WEBVTT

00:00:18.766 --> 00:00:20.007
Welcome to Community & Code,

00:00:20.027 --> 00:00:21.407
the podcast about the

00:00:21.468 --> 00:00:23.549
people behind the commits.

00:00:23.971 --> 00:00:25.791
I'm here today with Brendan O'Leary.

00:00:26.151 --> 00:00:27.634
He is a developer relations

00:00:27.753 --> 00:00:29.255
engineer at Kilo Code.

00:00:29.875 --> 00:00:30.836
Welcome to the show, Brendan.

00:00:30.875 --> 00:00:31.617
Tell us about yourself.

00:00:32.442 --> 00:00:33.882
Yeah, thanks for having me, Chris.

00:00:34.463 --> 00:00:35.283
Glad to be here.

00:00:36.004 --> 00:00:36.664
Yeah, I'm Brendan.

00:00:37.344 --> 00:00:38.524
I work in developer relations.

00:00:39.045 --> 00:00:39.824
I've been in developer

00:00:39.844 --> 00:00:41.425
relations for a long time,

00:00:41.525 --> 00:00:43.786
but longer than that,

00:00:43.846 --> 00:00:45.086
just kind of been involved

00:00:45.246 --> 00:00:46.827
in coding and shipping code

00:00:46.867 --> 00:00:48.586
to production and kind of

00:00:48.628 --> 00:00:49.987
everything that goes along with that.

00:00:50.067 --> 00:00:51.648
You know, I've worked in product,

00:00:51.707 --> 00:00:54.168
I've worked in engineering, and then,

00:00:54.289 --> 00:00:54.469
yeah,

00:00:54.509 --> 00:00:56.990
kind of found my passion in

00:00:57.250 --> 00:00:58.670
developer relations and

00:00:58.689 --> 00:01:00.731
helping other developers ship code, so...

00:01:01.411 --> 00:01:03.914
So I've been in DevRel for

00:01:04.054 --> 00:01:04.835
almost a full year,

00:01:04.875 --> 00:01:06.115
although I feel like I mean,

00:01:06.155 --> 00:01:07.317
I've said this before, too.

00:01:07.536 --> 00:01:09.820
I was doing forms of

00:01:09.880 --> 00:01:11.240
developer relations before

00:01:11.280 --> 00:01:13.523
I was actually like had that as a title.

00:01:14.524 --> 00:01:15.985
And also one of the things

00:01:16.006 --> 00:01:17.286
that I've observed since

00:01:17.347 --> 00:01:18.508
kind of being in the space

00:01:18.548 --> 00:01:19.628
more and meeting other

00:01:19.668 --> 00:01:20.530
people that are doing

00:01:20.590 --> 00:01:21.892
DevRel is that it means a

00:01:21.951 --> 00:01:23.653
lot of different things depending on

00:01:24.334 --> 00:01:25.718
the organization, the team,

00:01:26.019 --> 00:01:28.686
where it sits in the org structure.

00:01:29.509 --> 00:01:30.350
What does developer

00:01:30.370 --> 00:01:32.537
relations look like for you at KiloCode?

00:01:33.466 --> 00:01:34.427
Yeah, that's a great question.

00:01:34.447 --> 00:01:34.688
And yeah,

00:01:34.748 --> 00:01:36.088
where it sits and all of that is

00:01:36.129 --> 00:01:38.551
like the best part of

00:01:38.631 --> 00:01:39.772
developer relations is bike

00:01:39.792 --> 00:01:41.192
shedding what developer relations is,

00:01:41.253 --> 00:01:41.474
right?

00:01:44.016 --> 00:01:47.018
Yeah, so Aquilo Code, you know,

00:01:47.037 --> 00:01:48.599
we're really focused on

00:01:49.980 --> 00:01:52.022
delivering an agentic AI.

00:01:52.061 --> 00:01:54.063
It's an open source application.

00:01:54.623 --> 00:01:55.746
a VS Code extension that

00:01:55.787 --> 00:01:59.373
helps folks use all the

00:01:59.414 --> 00:02:00.335
different AI models that

00:02:00.376 --> 00:02:01.900
are out there to try and

00:02:02.921 --> 00:02:03.582
supplement what they're

00:02:03.623 --> 00:02:04.424
doing as an engineer.

00:02:05.603 --> 00:02:06.504
And so developer relations

00:02:06.525 --> 00:02:09.628
was an early focus for us

00:02:09.688 --> 00:02:11.730
because all of our customers,

00:02:11.790 --> 00:02:13.131
all of our users are developers.

00:02:14.171 --> 00:02:17.314
And I think that what I

00:02:17.355 --> 00:02:18.495
always say about developer

00:02:18.516 --> 00:02:20.037
relations in general is

00:02:20.078 --> 00:02:22.280
like a lot of folks like to say, oh,

00:02:22.299 --> 00:02:22.799
it should sit in

00:02:22.840 --> 00:02:23.620
engineering or it should

00:02:23.641 --> 00:02:24.221
sit in marketing.

00:02:25.262 --> 00:02:26.763
And at the end of the day,

00:02:27.383 --> 00:02:28.663
like some developer

00:02:28.683 --> 00:02:29.944
relations folks get mad at me about this,

00:02:30.025 --> 00:02:30.825
but at the end of the day,

00:02:30.846 --> 00:02:32.206
it is developer marketing, right?

00:02:32.227 --> 00:02:33.307
Like that is what it is like.

00:02:33.627 --> 00:02:34.487
Like, obviously,

00:02:34.848 --> 00:02:35.588
I think it's obvious to

00:02:35.649 --> 00:02:37.770
everyone or people that are in, you know,

00:02:38.651 --> 00:02:39.771
software that developers

00:02:39.831 --> 00:02:40.932
don't like being marketed

00:02:41.152 --> 00:02:42.593
to through traditional channels.

00:02:43.562 --> 00:02:45.623
um but would rather be able

00:02:45.644 --> 00:02:47.066
to try something put their

00:02:47.126 --> 00:02:49.008
hands on it understand how

00:02:49.048 --> 00:02:50.949
it works maybe even inspect

00:02:50.969 --> 00:02:51.971
the code behind it right

00:02:51.991 --> 00:02:52.912
that's why open source is

00:02:52.951 --> 00:02:56.294
such a popular methodology

00:02:56.375 --> 00:02:58.456
and so in order to enable

00:02:58.477 --> 00:02:59.578
that I think you need

00:03:00.318 --> 00:03:01.120
a human connection with

00:03:01.139 --> 00:03:02.080
people and so that's what

00:03:02.100 --> 00:03:03.141
we are really focused on is

00:03:03.162 --> 00:03:04.021
how can we connect with

00:03:04.062 --> 00:03:05.544
people and make them

00:03:05.584 --> 00:03:06.905
successful with the tool uh

00:03:07.945 --> 00:03:09.407
and then kind of the rest

00:03:09.447 --> 00:03:10.427
of it is the score takes

00:03:10.448 --> 00:03:11.468
care of itself right like

00:03:11.549 --> 00:03:12.669
all of all of the revenue

00:03:12.689 --> 00:03:13.350
and all the other things

00:03:13.371 --> 00:03:15.192
can follow that uh but if

00:03:15.211 --> 00:03:16.092
you don't execute on that

00:03:16.133 --> 00:03:17.895
well you're just constantly

00:03:17.935 --> 00:03:19.757
fighting uphill uh if

00:03:19.776 --> 00:03:20.796
you're if you want to you

00:03:20.817 --> 00:03:21.858
know try and make something

00:03:21.918 --> 00:03:23.460
work uh with developers

00:03:24.823 --> 00:03:29.324
So tell me about what Kilo

00:03:29.344 --> 00:03:31.746
code is or does and about

00:03:31.907 --> 00:03:36.848
sort of the AI integration

00:03:36.908 --> 00:03:39.090
that you're building or

00:03:39.110 --> 00:03:40.510
working on or sharing with other people.

00:03:40.531 --> 00:03:41.091
Sure.

00:03:41.151 --> 00:03:41.350
Yeah.

00:03:41.411 --> 00:03:42.731
Yeah, that would be great.

00:03:42.771 --> 00:03:43.771
I think, yeah,

00:03:43.792 --> 00:03:44.432
so I'll kind of give you

00:03:44.453 --> 00:03:45.353
like maybe a little bit of

00:03:45.372 --> 00:03:46.114
my path to there,

00:03:46.133 --> 00:03:46.954
which I think then answers

00:03:48.555 --> 00:03:49.474
what we're doing and why I

00:03:49.514 --> 00:03:51.336
think it is a way of

00:03:51.376 --> 00:03:52.276
attacking this problem.

00:03:53.176 --> 00:03:53.697
You know,

00:03:53.736 --> 00:03:56.217
I think there's a lot of like AI

00:03:56.258 --> 00:03:57.899
maximalists out there that will say,

00:03:58.378 --> 00:03:59.598
you know, for the past two or three years,

00:03:59.618 --> 00:04:00.179
we've said, oh,

00:04:00.199 --> 00:04:01.060
we're six months away from

00:04:01.080 --> 00:04:02.120
the never being an software

00:04:02.139 --> 00:04:02.840
engineer anymore.

