the solution and we were able to do that in like
I don't know two days or so and they were

01:13:42.580 --> 01:13:46.620
then building and and testing so they were
actually having the fix out very well but they

01:13:46.620 --> 01:13:52.100
were having a responsible disclosure and they
were working with partners and because PHP goes

01:13:52.100 --> 01:13:57.020
through distributions and people download it so
it needs to be a large coordinated effort on

01:13:57.020 --> 01:13:58.140
uh you know making sure

01:13:59.500 --> 01:14:01.500
to provide the fixes on time.

01:13:59.960 --> 01:14:00.860
Mm-hmm.

01:14:01.660 --> 01:14:04.980
So they're working with all these distribution
partners and everyone signed an agreement and

01:14:04.980 --> 01:14:07.240
they're in this responsible disclosure group.

01:14:07.240 --> 01:14:13.020
So they have a responsibility on following the
rules because you're privileged.

01:14:08.420 --> 01:14:09.560
Oh yeah.

01:14:11.680 --> 01:14:14.880
And then no doubt someone blurted out something.

01:14:15.220 --> 01:14:20.520
Yeah, so you're privileged to get this early
information about, hey, there's an RCE, here's

01:14:20.520 --> 01:14:25.180
the fix, make sure it's available. We're going
to make the announcement of the RCE and the

01:14:25.180 --> 01:14:28.460
release is available by, I don't know, Thursday,
2 p.m.

01:14:29.500 --> 01:14:34.500
And what happened is Chainguard, what we were
just talking about, they maintain secure

01:14:34.500 --> 01:14:39.920
images, right? Base images. And they broke the
embargo.

01:14:40.160 --> 01:14:46.300
They released like six hours ahead of the
scheduled time, which

01:14:40.280 --> 01:14:41.040
All right.

01:14:46.300 --> 01:14:49.900
forced everyone else to be on an accelerated
timeline.

01:14:49.900 --> 01:14:56.240
and the PHP foundation was quite upset with it
because all the other distribution

01:14:56.240 --> 01:15:02.040
partners in the Scrum within like four hours
instead of eight, I think they had to go and

01:15:02.040 --> 01:15:08.060
push. And a lot of comments were out there like,
what Chainguard did deserves

01:15:08.060 --> 01:15:12.440
them to get kicked out of this privileged
information group, right?

01:15:12.500 --> 01:15:16.620
you shouldn't be informed about RCEs anymore if
you're not going to follow embargoes.

01:15:15.280 --> 01:15:18.460
Yeah, the trust has been broken sort of thing.

01:15:18.180 --> 01:15:18.700
Yeah.

01:15:20.100 --> 01:15:25.700
And then the guys from Chainguard said, look, we
had no choice because we saw it actively

01:15:25.700 --> 01:15:31.720
being exploited. And therefore, we decided it is
best for our customers to

01:15:31.720 --> 01:15:35.540
release the fix prior to the lift of the
embargo.

01:15:33.740 --> 01:15:34.500
Oh really?

01:15:35.520 --> 01:15:40.220
I thought these things they allow you to release
the fix but they don't allow you to announce

01:15:36.080 --> 01:15:36.260
Yeah.

01:15:40.220 --> 01:15:42.780
it because I think I'm Mozilla

01:15:42.780 --> 01:15:47.160
No, it's an agreed point at which they do the
release.

01:15:46.460 --> 01:15:47.860
Really?

01:15:48.500 --> 01:15:54.780
yeah so basically I thought that's a reasonable
response right but then

01:15:54.780 --> 01:15:58.820
I read some threads and people are like yeah
they talk about they saw exploits but they never

01:15:58.820 --> 01:16:03.520
say where and when and they have no proof they
just say it but no proof and then another

01:16:03.520 --> 01:16:05.060
thing is like AWS

01:16:06.380 --> 01:16:12.660
I think AWS jumped on and said we saw exploits
like in August or in September

01:16:12.660 --> 01:16:15.480
right we they kind of

01:16:15.700 --> 01:16:17.100
Hmm.