00:04:04.580 --> 00:04:05.480
And then there's a lot of,

00:04:05.621 --> 00:04:07.681
I think also skepticism from the,

00:04:07.762 --> 00:04:08.042
you know,

00:04:08.062 --> 00:04:10.103
like rightly earned skepticism

00:04:10.163 --> 00:04:11.682
from developers and

00:04:11.723 --> 00:04:14.663
software engineers who have seen either,

00:04:14.944 --> 00:04:15.205
you know,

00:04:15.504 --> 00:04:18.205
the slop that early AI models

00:04:18.246 --> 00:04:19.066
produced or,

00:04:19.945 --> 00:04:21.987
or the kind of vibe coding

00:04:22.448 --> 00:04:23.528
methodology that has not

00:04:23.549 --> 00:04:24.568
been super successful for

00:04:24.609 --> 00:04:25.188
people that don't

00:04:25.229 --> 00:04:26.389
understand how the code works.

00:04:28.031 --> 00:04:28.872
So I would say,

00:04:29.151 --> 00:04:31.212
I think that the reality is

00:04:31.272 --> 00:04:32.032
somewhere in the middle.

00:04:33.974 --> 00:04:34.214
I say,

00:04:34.235 --> 00:04:35.415
I don't think that AI is going to

00:04:35.435 --> 00:04:36.435
replace software engineers,

00:04:36.475 --> 00:04:37.355
but I do think a software

00:04:37.377 --> 00:04:38.737
engineer that uses AI is

00:04:38.757 --> 00:04:39.997
going to replace software engineers.

00:04:41.538 --> 00:04:43.879
I think it's a tool that

00:04:44.000 --> 00:04:46.000
represents probably, yes, a step change,

00:04:46.041 --> 00:04:46.302
but again,

00:04:46.341 --> 00:04:46.922
I don't even need to be a

00:04:46.982 --> 00:04:47.983
maximalist to say that

00:04:48.702 --> 00:04:50.105
If you look at statistics

00:04:50.125 --> 00:04:51.367
from Stack Overflow,

00:04:52.228 --> 00:04:53.329
if you look at what people

00:04:53.348 --> 00:04:54.110
are talking about in the

00:04:54.350 --> 00:04:57.074
SEO world just in general

00:04:57.093 --> 00:04:59.076
with the rise of models

00:04:59.476 --> 00:05:00.699
like ChatGPT that are

00:05:01.418 --> 00:05:02.461
making massive traction

00:05:02.500 --> 00:05:03.903
throughout the world,

00:05:03.942 --> 00:05:05.144
not just in software engineering.

00:05:06.004 --> 00:05:07.887
I think the reality is that

00:05:07.927 --> 00:05:11.533
if you're an engineer today that's saying,

00:05:11.574 --> 00:05:13.817
yeah, but AI, it's a fad, it'll pass.

00:05:14.377 --> 00:05:15.620
That's like somebody saying, oh,

00:05:15.639 --> 00:05:17.281
Stack Overflow, that's a fad, it'll pass.

00:05:17.302 --> 00:05:18.624
Or, oh, Google, yeah, that's a fad,

00:05:18.644 --> 00:05:20.887
it'll pass at those times.

00:05:22.101 --> 00:05:23.562
Is it overhyped and probably

00:05:23.583 --> 00:05:26.843
the latest overhyped

00:05:27.103 --> 00:05:28.064
venture capital bubble?

00:05:28.125 --> 00:05:28.524
Sure.

00:05:29.884 --> 00:05:31.125
But is it going to just go

00:05:31.185 --> 00:05:33.406
away one day because of that?

00:05:33.447 --> 00:05:33.627
No.

00:05:33.966 --> 00:05:35.307
There was a time when people

00:05:35.327 --> 00:05:36.187
thought of the Internet that way,

00:05:36.208 --> 00:05:38.129
or they thought of the web

00:05:38.189 --> 00:05:39.348
and dot coms that way.

00:05:39.369 --> 00:05:40.470
And yeah,

00:05:40.490 --> 00:05:41.889
there had to be this equilibrium

00:05:42.170 --> 00:05:43.091
to find reality.

00:05:43.151 --> 00:05:44.891
But the reality now is that

00:05:44.911 --> 00:05:46.012
we all have the Internet in

00:05:46.031 --> 00:05:47.072
our pocket and at our

00:05:47.093 --> 00:05:48.052
fingertips every minute.

00:05:49.122 --> 00:05:49.803
I think that's going to be

00:05:49.824 --> 00:05:50.504
the rally for AI.

00:05:50.564 --> 00:05:50.944
So sorry,

00:05:50.963 --> 00:05:52.225
that's like a very philosophical

00:05:52.245 --> 00:05:54.685
way of coming to, what does Kilo Code do?

00:05:56.247 --> 00:05:57.047
I've got questions,

00:05:57.267 --> 00:05:58.168
but we'll come back to them.

00:05:58.528 --> 00:06:01.110
Yeah.

00:06:01.129 --> 00:06:03.170
So yeah, Kilo Code lives in VS Code.

00:06:03.190 --> 00:06:04.812
So it's a VS Code extension.

00:06:06.572 --> 00:06:09.293
And then we just interface to the models,

00:06:09.334 --> 00:06:09.494
right?

00:06:09.514 --> 00:06:10.615
So like we're not inventing

00:06:11.175 --> 00:06:12.776
any foundational model.

00:06:13.735 --> 00:06:14.916
But we allow you to say, oh,

00:06:14.978 --> 00:06:16.778
I want to connect to Sonnet

00:06:17.240 --> 00:06:20.603
four or I want to connect to Gemini.

00:06:20.622 --> 00:06:20.802
Right.

00:06:20.822 --> 00:06:22.884
Like what model do I want to connect to?

00:06:23.886 --> 00:06:26.367
And then our tool has a

00:06:26.408 --> 00:06:27.389
number of agents that can

00:06:27.428 --> 00:06:28.689
then take action.

00:06:28.730 --> 00:06:28.889
Right.

00:06:28.910 --> 00:06:30.192
Like an orchestrator that

00:06:30.612 --> 00:06:32.653
that take plans out with the agents,

00:06:32.733 --> 00:06:34.415
a architect that doesn't

00:06:34.456 --> 00:06:35.315
just code right away.

00:06:35.757 --> 00:06:37.298
It plans out the architecture,

00:06:37.317 --> 00:06:37.798
what it's going to do.

00:06:38.319 --> 00:06:39.259
a code agent that codes

00:06:39.300 --> 00:06:40.500
right and so it's kind of

00:06:40.540 --> 00:06:42.120
this different world where

00:06:42.620 --> 00:06:43.701
again agent is like an

00:06:43.862 --> 00:06:45.122
overloaded term right now

00:06:45.142 --> 00:06:45.723
but it's this different

00:06:45.742 --> 00:06:47.064
world where you don't just

00:06:47.084 --> 00:06:48.663
throw code slop at a person

00:06:48.704 --> 00:06:50.685
from from a prompt but you

00:06:50.725 --> 00:06:52.346
actually like pair program

00:06:52.365 --> 00:06:53.586
with the ai that's that's

00:06:53.607 --> 00:06:55.387
the idea behind behind kilo

00:06:56.809 --> 00:06:59.790
So I've been using Klein in

00:06:59.810 --> 00:07:02.151
VS Code recently,

00:07:03.250 --> 00:07:07.192
instead of GitHub Copilot,

00:07:07.211 --> 00:07:08.192
which is what everyone

00:07:08.233 --> 00:07:09.932
obviously uses and talks about.

00:07:10.552 --> 00:07:12.353
So and Klein has a

00:07:12.613 --> 00:07:14.394
plan mode and an act mode.

00:07:14.435 --> 00:07:16.096
And plan mode is kind of like the,

00:07:16.115 --> 00:07:16.975
we're going to talk about

00:07:17.016 --> 00:07:18.516
the thing and come up with a solution.

00:07:19.076 --> 00:07:21.237
And then the act mode is like, okay,

00:07:21.637 --> 00:07:22.458
I've done this thing.

00:07:22.478 --> 00:07:23.639
Do you want me to do the thing?

00:07:23.658 --> 00:07:24.338
Because if you do,

00:07:24.379 --> 00:07:25.740
then click over to act mode,

00:07:25.759 --> 00:07:26.660
and then I'll do the thing.

00:07:26.699 --> 00:07:27.880
And then you can say, yeah, okay.

00:07:28.281 --> 00:07:29.221
So it sounds like you have a

00:07:29.321 --> 00:07:30.341
similar sort of approach,

00:07:30.362 --> 00:07:31.322
but maybe like another

00:07:31.382 --> 00:07:32.723
layer in between kind of

00:07:33.853 --> 00:07:34.014
Yep.

00:07:34.194 --> 00:07:37.995
And the fun fact there is we

00:07:38.115 --> 00:07:39.596
actually forked a project

00:07:39.617 --> 00:07:41.437
called root code when we

00:07:41.478 --> 00:07:42.238
started the company,

00:07:42.879 --> 00:07:44.339
which the fun fact there is

00:07:44.399 --> 00:07:45.701
root code is a fork of Klein.