01:16:17.160 --> 01:16:23.580
that story yeah but um yeah the whole PHP thing
was quite an you know quite the tea it

01:16:23.580 --> 01:16:28.260
was interesting and it's also it's also very
interesting because it was in a week of um you

01:16:28.260 --> 01:16:32.940
know Brotli from Google the compression
algorithm having you know they say Google is not

01:16:32.940 --> 01:16:39.080
having any maintainer on it and there were
exploits in it and Brotli is actually
implemented

01:16:39.080 --> 01:16:44.140
the you know the algorithm and the libraries are
ported over into many languages and it's

01:16:44.140 --> 01:16:45.420
integrated into Golang it's

01:16:46.520 --> 01:16:50.320
Oh, is that where the, oh yeah, the Brotli,
yeah.

01:16:50.360 --> 01:16:52.900
So they said there was not to shell.

01:16:50.820 --> 01:16:51.660
Is that where?

01:16:53.220 --> 01:16:55.660
Because in that week, you have so many exploits
coming out.

01:16:55.840 --> 01:16:59.520
You had the eMudhra intermediate certificate
leak.

01:16:58.100 --> 01:16:59.540
That was hilarious.

01:16:59.980 --> 01:17:00.940
That was hilarious.

01:17:02.320 --> 01:17:07.580
Basically, OpenAI or ChatGPT or Claude, I don't
know which LLM,

01:17:08.380 --> 01:17:13.120
was able to reverse engineer the actual prime
numbers that made up the private key.

01:17:13.260 --> 01:17:16.600
And it turns out that the complicated to cast
prime number was number four.

01:17:17.160 --> 01:17:20.700
have all these shirts, and Private Key 4, number
4.

01:17:21.100 --> 01:17:23.100
So that was hilarious.

01:17:23.360 --> 01:17:25.120
But that was a wild week.

01:17:23.980 --> 01:17:28.140
What I find even more hilarious is the xkcd
comic.

01:17:30.580 --> 01:17:35.060
I mean, this is an old one, right? I don't know
how you see when it was released.

01:17:37.020 --> 01:17:38.740
How do you see when it...

01:17:38.440 --> 01:17:40.320
Maybe the numbers are serial.

01:17:43.140 --> 01:17:48.820
Yeah, 221. I mean the latest one is 3319.

01:17:45.420 --> 01:17:46.380
Go to

01:17:47.520 --> 01:17:48.080
internet

01:17:49.420 --> 01:17:51.940
It's like predicted the future somehow.

01:17:49.820 --> 01:17:51.340
Wow, it's very early

01:17:53.280 --> 01:17:54.780
Yeah, there's quite a few

01:17:56.620 --> 01:17:57.960
XKCDs that are applicable

01:17:58.020 --> 01:18:00.560
But do you think it's just like four?

01:18:00.940 --> 01:18:09.040
I mean it's simulating a dice roll, but like
it's just so seven different so yeah

01:18:08.720 --> 01:18:09.280
Coincident

01:18:09.040 --> 01:18:11.020
it's so amazing

01:18:10.520 --> 01:18:12.000
Coinkidoink? I think not.

01:18:13.780 --> 01:18:18.960
So the eMudhra intermediate certificate private
key guest

01:18:19.820 --> 01:18:25.840
being extremely weak. Node2Shell being Node.js
needing an

01:18:25.840 --> 01:18:31.760
urgent patch because you're able to exploit the
compression algorithm and then you're able to

01:18:31.760 --> 01:18:37.920
get RCE. And then that became Web2Shell because
Brotli is implemented or this library

01:18:37.920 --> 01:18:39.940
lives in a lot of other languages.

01:18:40.320 --> 01:18:43.760
any web server that supports Brotli is actually
exploitable.

01:18:43.540 --> 01:18:44.680
Mm-hmm.

01:18:45.280 --> 01:18:50.860
Golang having to, with the certificate
revocation, very quickly release a patch because

01:18:50.860 --> 01:18:53.460
Golang implemented its own certificate

01:18:54.860 --> 01:18:56.980
revocation list verification or something.

01:18:57.500 --> 01:19:01.560
Basically, you had to update your Golang. Yeah,
that was a wild week, and I don't think it's

01:19:01.560 --> 01:19:03.880
done. I think it's still going.

01:19:03.980 --> 01:19:08.960
I think this is the new future, because we were
talking about Chainguard and security and...