00:07:45.761 --> 00:07:47.541
So so there's there's a lot

00:07:47.661 --> 00:07:48.461
of there's a lot of

00:07:48.721 --> 00:07:49.923
relation between those

00:07:50.023 --> 00:07:50.963
those three packages.

00:07:51.564 --> 00:07:53.485
yeah um and we're actually

00:07:53.545 --> 00:07:54.404
pretty big contributors

00:07:54.444 --> 00:07:55.764
back to rue because of that

00:07:56.446 --> 00:07:57.906
ruin ruin klein kind of

00:07:57.966 --> 00:08:00.247
split six months ago or so

00:08:00.387 --> 00:08:01.547
like in terms of like kind

00:08:01.567 --> 00:08:02.288
of the direction they're

00:08:02.408 --> 00:08:03.987
going as projects but

00:08:04.007 --> 00:08:04.988
they're all open source

00:08:05.028 --> 00:08:06.528
right and so I think that's

00:08:06.569 --> 00:08:08.889
actually why we see now vs

00:08:08.930 --> 00:08:10.050
code reacting and saying oh

00:08:10.069 --> 00:08:10.629
we're going to make

00:08:10.990 --> 00:08:12.411
co-pilot chat open source

00:08:12.430 --> 00:08:13.190
right they just recently

00:08:13.230 --> 00:08:15.031
did that um in fact the

00:08:15.132 --> 00:08:17.452
code was just released like last week um

00:08:18.392 --> 00:08:20.394
I think that's a reaction to, oh,

00:08:21.654 --> 00:08:23.156
these open source tools are

00:08:23.196 --> 00:08:24.576
really gaining a lot of traction because,

00:08:24.817 --> 00:08:25.057
again,

00:08:25.076 --> 00:08:25.877
developers want to be able to

00:08:25.898 --> 00:08:26.677
inspect what they're doing.

00:08:26.838 --> 00:08:27.259
They don't want to

00:08:27.338 --> 00:08:29.220
necessarily just send a

00:08:29.259 --> 00:08:30.401
chat into the ether and not

00:08:30.440 --> 00:08:31.500
understand what happens back.

00:08:32.581 --> 00:08:32.982
So yes,

00:08:33.923 --> 00:08:37.004
it is not that different in kind of

00:08:37.045 --> 00:08:37.745
the methodology,

00:08:37.925 --> 00:08:41.247
but the ways that we're our

00:08:41.286 --> 00:08:42.427
thought is that it is a

00:08:42.468 --> 00:08:43.668
very fragmented space.

00:08:44.269 --> 00:08:46.651
And what we want to do is be

00:08:46.690 --> 00:08:48.271
the superset of those tools, right?

00:08:48.312 --> 00:08:49.372
Because again,

00:08:49.413 --> 00:08:52.674
a thing that that today again,

00:08:52.715 --> 00:08:53.836
this will be out of date by

00:08:53.855 --> 00:08:55.657
the time someone listens to it.

00:08:55.777 --> 00:08:56.738
Yeah, exactly.

00:08:57.677 --> 00:08:58.778
But the thing today that

00:08:58.798 --> 00:08:59.519
like a root code or a

00:08:59.558 --> 00:09:00.600
client doesn't have is any

00:09:00.639 --> 00:09:01.921
like tab completion, right?

00:09:01.941 --> 00:09:02.741
Which is what I think.

00:09:03.841 --> 00:09:04.001
You know,

00:09:04.062 --> 00:09:05.783
Cursor has a big share in the

00:09:05.823 --> 00:09:06.884
market because of so that's

00:09:06.984 --> 00:09:08.164
something we recently added.

00:09:08.184 --> 00:09:09.885
And so we want to build a

00:09:09.946 --> 00:09:12.888
superset of all the features to say, like,

00:09:12.969 --> 00:09:14.328
not have to have seventeen

00:09:14.369 --> 00:09:16.571
different extensions to do do stuff.

00:09:17.491 --> 00:09:20.352
But conceptually, this idea of

00:09:21.313 --> 00:09:21.615
you know,

00:09:23.035 --> 00:09:24.158
an agent that has kind of a

00:09:24.197 --> 00:09:25.359
different set of rules for

00:09:25.418 --> 00:09:27.260
it and a different kind of

00:09:27.581 --> 00:09:31.326
what are you about is where

00:09:31.745 --> 00:09:34.009
those three kind of family

00:09:34.469 --> 00:09:35.650
of extensions came from.

00:09:36.672 --> 00:09:37.673
And the beauty of like Aru

00:09:37.712 --> 00:09:38.514
or us is like you can

00:09:38.553 --> 00:09:39.634
actually create your own

00:09:42.725 --> 00:09:44.267
And so like if you look at

00:09:44.287 --> 00:09:46.568
our documentation or ruse documentation,

00:09:47.330 --> 00:09:49.032
there's a documentation writer mode,

00:09:49.272 --> 00:09:49.412
right?

00:09:49.432 --> 00:09:51.173
That is very focused on like, hey,

00:09:51.594 --> 00:09:52.794
I'm going to give you a bunch of input,

00:09:52.835 --> 00:09:53.596
but what your output is

00:09:53.615 --> 00:09:55.278
going to be isn't code.

00:09:55.317 --> 00:09:56.538
It's going to be well

00:09:56.558 --> 00:09:58.160
formatted documentation with yes,

00:09:58.541 --> 00:10:00.182
snippet examples, but not, you know,

00:10:00.663 --> 00:10:01.244
just code.

00:10:02.664 --> 00:10:03.846
So, yeah.

00:10:03.905 --> 00:10:04.306
Interesting.

00:10:05.407 --> 00:10:06.988
I mean,

00:10:07.067 --> 00:10:09.669
I've talked about AI quite a bit on

00:10:09.710 --> 00:10:10.250
this show,

00:10:10.970 --> 00:10:14.653
and there's lots of stuff happening.

00:10:14.673 --> 00:10:18.696
I mean, it's the new shit, right?

00:10:19.056 --> 00:10:21.177
What brought you or

00:10:21.337 --> 00:10:24.799
interested you in this space initially?

00:10:26.760 --> 00:10:29.003
What about AI is...

00:10:31.668 --> 00:10:33.389
is most exciting or

00:10:33.490 --> 00:10:36.510
interesting yeah uh I would

00:10:36.551 --> 00:10:37.551
say I wasn't interested at

00:10:37.571 --> 00:10:38.672
all to begin with which is

00:10:38.731 --> 00:10:39.951
I think great that's what I

00:10:39.991 --> 00:10:43.014
like to hear um I never

00:10:43.053 --> 00:10:43.933
wanted to learn what

00:10:44.073 --> 00:10:45.434
different models meant I

00:10:45.475 --> 00:10:46.975
never wanted to learn like

00:10:48.155 --> 00:10:49.416
I I was not I was very much

00:10:49.475 --> 00:10:51.817
not a maximalist I I was

00:10:51.856 --> 00:10:52.618
like one of those people

00:10:52.638 --> 00:10:53.798
that early on was like I'm

00:10:53.837 --> 00:10:55.139
an elder millennial so I was like

00:10:55.698 --> 00:10:57.320
It's not AI, it's machine learning,

00:10:57.360 --> 00:10:58.159
you know, for certain things.

00:10:58.240 --> 00:11:00.160
And like kind of really like

00:11:01.042 --> 00:11:03.082
an LLM is just makes you average.

00:11:03.143 --> 00:11:03.842
Like you don't want to you

00:11:03.863 --> 00:11:04.864
want to be better than average.

00:11:04.884 --> 00:11:05.604
Like, you know,

00:11:05.624 --> 00:11:06.605
I've been kind of that kind

00:11:06.644 --> 00:11:09.746
of like naysayer for some time.

00:11:09.767 --> 00:11:10.767
And I think that's probably

00:11:10.807 --> 00:11:13.089
a big chunk of software engineers, again,

00:11:13.109 --> 00:11:14.408
because they tried it early.

00:11:14.828 --> 00:11:15.909
They're early adopters, right?

00:11:15.970 --> 00:11:17.770
Us software engineers are early adopters.

00:11:18.250 --> 00:11:19.251
And it was really bad.

00:11:19.591 --> 00:11:20.753
Like even twelve

00:11:21.754 --> 00:11:23.236
you know, fifteen, eighteen months ago.

00:11:24.017 --> 00:11:24.878
The thing I the thing I

00:11:24.937 --> 00:11:26.320
point to there for folks,

00:11:26.360 --> 00:11:27.421
if I'm trying to convince them,

00:11:27.480 --> 00:11:28.201
but then I could talk about

00:11:28.221 --> 00:11:30.024
what convinced me is like

00:11:30.043 --> 00:11:32.245
if you look at the classic, you know,

00:11:32.326 --> 00:11:34.227
video generation prompt,

00:11:34.268 --> 00:11:37.650
which is Will Smith eating spaghetti like,

00:11:37.912 --> 00:11:39.833
yes, like twelve, eighteen months ago,

00:11:39.854 --> 00:11:40.875
it was horrible.

00:11:40.955 --> 00:11:42.817
It was like like fear,

00:11:42.836 --> 00:11:44.498
like like nightmare inducing.