01:19:06.380 --> 01:19:12.640
Yeah, but like if you didn't use PHP, I can't
help but think you are

01:19:12.640 --> 01:19:15.340
setting yourself up for success.

01:19:17.320 --> 01:19:23.160
I just said there's so many. I said the
intermediate certificate leak, there's the
because an

01:19:23.160 --> 01:19:24.800
LLM was able to guess the number.

01:19:23.820 --> 01:19:26.500
Okay, okay. It's not, you're right.

01:19:26.060 --> 01:19:28.760
I just said there's the Brotli algorithm.

01:19:26.660 --> 01:19:30.220
It's not PHP, it's Brotli, it's other things.

01:19:30.220 --> 01:19:36.300
but like but PHP has that surface area like at
the minute you can somehow leverage that

01:19:35.680 --> 01:19:39.340
Stop your PHP hate. The H doesn't stand for
hate, Kai.

01:19:42.200 --> 01:19:43.440
well I I

01:19:42.620 --> 01:19:44.480
It's pre-hypertext processor.

01:19:44.760 --> 01:19:50.700
all right well my client is using this thing
called PrivateBin have you heard of that for

01:19:50.700 --> 01:19:56.720
sharing secrets and that uses uh that uses

01:19:54.920 --> 01:19:57.720
Sounds familiar. It's PHP, I guess.

01:19:56.720 --> 01:19:59.880
PHP and I'm like I'm just like oh my god no no

01:20:01.220 --> 01:20:04.860
Isn't Facebook still on PHP? Didn't they build
their own PHP...

01:20:04.700 --> 01:20:06.460
Hack or something. I can't remember.

01:20:06.320 --> 01:20:08.700
No, engine, like, unoptimized.

01:20:10.320 --> 01:20:13.500
from Zend, they have their own engine, I think.

01:20:12.980 --> 01:20:15.760
Yeah, maybe you're right. Maybe I should stop
hating on PHP.

01:20:16.600 --> 01:20:18.700
And you accept JavaScript in your life.

01:20:19.580 --> 01:20:21.840
Not really. I don't really.

01:20:20.940 --> 01:20:24.340
Not really, right? But that's the crazy thing
now, right?

01:20:24.420 --> 01:20:26.840
JavaScript is everywhere. Durable Objects, Bun.

01:20:27.700 --> 01:20:29.640
I'm not a thousand percent sure Go

01:20:31.660 --> 01:20:37.660
But I believe the linker with Go, it basically
only pulls

01:20:32.200 --> 01:20:33.280
Sorry, it's Deno.

01:20:37.660 --> 01:20:41.300
in the functions that that particular app uses.

01:20:42.320 --> 01:20:48.580
So if you have a Go binary and you try to game
it and run Brotli decompress

01:20:48.580 --> 01:20:49.020
or something,

01:20:50.640 --> 01:20:55.140
nine times out of ten, that function wouldn't be
in the bundle.

01:20:55.340 --> 01:20:58.000
It wouldn't be in the static binary, so you
couldn't exploit it.

01:20:58.000 --> 01:21:04.440
So this is why I like to use Go because the
function surface area is much lower because

01:21:04.440 --> 01:21:05.500
the distribution doesn't

01:21:06.660 --> 01:21:09.280
ship up the standard library and then some,
right?

01:21:10.320 --> 01:21:17.480
You had to bump because every web server in
Golang is enabling the Linux kernel and TCP
flag,

01:21:17.520 --> 01:21:23.720
which has also been exploited. Like, if you
don't, you have to run go with a

01:21:21.160 --> 01:21:22.080
Yeah, you

01:21:23.720 --> 01:21:28.980
certain environment variable to prevent the TCP
usage, no matter how it's bundled.

01:21:26.980 --> 01:21:33.640
You're right. I mean, it doesn't make it immune
to those issues, but it does

01:21:33.640 --> 01:21:39.200
reduce the surface area in my mind, but I'm
probably just clinging on to nonsense.

01:21:38.620 --> 01:21:45.020
Go is so much better. But the thing is, when I
build with JavaScript,

01:21:45.520 --> 01:21:49.280
you have to bundle, which is exactly what you
described.

01:21:49.680 --> 01:21:52.520
You're not shipping node modules.

01:21:52.520 --> 01:21:58.980
You're resolving every library's

01:21:53.160 --> 01:21:54.560
Linking.