00:11:45.538 --> 00:11:45.899
Um,

00:11:45.958 --> 00:11:47.820
and then if you look at the like V VO

00:11:47.840 --> 00:11:50.461
three or whatever, like from Google, it's,

00:11:50.861 --> 00:11:53.503
it's pretty realistic and kind of crazy.

00:11:53.582 --> 00:11:55.903
And I think the models have

00:11:55.943 --> 00:11:58.644
done just as big of a jump, uh,

00:11:58.684 --> 00:12:01.366
in what they're able to do with code, uh,

00:12:01.407 --> 00:12:02.206
and what they're able to do

00:12:02.846 --> 00:12:04.388
acting again as an agent.

00:12:05.408 --> 00:12:05.668
Um,

00:12:05.708 --> 00:12:07.289
which to me just means working with the

00:12:07.309 --> 00:12:08.009
human, not just.

00:12:09.871 --> 00:12:11.552
know spitting out the slop

00:12:11.952 --> 00:12:14.232
um that they're good at and

00:12:14.513 --> 00:12:15.394
and so what kind of

00:12:15.433 --> 00:12:17.913
convinced me again is what

00:12:17.933 --> 00:12:18.673
I was talking about earlier

00:12:18.714 --> 00:12:19.495
is like I kind of got

00:12:19.514 --> 00:12:20.634
convinced that like okay

00:12:21.174 --> 00:12:23.235
once I started seeing what

00:12:23.375 --> 00:12:24.416
ai really looked like again

00:12:24.436 --> 00:12:25.436
because again I had tried

00:12:25.456 --> 00:12:27.076
it a while ago I was using

00:12:27.116 --> 00:12:27.956
cursor and I was hitting

00:12:28.017 --> 00:12:29.356
tab a lot which was helpful

00:12:29.917 --> 00:12:31.118
but was it that much

00:12:31.158 --> 00:12:32.177
different than autocomplete

00:12:32.197 --> 00:12:35.239
right or you know language model yeah

00:12:36.919 --> 00:12:38.301
But what kind of convinced me is again,

00:12:38.360 --> 00:12:40.442
I think that once I saw the

00:12:40.482 --> 00:12:42.344
speed that you can get with,

00:12:44.145 --> 00:12:45.287
like when you understand

00:12:45.326 --> 00:12:46.827
the context and give it context,

00:12:46.908 --> 00:12:48.028
how much of a great pair

00:12:48.089 --> 00:12:50.129
program that becomes, it's incredible.

00:12:50.149 --> 00:12:52.913
Like just even on my portfolio site,

00:12:52.972 --> 00:12:53.133
right?

00:12:53.153 --> 00:12:53.373
It's like,

00:12:53.393 --> 00:12:54.994
if I have context to something and I say,

00:12:55.014 --> 00:12:56.836
hey, like see this, I want another model.

00:12:56.855 --> 00:12:58.417
Like this, here's an example.

00:12:58.616 --> 00:13:01.239
And my portfolio sites like is public too.

00:13:01.259 --> 00:13:01.960
So people could look.

00:13:02.600 --> 00:13:03.721
Like, I have a portfolio site.

00:13:03.740 --> 00:13:04.962
It's got, like, talks.

00:13:05.062 --> 00:13:05.964
This is very DevRel-y.

00:13:06.423 --> 00:13:07.345
It's got talks.

00:13:07.365 --> 00:13:08.586
It's got a blog.

00:13:09.126 --> 00:13:09.788
And then my wife and I

00:13:10.067 --> 00:13:11.068
recently went on a trip to Europe,

00:13:11.089 --> 00:13:11.570
and she's like,

00:13:11.590 --> 00:13:13.010
I really wish we just have, like, one.

00:13:13.731 --> 00:13:15.073
We were taking a journal, like,

00:13:15.114 --> 00:13:16.475
actually an Apple's journaling app,

00:13:16.495 --> 00:13:17.576
which you can't share easily,

00:13:17.596 --> 00:13:18.977
it turns out, which makes sense.

00:13:19.739 --> 00:13:20.219
But she's like, oh,

00:13:20.239 --> 00:13:20.980
I just wish we could have

00:13:21.000 --> 00:13:21.900
that just on a site that I

00:13:21.941 --> 00:13:22.461
could just send people.

00:13:22.522 --> 00:13:23.982
And I was like, oh, I could do that.

00:13:24.792 --> 00:13:26.232
And so I said to it, you know,

00:13:26.472 --> 00:13:27.953
Tequila or I mean, again,

00:13:27.994 --> 00:13:29.274
to many of these agents,

00:13:29.294 --> 00:13:30.196
it would work the same.

00:13:31.015 --> 00:13:32.756
I was like, hey, like, see this blog,

00:13:32.797 --> 00:13:33.918
see these talks like I want that,

00:13:33.957 --> 00:13:36.159
but model it for like a trip diary.

00:13:36.940 --> 00:13:38.721
And like when it has that context,

00:13:38.841 --> 00:13:40.562
it's it's like I could have

00:13:40.602 --> 00:13:42.264
coded one hundred percent of what it did,

00:13:42.884 --> 00:13:44.865
but it took it literally thirty seconds.

00:13:46.389 --> 00:13:47.210
And it would have taken me

00:13:47.269 --> 00:13:48.691
an hour or two and it would

00:13:48.711 --> 00:13:50.251
have probably been the same.

00:13:51.533 --> 00:13:55.274
And so once I kind of learned how, again,

00:13:55.335 --> 00:13:55.914
I think I was a big

00:13:55.955 --> 00:13:56.796
naysayer about was this

00:13:57.015 --> 00:13:57.996
concept of prompt engineering,

00:13:58.037 --> 00:13:59.778
but it kind of is a thing.

00:13:59.798 --> 00:14:01.018
It's like this context

00:14:01.077 --> 00:14:01.739
engineering is what I'm

00:14:01.759 --> 00:14:02.740
hearing a lot of people talk about now.

00:14:02.759 --> 00:14:04.620
It's like when you can give it context,

00:14:05.140 --> 00:14:06.682
it is really fast at the

00:14:06.721 --> 00:14:08.222
things that I already know how to do.

00:14:10.376 --> 00:14:11.177
And it's really good at

00:14:11.216 --> 00:14:12.577
figuring it out too and

00:14:12.638 --> 00:14:13.879
working with me right like it.

00:14:13.938 --> 00:14:14.999
Now if I want to download

00:14:15.019 --> 00:14:15.840
like an open source tool

00:14:15.860 --> 00:14:17.139
that I'm interested in,

00:14:17.179 --> 00:14:18.201
the first thing I do is I

00:14:18.221 --> 00:14:19.260
go into ask mode which

00:14:19.341 --> 00:14:20.402
doesn't write any code.

00:14:20.562 --> 00:14:22.062
It's like I'm not allowed to

00:14:22.702 --> 00:14:23.423
and I say like hey,

00:14:23.462 --> 00:14:24.703
how does this project work right?

00:14:25.144 --> 00:14:26.024
And it'll show me it'll

00:14:26.264 --> 00:14:27.426
it'll write diagrams.

00:14:27.885 --> 00:14:29.947
It'll write you know whole like flow.

00:14:30.246 --> 00:14:31.908
It'll really dig in and

00:14:31.967 --> 00:14:33.389
explain because these these

00:14:33.428 --> 00:14:35.789
models now have enough context to take in,

00:14:36.269 --> 00:14:37.890
you know, a massive amount of code.

00:14:38.774 --> 00:14:39.715
It can really explain it a

00:14:39.754 --> 00:14:40.796
lot better and works as a

00:14:40.836 --> 00:14:43.356
pair programmer better than

00:14:43.437 --> 00:14:44.397
I ever expected it would.

00:14:46.116 --> 00:14:47.197
Yeah, I recently,

00:14:47.337 --> 00:14:48.957
I noticed in one of the

00:14:48.998 --> 00:14:52.620
most recent chat GPT updates,

00:14:52.779 --> 00:14:55.059
they've added integrations

00:14:55.200 --> 00:14:58.140
with like deeper

00:14:58.181 --> 00:14:59.922
integrations with like Google Drive.

00:15:00.785 --> 00:15:01.865
and also integrations with

00:15:01.966 --> 00:15:03.326
GitHub so that you can

00:15:03.386 --> 00:15:04.486
authenticate with your

00:15:04.846 --> 00:15:06.408
Google user or your GitHub user,

00:15:06.447 --> 00:15:08.708
and then you can feed it an

00:15:08.889 --> 00:15:10.729
actual project repository

00:15:10.849 --> 00:15:12.610
or a document or whatever,

00:15:12.750 --> 00:15:14.350
and then start asking it

00:15:14.390 --> 00:15:18.951
questions about the context within that.

00:15:19.111 --> 00:15:21.293
I half-formed a project with

00:15:22.253 --> 00:15:26.214
AI to build... I had this idea for...

00:15:27.955 --> 00:15:31.857
my perfect AI-generated playlist app.

00:15:34.440 --> 00:15:36.100
I go back to back in the

00:15:36.240 --> 00:15:37.741
olden days when Pandora was

00:15:37.802 --> 00:15:38.442
actually good.

00:15:39.323 --> 00:15:40.504
The reason why Pandora was

00:15:40.565 --> 00:15:41.565
good is because they hired

00:15:41.625 --> 00:15:43.706
human beings with music

00:15:44.346 --> 00:15:45.648
history and music knowledge

00:15:45.869 --> 00:15:47.370
backgrounds to create

00:15:47.929 --> 00:15:49.192
what they called phenomes

00:15:49.272 --> 00:15:50.552
that they applied to all

00:15:50.592 --> 00:15:51.693
the different songs and

00:15:51.734 --> 00:15:52.916
stuff that were in their database,

00:15:53.196 --> 00:15:54.136
so that when you said,

00:15:54.636 --> 00:15:56.259
I want a playlist or a

00:15:56.298 --> 00:15:57.821
station based on this song,

00:15:57.921 --> 00:15:58.981
it was not just doing

00:15:59.062 --> 00:16:01.544
genre-based connections like Spotify does,

00:16:01.585 --> 00:16:02.725
it was doing like, no,

00:16:02.785 --> 00:16:03.667
these are the component

00:16:03.687 --> 00:16:04.648
parts of that song,

00:16:04.967 --> 00:16:05.908
and here are other things

00:16:05.928 --> 00:16:07.230
that have similar component parts.

00:16:07.671 --> 00:16:08.471
And there's never been

00:16:08.511 --> 00:16:09.652
anything that's even remotely,

00:16:09.672 --> 00:16:10.972
and I'm like a music snob, right?

00:16:10.993 --> 00:16:12.673
I used to be a DJ.

00:16:12.714 --> 00:16:14.836
So like I was like,

00:16:14.975 --> 00:16:16.856
I'm all about like creating

00:16:16.897 --> 00:16:18.018
like the perfect playlist.

00:16:18.658 --> 00:16:19.479
And that was the only thing

00:16:19.499 --> 00:16:20.759
that really created things

00:16:20.779 --> 00:16:22.600
that felt like a thing that

00:16:22.681 --> 00:16:23.542
I could have done or like

00:16:23.601 --> 00:16:25.243
made sense in the same sort of way.

00:16:25.942 --> 00:16:28.985
I dislike how Spotify tries

00:16:29.004 --> 00:16:29.885
to do that and fails.

00:16:29.966 --> 00:16:31.807
Like it gets lumped into a

00:16:31.866 --> 00:16:33.869
very specific like rut and

00:16:33.889 --> 00:16:35.149
then that's all it ever really gives you.

00:16:35.830 --> 00:16:37.030
So I've started going to like,

00:16:37.971 --> 00:16:40.052
chat GPT and like asking it questions like,

00:16:40.371 --> 00:16:40.673
Hey,

00:16:40.812 --> 00:16:42.133
I'm kind of feeling like this and

00:16:42.173 --> 00:16:43.453
have a conversation with it.

00:16:43.774 --> 00:16:44.875
And then at the end, it's like, okay,

00:16:44.914 --> 00:16:45.955
here's, here's some stuff.

00:16:45.975 --> 00:16:46.674
And then I have to take

00:16:46.695 --> 00:16:47.916
those songs and put it into,

00:16:48.416 --> 00:16:49.136
into Spotify.

00:16:49.157 --> 00:16:50.096
And it's a manual process.

00:16:50.116 --> 00:16:52.138
So my ideal, my idea, uh,

00:16:52.158 --> 00:16:54.778
my ideal reality is I can just do that.

00:16:54.899 --> 00:16:55.960
And then it does the thing

00:16:56.000 --> 00:16:57.120
for me and it creates the

00:16:57.160 --> 00:16:58.041
playlist just so it

00:16:58.100 --> 00:16:59.081
authenticates with Spotify

00:16:59.101 --> 00:16:59.942
and it creates a playlist for me.

00:17:00.782 --> 00:17:02.503
So I had this half-formed

00:17:02.903 --> 00:17:04.543
thing with bits of code

00:17:04.563 --> 00:17:05.764
that don't connect to anything.

00:17:05.784 --> 00:17:07.326
There's some API stuff in there,

00:17:07.365 --> 00:17:10.187
but it's not fully... But I

00:17:10.246 --> 00:17:11.268
noticed that it had this feature.

00:17:11.288 --> 00:17:11.827
I'm like, okay,

00:17:12.227 --> 00:17:13.769
tell me what's missing and

00:17:13.808 --> 00:17:15.769
what's wrong and the things

00:17:15.789 --> 00:17:16.691
that don't connect.

00:17:17.030 --> 00:17:17.931
Because I know that this

00:17:17.990 --> 00:17:18.792
isn't going to work,

00:17:18.871 --> 00:17:20.913
but I also... I haven't really...

00:17:22.736 --> 00:17:23.977
I'm mostly just copying and

00:17:24.017 --> 00:17:25.057
pasting stuff at this point.

00:17:25.077 --> 00:17:26.137
I'm not even really thinking

00:17:26.178 --> 00:17:26.979
about the code yet.

00:17:27.219 --> 00:17:28.578
So just tell me the things

00:17:28.598 --> 00:17:29.759
that are there and not there.

00:17:29.799 --> 00:17:31.740
And it gave me a really,

00:17:31.861 --> 00:17:35.423
really detailed report on, it's got this,

00:17:35.442 --> 00:17:36.923
and this is good, and this stuff is bad,

00:17:36.963 --> 00:17:37.864
and this stuff is missing,

00:17:37.903 --> 00:17:38.483
and this stuff you would

00:17:38.523 --> 00:17:39.224
need to do before.

00:17:39.244 --> 00:17:41.365
It was pretty comprehensive.

00:17:42.526 --> 00:17:44.667
I could probably then use it

00:17:44.708 --> 00:17:46.989
to create a list of GitHub issues,

00:17:47.249 --> 00:17:48.348
things like task lists,

00:17:48.368 --> 00:17:49.230
things that need to be done.

00:17:50.269 --> 00:17:50.711
So yeah,

00:17:51.451 --> 00:17:55.614
giving it context of larger and

00:17:55.653 --> 00:17:56.493
larger swaths of

00:17:56.513 --> 00:17:59.115
information is really valuable.

00:17:59.977 --> 00:18:01.076
You were talking earlier about

00:18:02.565 --> 00:18:05.948
sort of how, you know,

00:18:06.367 --> 00:18:07.888
in the past people were using,

00:18:08.048 --> 00:18:10.070
would you just use Stack Overflow,

00:18:11.451 --> 00:18:14.614
how it's impacting the way that people,

00:18:14.815 --> 00:18:16.895
the way that developers get their,

00:18:17.076 --> 00:18:18.096
consume their information

00:18:18.136 --> 00:18:20.398
and also even having impacts on SEO.

00:18:21.900 --> 00:18:24.020
I've been seeing things

00:18:24.061 --> 00:18:26.623
about the decline of sites,

00:18:27.084 --> 00:18:29.585
the decline of clicks on

00:18:29.865 --> 00:18:31.105
the links in search engines,

00:18:31.145 --> 00:18:32.106
because as search engines

00:18:32.146 --> 00:18:33.048
integrate with AI,

00:18:33.448 --> 00:18:34.689
then you don't need to

00:18:34.729 --> 00:18:35.648
click on the things if AI

00:18:35.729 --> 00:18:36.509
is just going to give you

00:18:37.250 --> 00:18:38.431
the summary up at the top.

00:18:39.092 --> 00:18:41.073
So it's having SEO impacts there,

00:18:41.593 --> 00:18:45.175
but also the decline in use

00:18:45.215 --> 00:18:46.676
in things like Stack Overflow.

00:18:48.077 --> 00:18:51.660
And I wonder about, at a certain point,

00:18:52.500 --> 00:18:54.001
Because we're not at a point

00:18:54.362 --> 00:18:58.046
like at some point, maybe theoretically,

00:18:58.205 --> 00:18:58.546
I don't know.

00:18:58.625 --> 00:18:59.646
I don't I'm not sure.

00:18:59.666 --> 00:19:01.429
But at some point in theory,

00:19:01.788 --> 00:19:04.592
at least if we listen to, you know,

00:19:04.731 --> 00:19:06.173
all the all the gurus and

00:19:06.213 --> 00:19:07.275
CEOs of these companies.

00:19:09.157 --> 00:19:11.538
the belief is that we will

00:19:11.558 --> 00:19:13.640
have truly generative AI,

00:19:13.721 --> 00:19:14.780
where it can actually come

00:19:14.820 --> 00:19:15.882
up with its own,

00:19:16.622 --> 00:19:18.143
like answers to questions.

00:19:18.163 --> 00:19:18.942
But right now,

00:19:19.364 --> 00:19:20.943
it's just finding something

00:19:20.983 --> 00:19:23.965
that that exists in its data set,

00:19:24.425 --> 00:19:25.366
and matching it to

00:19:26.027 --> 00:19:27.086
the patterns that it's being

00:19:27.106 --> 00:19:27.926
given in the prompt.

00:19:28.988 --> 00:19:31.728
Which means that, for coding,

00:19:32.127 --> 00:19:33.528
it's looking at open source code,

00:19:33.568 --> 00:19:34.749
it's looking at the data

00:19:34.808 --> 00:19:35.729
from Stack Overflow or

00:19:35.769 --> 00:19:38.630
whatever that existed at

00:19:38.670 --> 00:19:40.009
the time that it was last indexed,

00:19:40.650 --> 00:19:43.151
and using that to inform its answers.