01:21:58.980 --> 01:22:02.100
function that you depend on and bundle it.

01:22:02.480 --> 01:22:10.640
So bundlers like esbuild, they generate a single
program

01:22:10.060 --> 01:22:10.980
Yeah. Yeah.

01:22:10.640 --> 01:22:16.060
that only inlines the functions that you depend
on and not the whole node modules, like the

01:22:16.060 --> 01:22:17.940
whole package that you depend on.

01:22:17.620 --> 01:22:23.980
But it still runs in this Node.js runtime that
has a lot of things

01:22:23.980 --> 01:22:24.640
in it, right?

01:22:26.200 --> 01:22:27.080
No? OK.

01:22:26.620 --> 01:22:30.440
It runs in QuickJS or it runs on the edge.

01:22:31.980 --> 01:22:33.080
OK, all right.

01:22:32.020 --> 01:22:34.740
Right now, JavaScript runs everywhere.

01:22:35.060 --> 01:22:36.380
And I mean,

01:22:37.480 --> 01:22:40.560
you can hate on PHP,

01:22:42.480 --> 01:22:46.200
but I think JavaScript is at least as bad.

01:22:47.420 --> 01:22:50.740
But it turns out you can write anywhere.

01:22:52.520 --> 01:22:56.680
you can you know what I got wrong earlier when I
said Cloudflare there they didn't acquire

01:22:53.820 --> 01:22:55.180
Yeah, yeah, yeah.

01:22:56.680 --> 01:22:59.840
Bun they acquired Deno and Deno had this

01:22:58.860 --> 01:23:04.880
I mean, yeah, anyone listening, I'm not just
being, I guess

01:23:04.880 --> 01:23:11.020
I'm a bit tribal, like I do have my my ways of
doing things because you have to you

01:23:11.020 --> 01:23:17.200
know and there's so much out there isn't there
Vincent and I do have my my way around

01:23:17.200 --> 01:23:23.220
managing all this complexity so yeah you got to
forgive me then when

01:23:23.220 --> 01:23:29.240
someone talks about a PHP vulnerability I'm just
like batting my eyelids and moving on

01:23:28.260 --> 01:23:32.640
Yeah, no, the reason I talk about PHP, because
there was some tea regarding Chainguard, and

01:23:29.240 --> 01:23:29.440
The

01:23:32.640 --> 01:23:37.740
Chainguard has this huge initiative about short-
lived secrets, which they pride themselves in

01:23:33.260 --> 01:23:36.700
Yeah, you're right. That was...

01:23:37.370 --> 01:23:40.760
And I'm still a bit bemused by that process.

01:23:37.740 --> 01:23:38.260
security.

01:23:41.200 --> 01:23:46.460
I thought you're allowed to release fixes, but
not announce it.

01:23:46.640 --> 01:23:50.660
I mean, a lot of companies have released
security fixes without telling them what the
thing

01:23:48.900 --> 01:23:54.500
What was this exploit? There was this thing that
leaked.

01:23:50.660 --> 01:23:50.840
is.

01:23:54.700 --> 01:24:00.780
I think it was on the Linux kernel. There was
this CVE that was announced,

01:24:00.780 --> 01:24:07.360
and then somebody identified a similar CVE, like
in an adjacent code

01:24:07.360 --> 01:24:12.840
path. And they created a branch, a public
branch, to implement a fix.

01:24:13.100 --> 01:24:19.300
And the name of the branch referenced the
family, like family of exploit.

01:24:19.580 --> 01:24:25.980
So somebody saw the branch, saw the fix being
worked on, and tried to identify if

01:24:25.980 --> 01:24:29.220
there was another exploit. And then it turned
out into a zero day.

01:24:29.420 --> 01:24:31.480
There was this thing that happened with, I
think, the Linux kernel.

01:24:31.740 --> 01:24:36.080
So when you say you're allowed to announce, I
think some of this work is difficult if it

01:24:36.080 --> 01:24:36.400
happens.

01:24:38.260 --> 01:24:40.740
that this type of RCEs are being worked on
privately.

01:24:42.740 --> 01:24:43.960
I'm gonna ask chatGBT.