00:19:43.310 --> 00:19:45.071
But if developers aren't

00:19:45.152 --> 00:19:46.192
using Stack Overflow,

00:19:46.372 --> 00:19:47.711
if developers aren't writing this code,

00:19:47.751 --> 00:19:48.471
and if all the code is

00:19:48.511 --> 00:19:49.633
being written by AI now,

00:19:49.913 --> 00:19:50.472
there's a certain...

00:19:52.394 --> 00:19:54.596
recursive loop that happens, right?

00:19:54.977 --> 00:19:57.299
Like what happens in the

00:19:57.400 --> 00:20:00.483
in-between time when we

00:20:00.544 --> 00:20:01.746
don't have a fully

00:20:02.026 --> 00:20:04.628
generative AI out there,

00:20:05.029 --> 00:20:07.613
but also no one's putting

00:20:07.992 --> 00:20:09.454
new stuff out there either

00:20:09.474 --> 00:20:10.876
because they're just using AI for it.

00:20:11.472 --> 00:20:11.692
Yep.

00:20:12.134 --> 00:20:12.213
Yeah.

00:20:12.233 --> 00:20:13.494
It's a really good question.

00:20:13.575 --> 00:20:14.654
And I don't know that I have

00:20:14.674 --> 00:20:16.497
the answer to it either, because like,

00:20:16.537 --> 00:20:18.238
again, there's the maximalists,

00:20:18.278 --> 00:20:19.278
the CEOs of these companies

00:20:19.298 --> 00:20:20.579
that are going to say, oh, well, you know,

00:20:20.640 --> 00:20:23.281
we have, oh, we were, you know,

00:20:23.342 --> 00:20:25.202
a few months from, you know,

00:20:25.323 --> 00:20:26.943
actual artificial intelligence.

00:20:26.983 --> 00:20:27.344
And it's like,

00:20:28.265 --> 00:20:31.247
I mean, again, if the argument, yeah, yeah,

00:20:31.287 --> 00:20:31.626
no, you're not.

00:20:31.666 --> 00:20:32.748
And the argument against that is like,

00:20:32.847 --> 00:20:35.410
you're still selling your LLM to people.

00:20:35.470 --> 00:20:37.570
Like if you were really that close,

00:20:37.590 --> 00:20:39.092
you would stop and you

00:20:39.112 --> 00:20:40.413
would just build every company.

00:20:41.473 --> 00:20:41.673
Right.

00:20:41.753 --> 00:20:41.894
Right.

00:20:41.913 --> 00:20:42.595
That's like, like that,

00:20:42.815 --> 00:20:43.555
that's the best argument

00:20:43.575 --> 00:20:44.336
I've ever heard against.

00:20:44.375 --> 00:20:46.317
Like, like, like, yeah,

00:20:46.356 --> 00:20:47.417
like obviously you're not

00:20:47.478 --> 00:20:49.298
because you're not, you,

00:20:49.318 --> 00:20:50.400
you would just write every other,

00:20:50.680 --> 00:20:52.000
you would just be the only

00:20:52.040 --> 00:20:54.083
company if you thought you

00:20:54.103 --> 00:20:54.722
could really do that.

00:20:56.021 --> 00:20:56.942
And I think you hit the nail

00:20:56.962 --> 00:20:58.025
on the head with this

00:20:58.065 --> 00:21:00.366
feedback loop that I'm very

00:21:00.406 --> 00:21:02.429
worried about too,

00:21:02.449 --> 00:21:04.411
which is why I think taking

00:21:04.431 --> 00:21:06.492
a human out of the loop in my mind,

00:21:06.532 --> 00:21:06.933
and again,

00:21:07.134 --> 00:21:08.194
this could get dated and in

00:21:08.234 --> 00:21:09.135
three months someone could

00:21:09.175 --> 00:21:10.037
laugh at me for saying this,

00:21:10.057 --> 00:21:11.157
but I just don't see the

00:21:11.198 --> 00:21:11.919
human coming out of the

00:21:11.959 --> 00:21:13.800
loop anytime soon.

00:21:14.481 --> 00:21:16.644
I see the humans getting a lot more done.

00:21:17.904 --> 00:21:19.165
um because it's like goes to

00:21:19.185 --> 00:21:19.967
your point right like I

00:21:20.007 --> 00:21:21.307
think I think spotify and

00:21:21.387 --> 00:21:22.648
pandora is the perfect

00:21:22.709 --> 00:21:24.391
example because yeah my

00:21:24.411 --> 00:21:26.311
experience with spotify is

00:21:26.531 --> 00:21:27.712
it gets into that feedback

00:21:27.792 --> 00:21:28.854
loop right like if I wanted

00:21:28.973 --> 00:21:30.355
to go if I want to shuffle

00:21:30.415 --> 00:21:31.896
it and like have it just

00:21:31.916 --> 00:21:33.498
kind of vibe with me it's

00:21:33.518 --> 00:21:34.419
like I end up listening to

00:21:34.459 --> 00:21:35.920
the same twenty songs

00:21:36.380 --> 00:21:37.000
because it gets this

00:21:37.020 --> 00:21:38.041
feedback loop of like yeah

00:21:38.061 --> 00:21:39.143
brandon likes that song so

00:21:39.163 --> 00:21:40.564
let me throw it back exactly right

00:21:42.065 --> 00:21:44.627
versus the Pandora example,

00:21:44.667 --> 00:21:46.849
which is the human curated

00:21:46.950 --> 00:21:48.090
algorithm in some ways.

00:21:49.192 --> 00:21:50.472
Pandora didn't have humans

00:21:51.314 --> 00:21:52.234
type in our computer when

00:21:52.275 --> 00:21:53.615
you were listening to a song and be like,

00:21:53.675 --> 00:21:54.396
oh, Chris is listening to this,

00:21:54.416 --> 00:21:55.458
let me give this one to him next.

00:21:55.979 --> 00:21:58.260
But it did have humans

00:21:58.721 --> 00:21:59.922
curating that input,

00:22:00.663 --> 00:22:00.843
Yeah.

00:22:00.863 --> 00:22:02.044
In such a way that it

00:22:02.223 --> 00:22:03.204
actually was something that

00:22:03.265 --> 00:22:04.305
like felt human to you.

00:22:04.484 --> 00:22:04.765
Right.

00:22:05.465 --> 00:22:05.605
Yeah.

00:22:05.685 --> 00:22:07.146
And I just I think that

00:22:07.166 --> 00:22:09.167
that's where my theory is.

00:22:09.188 --> 00:22:09.907
That's where we're going to

00:22:09.988 --> 00:22:11.348
end up for a pretty

00:22:11.409 --> 00:22:12.829
significant period of time,

00:22:14.050 --> 00:22:16.071
because like you said, if you give it

00:22:16.771 --> 00:22:18.412
If you give today's LMS more

00:22:18.432 --> 00:22:19.133
and more context,

00:22:19.192 --> 00:22:21.333
it's really good at then.

00:22:21.573 --> 00:22:23.294
Oh, I've got a lot of stuff to read.

00:22:24.114 --> 00:22:25.775
Let me like distill it down.

00:22:25.815 --> 00:22:27.494
Like I that is what I use it

00:22:27.535 --> 00:22:28.815
for in my day to day work

00:22:29.395 --> 00:22:30.296
is like if I have a whole

00:22:30.316 --> 00:22:31.076
bunch of stuff that I want

00:22:31.096 --> 00:22:32.236
to interpret and get some

00:22:32.296 --> 00:22:33.896
help with and like kind of

00:22:35.750 --> 00:22:36.851
know brainstorm with someone

00:22:36.911 --> 00:22:37.652
on like it's going to be

00:22:37.692 --> 00:22:39.051
really good at that but if

00:22:39.092 --> 00:22:39.711
I don't give it that

00:22:39.751 --> 00:22:41.113
context and I just opened

00:22:41.173 --> 00:22:42.893
up claude or chat gpt and

00:22:42.913 --> 00:22:44.493
say hey what should I do as

00:22:44.534 --> 00:22:45.673
developer relations

00:22:45.713 --> 00:22:46.594
engineer today like yeah

00:22:46.614 --> 00:22:47.575
it's it's going to be very

00:22:47.615 --> 00:22:50.214
generic ai feeling answers

00:22:50.275 --> 00:22:50.935
it's going to be the

00:22:50.996 --> 00:22:52.556
spotify dj which sorry

00:22:52.576 --> 00:22:55.176
spotify friends but it's not great um

00:22:56.576 --> 00:22:57.557
But if I give it a little

00:22:57.597 --> 00:22:58.739
bit of love and I give it a

00:22:58.759 --> 00:22:59.699
little bit of context,

00:23:01.079 --> 00:23:02.840
it then gets and it feels a lot better.

00:23:02.861 --> 00:23:03.721
I'm like, yes, okay,

00:23:03.761 --> 00:23:04.942
now this is enhancing what

00:23:04.982 --> 00:23:06.022
I'm able to do as a human,

00:23:07.584 --> 00:23:08.305
not replacing it.