01:24:43.140 --> 01:24:51.100
Yeah, I actually haven't really spent time to
understand this Linux remote

01:24:51.100 --> 01:24:52.080
exploit.

01:24:53.700 --> 01:24:56.520
There's some exploit in IPv4 network stack.

01:24:57.870 --> 01:24:59.980
I know they created a remote commode.

01:25:00.480 --> 01:25:02.840
I mean, these are very rare, aren't they?

01:25:10.640 --> 01:25:12.140
MPTCP, I didn't even know what that is.

01:25:13.820 --> 01:25:15.440
MPTCP, yeah.

01:25:15.620 --> 01:25:16.340
What is that?

01:25:21.450 --> 01:25:26.700
I thought it was like multi-channel TCP or
multiplex TCP.

01:25:27.140 --> 01:25:29.440
Oh, multipath. Yeah.

01:25:27.380 --> 01:25:27.940
Multipart. Oh,

01:25:34.200 --> 01:25:34.580
Interesting.

01:25:35.330 --> 01:25:36.400
Oh, DirtyFrag.

01:25:38.300 --> 01:25:44.360
So DirtyCow was public. Everyone had to roll
their Linux kernel or roll their Linux

01:25:44.360 --> 01:25:47.380
machines. And then right after there was...

01:25:47.380 --> 01:25:49.560
Oh, it was copy fail, sorry.

01:25:50.200 --> 01:25:54.460
So DirtyFrag was disclosed in May 2026, shortly
after copy fail.

01:25:54.700 --> 01:26:00.700
So the maintainers published a security fix in
the public netdev.net.git tree ahead of the

01:26:00.700 --> 01:26:06.680
planned coordinated disclosure. And that
resulted in somebody looking at this,

01:26:06.820 --> 01:26:11.060
like, what are they trying to fix? there's a
huge, so then it's a zero day, nothing is

01:26:11.060 --> 01:26:11.460
patched.

01:26:12.660 --> 01:26:18.840
In your current role, do you Are you like

01:26:18.840 --> 01:26:21.940
the security guy or you just wear that hat?

01:26:23.300 --> 01:26:25.180
I think it's everyone's responsibility.

01:26:25.100 --> 01:26:26.840
It's everyone's responsibility.

01:26:27.320 --> 01:26:32.860
I think I think I am very interested in like,
reading the backstory and the history and how,

01:26:33.140 --> 01:26:39.440
like the Phineas hack happened against what's it
called the security team you

01:26:35.480 --> 01:26:37.420
Yeah, it's a good idea.

01:26:39.440 --> 01:26:44.460
know how the spanish police officers were you
know hacked by phineas fisher as well like

01:26:44.460 --> 01:26:48.700
those are extremely interesting stories when you
read how they penetrated the network how

01:26:48.700 --> 01:26:55.880
they were able to you know move along inside the
network and it

01:26:55.880 --> 01:26:56.160
just

01:26:58.220 --> 01:27:04.360
it's just important I think to be aware um but
to be honest my position is today

01:27:01.540 --> 01:27:02.680
Yeah, true, true.

01:27:04.360 --> 01:27:08.840
like when I talked about earlier when I was
talking about setting up this egress proxy and

01:27:08.840 --> 01:27:15.500
managing root you know certificate authority
secret material like a private key that's

01:27:15.500 --> 01:27:20.560
already like you just don't want that
responsibility you don't want to hold on to
secret

01:27:20.560 --> 01:27:24.880
material like that that's the I'm all in on
managed solutions there.

01:27:21.180 --> 01:27:22.020
That's what I mean,

01:27:24.980 --> 01:27:26.000
I think it's

01:27:25.060 --> 01:27:31.140
And I think it's the position of the bank. If
you've run Linux servers or anything like an

01:27:31.140 --> 01:27:32.680
EC2 instance with public ingress,

01:27:34.840 --> 01:27:41.020
today you feel the pain because you need to be
constantly

01:27:40.360 --> 01:27:41.340
On it.

01:27:41.020 --> 01:27:43.520
rolling. I was joking.

01:27:43.720 --> 01:27:47.260
It was like, oh, I waited a week and now I can
jump ahead.

01:27:46.560 --> 01:27:52.680
So it makes it all the more important that
people use AWS and cloud

01:27:47.460 --> 01:27:50.400
I only have to do one rollout, not like three in
the last week.

01:27:52.680 --> 01:27:55.680
solutions because they have a... Yeah.

01:27:53.460 --> 01:27:58.820
Yes. And that's why I think it's double
important that when you build on the cloud
today, or

01:27:55.840 --> 01:27:56.020
That's why

01:27:58.820 --> 01:28:04.820
that's why all of these initiatives like Vercel,
the platform that they offer, they

01:28:04.820 --> 01:28:10.740
take on that responsibility. the AWS shared
ownership responsibility but if you run on AWS

01:28:06.160 --> 01:28:06.760
Oh really?

01:28:10.740 --> 01:28:17.120
you should be using these newer cloud managed
solutions and way of building where you just

01:28:17.120 --> 01:28:20.420
run your program and you don't manage the whole
network stack

01:28:20.540 --> 01:28:22.700
Yeah, you've got to choose your battles, haven't
you?

01:28:21.580 --> 01:28:27.700
because and that's where if you're stuck with
plain old HashiCorp language Terraform

01:28:27.700 --> 01:28:29.200
you're screwed

01:28:30.860 --> 01:28:31.300
Yeah,

01:28:34.180 --> 01:28:37.780
you're right. I mean, there's going to be so
much transformation happening.

01:28:39.100 --> 01:28:39.660
I mean,

01:28:41.560 --> 01:28:46.720
I'm pretty much stuck in the world of
enterprises where things move really slowly and
it

01:28:46.720 --> 01:28:52.520
needs to move faster? I'm just actually thinking
aloud, how do we get to start moving faster

01:28:52.520 --> 01:28:52.740
here?

01:28:53.880 --> 01:28:59.880
For me, the choice now is very simple. If it's
not a managed AWS service, it means it's

01:28:53.960 --> 01:28:54.660
Because... For me,

01:28:59.880 --> 01:29:03.340
going to take a long time because it's going to
need to go through a whole chain of

01:29:03.340 --> 01:29:07.660
approvals. it's going to have to go through a
whole review, it's going to take so much longer

01:29:05.280 --> 01:29:06.620
easier

01:29:07.660 --> 01:29:12.600
who's responsible for patching, who's
responsible for intrusion detection, who's
responsible

01:29:11.240 --> 01:29:11.880
Yeah.

01:29:12.600 --> 01:29:18.760
for all of this so the choice is easy, you want
to move fast, you want to move at as high

01:29:18.760 --> 01:29:24.220
as possible a level you want to be able to ride
on their compliance and their

01:29:24.840 --> 01:29:28.400
security guarantees but don't use Northflank or
Railway

01:29:25.560 --> 01:29:26.300
Yeah. Yeah.

01:29:27.960 --> 01:29:30.880
But I mean, there's a serious...

01:29:30.880 --> 01:29:31.040
Yeah.

01:29:32.280 --> 01:29:36.480
did you hear also about Northflank and Railway
oh we haven't even talked about the Binance

01:29:32.760 --> 01:29:33.140
But...

01:29:36.480 --> 01:29:36.700
hack

01:29:37.200 --> 01:29:41.060
Oh my god. Well, I've washed my hands of crypto
long ago.

01:29:41.640 --> 01:29:45.460
I mean, Binance for me is like crypto is dead,
surely.

01:29:45.700 --> 01:29:46.520
It must be dead now.

01:29:47.200 --> 01:29:49.690
oh my god did I said the same at Mt.

01:29:49.690 --> 01:29:52.790
Gox hack back in 2012 I thought the Mt.

01:29:52.790 --> 01:29:58.820
Gox hack that's you know crypto Bitcoin went
down I was like um see I'm right and then right

01:29:57.320 --> 01:29:58.820
It's over. It's over, guys.

01:29:58.820 --> 01:30:06.060
after I think at the time I don't know it was
like a hundred dollars I

01:30:01.480 --> 01:30:04.020
You're a dinosaur if you remember

01:30:05.480 --> 01:30:07.040
Okay I need to take...

01:30:06.060 --> 01:30:10.400
remember one guy no there was one more thing I
wanted to say about this Binance hack because

01:30:10.400 --> 01:30:13.880
it accepts me very much Coffeezilla did a
breakdown

01:30:12.220 --> 01:30:14.280
All right are you a crypto guy?