00:23:08.325 --> 00:23:08.505
Again,

00:23:08.785 --> 00:23:10.006
that's kind of my whole thesis for

00:23:10.046 --> 00:23:11.807
why I joined Kilo because I

00:23:11.886 --> 00:23:15.809
do think that an engineer

00:23:15.849 --> 00:23:17.171
that can use AI well is

00:23:17.211 --> 00:23:18.270
going to be a lot more

00:23:18.310 --> 00:23:19.612
productive and thus replace

00:23:20.252 --> 00:23:21.433
engineers that are kind of like,

00:23:22.994 --> 00:23:25.155
I'm never going to use AI anymore.

00:23:27.025 --> 00:23:27.925
And I think on the other end

00:23:27.945 --> 00:23:28.946
of that extreme, to your point,

00:23:28.967 --> 00:23:29.807
it's like the people that

00:23:29.827 --> 00:23:30.829
just think that they can

00:23:30.869 --> 00:23:32.131
just go vibe it tomorrow.

00:23:32.151 --> 00:23:33.893
It's like that eventually

00:23:33.992 --> 00:23:36.977
returns to the mean, you know, with LLMs,

00:23:37.017 --> 00:23:37.136
right?

00:23:37.156 --> 00:23:38.637
The concept of a large

00:23:38.679 --> 00:23:39.599
language model giving you

00:23:39.640 --> 00:23:42.603
the next most statistically likely token,

00:23:43.544 --> 00:23:46.527
it will necessarily revert to the mean.

00:23:47.521 --> 00:23:48.623
And they're getting around

00:23:48.663 --> 00:23:49.463
that in a lot of ways, right?

00:23:49.523 --> 00:23:51.385
It's using reasoning inside

00:23:51.405 --> 00:23:52.126
to try and go back and

00:23:52.166 --> 00:23:53.067
forth a few times to make

00:23:53.087 --> 00:23:53.647
sure it's right.

00:23:55.690 --> 00:23:57.853
But there just has to be a

00:23:57.893 --> 00:23:59.875
limit on that road that

00:23:59.894 --> 00:24:00.575
we're walking down.

00:24:00.756 --> 00:24:01.276
And sure,

00:24:01.316 --> 00:24:03.337
there may be something different

00:24:03.377 --> 00:24:04.880
that comes after it that is

00:24:05.420 --> 00:24:06.842
in a post-LLM world.

00:24:07.748 --> 00:24:08.127
Um,

00:24:08.188 --> 00:24:10.528
but right now I think it just requires

00:24:10.548 --> 00:24:12.490
that human context, right?

00:24:12.730 --> 00:24:13.730
You're talking about task lists.

00:24:13.750 --> 00:24:14.029
It's like,

00:24:14.369 --> 00:24:15.931
it's really good at consuming a

00:24:15.951 --> 00:24:18.211
task list and going point

00:24:18.231 --> 00:24:19.031
by point right now.

00:24:19.432 --> 00:24:20.553
Um,

00:24:20.692 --> 00:24:22.113
and it can be good given context coming

00:24:22.133 --> 00:24:22.792
up with those tasks,

00:24:22.813 --> 00:24:23.534
but like that's still,

00:24:23.554 --> 00:24:25.034
those are still very human in the loop,

00:24:25.874 --> 00:24:26.734
uh, situations,

00:24:27.095 --> 00:24:29.195
but like if you are human

00:24:29.215 --> 00:24:30.056
in the loop and you start

00:24:30.076 --> 00:24:30.875
to learn how to use it,

00:24:30.935 --> 00:24:32.717
you can go a lot faster.

00:24:32.836 --> 00:24:33.017
Yeah.

00:24:33.728 --> 00:24:35.509
The other conversation that

00:24:35.630 --> 00:24:37.692
I often bring up when I'm

00:24:37.711 --> 00:24:39.634
talking about AI with other

00:24:39.653 --> 00:24:41.655
developers is the idea that,

00:24:42.897 --> 00:24:44.519
especially with a lot of

00:24:44.578 --> 00:24:46.461
companies going hard on AI

00:24:46.602 --> 00:24:49.064
and using that to justify

00:24:50.045 --> 00:24:50.786
laying off junior

00:24:50.806 --> 00:24:51.747
developers or not hiring

00:24:51.767 --> 00:24:54.869
junior developers, is the idea that, yes,

00:24:56.076 --> 00:24:57.955
At this point, probably,

00:24:59.537 --> 00:25:01.457
LLMs can get us to a point

00:25:01.596 --> 00:25:03.097
where maybe you don't

00:25:03.137 --> 00:25:04.917
necessarily need junior developers.

00:25:05.458 --> 00:25:07.298
But the analog that I bring

00:25:07.337 --> 00:25:09.638
up is we don't have a lot

00:25:09.659 --> 00:25:11.118
of COBOL programmers either.

00:25:11.459 --> 00:25:12.779
There's a lot of COBOL code

00:25:12.940 --> 00:25:13.619
out in the world,

00:25:13.660 --> 00:25:15.079
particularly in banks and

00:25:15.119 --> 00:25:16.480
government institutions and whatever.

00:25:16.941 --> 00:25:18.800
And the reason why YIIK was

00:25:18.820 --> 00:25:20.421
a thing was because nobody

00:25:20.441 --> 00:25:21.902
wrote COBOL anymore because

00:25:21.922 --> 00:25:23.261
they had all aged out and retired.

00:25:24.643 --> 00:25:26.365
And if no one's hiring

00:25:26.404 --> 00:25:27.905
junior developers anymore

00:25:27.945 --> 00:25:28.646
and we have senior

00:25:28.666 --> 00:25:29.848
developers or

00:25:30.308 --> 00:25:31.830
mid-developers that are

00:25:31.851 --> 00:25:33.673
reviewing the code that is written by AI,

00:25:34.192 --> 00:25:35.734
then no one's going to be a

00:25:35.755 --> 00:25:37.517
developer anymore because

00:25:37.876 --> 00:25:38.959
they're just going to go up

00:25:39.358 --> 00:25:40.140
and no one's coming in at

00:25:40.160 --> 00:25:40.800
the entry level.

00:25:41.181 --> 00:25:42.643
And then we're going to be like...

00:25:43.944 --> 00:25:45.967
what like ai just writing

00:25:46.007 --> 00:25:47.548
its own stuff and without

00:25:47.588 --> 00:25:48.910
with like off the rails

00:25:48.950 --> 00:25:51.472
again um and like again

00:25:51.512 --> 00:25:53.094
with like theoretically

00:25:54.375 --> 00:25:56.057
maybe ai gets to the point

00:25:56.076 --> 00:25:56.817
where it can generate its

00:25:56.917 --> 00:25:59.119
own code and like check

00:25:59.160 --> 00:26:00.181
itself and make sure that

00:26:00.221 --> 00:26:03.163
it's not blowing up bank

00:26:03.223 --> 00:26:05.606
sites or you know other things but like

00:26:06.720 --> 00:26:07.099
Yeah,

00:26:07.160 --> 00:26:08.880
that's the other thing that I think about,

00:26:08.980 --> 00:26:15.882
like how the rise of AI is currently now,

00:26:16.142 --> 00:26:18.761
but even like later, like five years,

00:26:18.821 --> 00:26:19.342
ten years,

00:26:19.422 --> 00:26:22.883
how that impacts the roles of

00:26:22.982 --> 00:26:24.903
engineers in the tech industry.

00:26:24.923 --> 00:26:27.463
Again, happy to be wrong,

00:26:27.503 --> 00:26:28.743
but I think it's just a

00:26:28.784 --> 00:26:30.263
shift in skill set, right?

00:26:30.304 --> 00:26:32.005
Like my current theory is

00:26:32.045 --> 00:26:33.384
it's a shift in skill set, right?

00:26:34.567 --> 00:26:37.588
like back in the day in automotive,

00:26:37.608 --> 00:26:37.769
I don't,

00:26:37.788 --> 00:26:39.289
I don't want to like oversimplify it,

00:26:39.351 --> 00:26:39.550
but like,

00:26:39.570 --> 00:26:40.631
if you think about an automotive

00:26:40.651 --> 00:26:41.551
factory back in the day,

00:26:41.571 --> 00:26:42.113
there were probably like

00:26:42.873 --> 00:26:43.874
super highly skilled

00:26:43.933 --> 00:26:45.815
welders at every station.

00:26:46.316 --> 00:26:46.756
Yeah.

00:26:46.935 --> 00:26:47.717
And like, yeah,

00:26:47.777 --> 00:26:49.238
now there's a robot welder

00:26:49.357 --> 00:26:50.618
that's really good at it.

00:26:50.818 --> 00:26:52.279
And so there's probably less

00:26:52.720 --> 00:26:54.260
super highly skilled welders.

00:26:55.382 --> 00:26:56.722
There are lots of like,

00:26:57.324 --> 00:26:59.045
like verticals where that not true.

00:26:59.065 --> 00:26:59.164
Right.