01:30:14.560 --> 01:30:16.660
I don't know I can't remember. Are you into
crypto?

01:30:16.160 --> 01:30:19.140
no I'm the guy who watches from the sideline and
grinds his teeth

01:30:16.880 --> 01:30:17.020
No I'm

01:30:19.580 --> 01:30:22.800
Yeah that's me that's me right there yeah I'm
the same.

01:30:22.820 --> 01:30:28.740
because what is the most crazy thing about this
Binance hack that Coffeezilla covered so I

01:30:28.740 --> 01:30:33.520
read quite a few breakdowns of this hack you
know apparently the DevOps engineer was lured by

01:30:33.520 --> 01:30:39.240
interviews and they sent an assignment to do the
assignment on his laptop which included the

01:30:37.800 --> 01:30:41.240
Oh my goodness, I've heard of things like this,
yeah.

01:30:39.240 --> 01:30:44.880
code that compromised his laptop and he had the
credentials to basically talk to the service

01:30:44.880 --> 01:30:46.820
and hit the hot wallet

01:30:45.520 --> 01:30:49.480
So someone was interviewing a Binance employee,
oh my god.

01:30:49.680 --> 01:30:54.860
yes but what was the interesting side that
Coffeezilla covered not the way that they got in

01:30:54.860 --> 01:30:59.700
and how they exfiltrated all the money but what
happened right after so it was about what was

01:30:59.700 --> 01:31:05.680
it 6 billion 5.6 billion of money and
Coffeezilla identified that immediately Binance
they

01:31:05.680 --> 01:31:10.560
talk with these there's two types of crypto
apparently like you have the ones that are uh
you

01:31:10.560 --> 01:31:13.920
know the ledger rules and then you have another
ones where it's it's not really decentralized

01:31:13.920 --> 01:31:19.040
you still have an authority and they're able to
freeze the crypto so the hackers they got a

01:31:19.040 --> 01:31:23.520
handful of of like a mix of coins whatever was
in the hot wallet right they empty it so the

01:31:23.520 --> 01:31:31.020
hot wallet is the one that's online that's there
for doing you know the transactions so

01:31:26.660 --> 01:31:29.780
Yeah, yeah, how the hell does it have so much
money? But yeah, you're going to get to it, I'm

01:31:29.780 --> 01:31:29.940
sure.

01:31:31.020 --> 01:31:36.160
no I I don't know why it had so much I guess
because Binance is so big and handles so much

01:31:36.160 --> 01:31:40.240
traffic and transactions that but its hot wallet
probably has so much.

01:31:38.720 --> 01:31:40.640
Billions! Okay, okay.

01:31:40.640 --> 01:31:41.800
Binance is huge.

01:31:43.220 --> 01:31:49.240
So what I was going to say that Coffeezilla
highlighted, which was, I think, kind of very

01:31:49.240 --> 01:31:55.340
farcical, is that of those 5.6 billion, 2.5
billion was frozen by those

01:31:55.340 --> 01:31:59.400
coins that are, I don't know, freezable by an
authority.

01:31:59.600 --> 01:32:05.420
So what's the point of decentralized currency if
you have an authority that can freeze it?

01:32:06.440 --> 01:32:10.540
Well, it's a guardrail. Anyway, listen, I have
to take my kids to school.

01:32:10.960 --> 01:32:13.540
I don't really know too much about it. I'm not
an authority.

01:32:13.040 --> 01:32:18.240
watch the Coffeezilla video of the Binance hack
and how the hackers only got 2.5 billion.

01:32:16.400 --> 01:32:19.920
Okay, we can talk about it. Anyway, good
catching up with you, Vincent.

01:32:21.660 --> 01:32:25.080
Oh, we're in for the right. That's just, what,
two weeks of hacks.

01:32:21.960 --> 01:32:22.680
We'll catch up.

01:32:25.460 --> 01:32:27.600
Okay, yeah, I mean, we're going to have a lot.

01:32:27.680 --> 01:32:34.120
We We should probably change this to the AI hack
or something like that or the

01:32:34.120 --> 01:32:37.440
security podcast or something like that. I mean
it's going to get that way isn't it.

01:32:38.060 --> 01:32:39.060
See you, bye.