00:26:59.184 --> 00:26:59.265
Like,

00:26:59.285 --> 00:27:00.266
I don't know if you know about like

00:27:00.465 --> 00:27:02.867
underwater welding, but it's like very,

00:27:02.907 --> 00:27:03.107
very,

00:27:03.700 --> 00:27:05.241
well-paid you know skilled

00:27:05.342 --> 00:27:07.344
labor but there's also

00:27:07.403 --> 00:27:09.204
labor in factories around

00:27:09.486 --> 00:27:11.748
how the robots work and and

00:27:12.028 --> 00:27:13.308
and making them work more

00:27:13.328 --> 00:27:13.990
efficiently and

00:27:14.109 --> 00:27:15.371
understanding how to use

00:27:15.391 --> 00:27:17.032
them and understanding how

00:27:17.053 --> 00:27:19.115
to design cars that can

00:27:19.335 --> 00:27:20.635
then be manufactured well

00:27:20.655 --> 00:27:21.517
like I think we'll see

00:27:21.576 --> 00:27:24.680
similar skill shifting um

00:27:24.759 --> 00:27:27.863
in the mid to long term here because like

00:27:29.692 --> 00:27:29.932
Yes,

00:27:29.992 --> 00:27:32.433
I think it will take maybe less total

00:27:32.493 --> 00:27:33.493
people to do things.

00:27:33.534 --> 00:27:34.714
But every time that we've

00:27:34.755 --> 00:27:36.494
seen that happen since the

00:27:36.515 --> 00:27:37.295
Industrial Revolution,

00:27:37.535 --> 00:27:38.296
it doesn't mean that

00:27:38.336 --> 00:27:39.776
there's like no jobs anymore.

00:27:40.576 --> 00:27:41.957
I don't think this will be it either.

00:27:41.997 --> 00:27:42.277
Again,

00:27:42.297 --> 00:27:44.978
I could my kids might be mad at me

00:27:45.018 --> 00:27:46.317
saying this in ten years,

00:27:46.357 --> 00:27:47.637
but when they don't have a job.

00:27:47.678 --> 00:27:49.719
But I just think it's a shift in skills.

00:27:49.759 --> 00:27:49.858
Right.

00:27:49.878 --> 00:27:49.979
Like,

00:27:50.038 --> 00:27:51.519
I think you become more of this

00:27:51.599 --> 00:27:52.859
product focused engineer.

00:27:52.900 --> 00:27:53.079
Right.

00:27:53.500 --> 00:27:55.980
I think if we were to see if

00:27:56.661 --> 00:27:57.520
If everything you just described,

00:27:57.540 --> 00:27:58.402
which is again,

00:27:58.422 --> 00:27:59.102
a lot of people are talking

00:27:59.122 --> 00:28:00.702
about like it's possible and maybe it is,

00:28:01.423 --> 00:28:01.722
but it's like,

00:28:01.823 --> 00:28:03.223
why do we had product

00:28:03.723 --> 00:28:05.164
product and product

00:28:05.184 --> 00:28:08.026
managers as a as a like a

00:28:08.086 --> 00:28:09.666
function this whole time?

00:28:09.707 --> 00:28:09.946
Right.

00:28:10.387 --> 00:28:11.367
Like, oh, why?

00:28:11.428 --> 00:28:11.788
Why?

00:28:11.867 --> 00:28:12.509
Why don't we just have

00:28:12.548 --> 00:28:13.969
engineers like and there's

00:28:13.989 --> 00:28:14.690
lots of companies that have

00:28:14.710 --> 00:28:16.170
tried that and it hasn't gone well.

00:28:16.210 --> 00:28:16.450
Why?

00:28:16.509 --> 00:28:17.111
Because it's kind of

00:28:17.151 --> 00:28:18.570
different specialized skills of like,

00:28:18.631 --> 00:28:19.011
yeah,

00:28:19.051 --> 00:28:20.392
understanding why and what we're

00:28:20.412 --> 00:28:21.752
building versus understanding how

00:28:22.753 --> 00:28:23.595
We're going to build it.

00:28:25.036 --> 00:28:25.855
And so I think it'll just be

00:28:25.875 --> 00:28:26.717
a shift in that kind of

00:28:26.757 --> 00:28:29.218
skill set where I think it

00:28:29.258 --> 00:28:30.419
will require a different

00:28:30.459 --> 00:28:32.099
kind of methodology where

00:28:32.621 --> 00:28:34.122
you still have to tell it what to build,

00:28:34.182 --> 00:28:34.402
right?

00:28:34.461 --> 00:28:37.124
And like, again,

00:28:37.144 --> 00:28:37.763
those of us that have been

00:28:37.784 --> 00:28:38.805
around software long enough

00:28:38.904 --> 00:28:40.945
know that like saying this

00:28:40.986 --> 00:28:42.406
is what we want the software to do.

00:28:43.208 --> 00:28:44.929
And then like what that really means,

00:28:44.969 --> 00:28:46.369
when it hits the real world

00:28:46.470 --> 00:28:48.651
are like a very tough

00:28:48.671 --> 00:28:49.971
translation and always has

00:28:50.011 --> 00:28:52.273
been and always, I think always will be.

00:28:53.335 --> 00:28:56.977
Because in your example of, oh,

00:28:56.997 --> 00:28:57.857
it's a bank and they've got

00:28:57.938 --> 00:29:00.220
zero software engineers or

00:29:00.700 --> 00:29:04.022
zero people mining the AI

00:29:04.123 --> 00:29:05.483
self-governing thing.

00:29:05.943 --> 00:29:06.304
It's like,

00:29:06.565 --> 00:29:07.585
I don't think that's going to be real.

00:29:07.625 --> 00:29:08.526
I think if that were real today,

00:29:08.566 --> 00:29:09.946
the banks wouldn't have

00:29:10.007 --> 00:29:10.907
product people and they

00:29:10.928 --> 00:29:11.709
wouldn't have business

00:29:11.749 --> 00:29:12.489
people and they wouldn't

00:29:12.528 --> 00:29:13.589
have... They would just

00:29:13.609 --> 00:29:14.990
have five engineers and it would be fine.

00:29:15.010 --> 00:29:15.991
Why is that not fine?

00:29:16.412 --> 00:29:16.551
Well,

00:29:16.571 --> 00:29:18.073
because it's a diversity of skill sets.

00:29:18.093 --> 00:29:20.755
And so if AI becomes better at...

00:29:21.817 --> 00:29:22.617
like you said maybe the

00:29:22.678 --> 00:29:23.839
junior engineering kind of

00:29:23.859 --> 00:29:26.123
things um or maybe kind of

00:29:26.222 --> 00:29:28.527
more of the coding right um

00:29:28.606 --> 00:29:29.827
and and I think it's just

00:29:29.888 --> 00:29:30.930
an abstracted layer off

00:29:30.990 --> 00:29:32.573
right like to your point we

00:29:32.593 --> 00:29:33.354
don't have a lot of cobalt

00:29:33.394 --> 00:29:34.055
engineers because there's

00:29:34.075 --> 00:29:35.195
not a lot of cobalt we also

00:29:35.237 --> 00:29:36.578
have like very few people

00:29:38.241 --> 00:29:39.903
relatively that are doing

00:29:39.943 --> 00:29:41.286
like machine coding, right?

00:29:41.306 --> 00:29:43.167
Like assembly level coding, right?

00:29:43.188 --> 00:29:43.867
Like why?

00:29:44.368 --> 00:29:44.669
Why?

00:29:44.729 --> 00:29:46.111
Because we have higher level

00:29:46.151 --> 00:29:47.573
languages on top of them, right?

00:29:48.012 --> 00:29:50.155
And so like in my mind, this just might be,

00:29:50.215 --> 00:29:51.396
let's say we don't look at

00:29:51.517 --> 00:29:52.518
JavaScript anymore.

00:29:53.578 --> 00:29:54.940
okay maybe I could see that

00:29:54.980 --> 00:29:57.421
world um there will still

00:29:57.480 --> 00:29:59.021
be the english language or

00:29:59.041 --> 00:30:00.583
whatever language were it's

00:30:00.603 --> 00:30:01.542
just another level of

00:30:01.603 --> 00:30:02.784
abstraction on top of it

00:30:02.864 --> 00:30:05.005
right um and they'll need

00:30:05.025 --> 00:30:07.486
people um to handle it and

00:30:07.506 --> 00:30:09.106
so so to say that they're

00:30:09.146 --> 00:30:11.428
not in my mind is kind of

00:30:11.468 --> 00:30:13.209
like in the early days of

00:30:13.489 --> 00:30:14.808
very very early days of

00:30:14.848 --> 00:30:15.970
computing right this is like

00:30:17.342 --> 00:30:18.863
talking about Admiral Grace

00:30:18.903 --> 00:30:20.284
Hopper and the early

00:30:20.663 --> 00:30:22.144
founders of our industry,

00:30:23.025 --> 00:30:24.645
it was very much seen as

00:30:25.467 --> 00:30:26.867
when computers were becoming a thing,

00:30:26.907 --> 00:30:27.887
let's remember if you've

00:30:27.907 --> 00:30:29.608
ever seen Hidden Figures,

00:30:29.648 --> 00:30:31.190
computer used to mean a person.

00:30:31.750 --> 00:30:34.152
And typically- A computer, a person.

00:30:34.192 --> 00:30:34.632
Yeah,

00:30:34.652 --> 00:30:36.692
a computer was a person who computes.

00:30:37.192 --> 00:30:38.334
And it was seen as this kind

00:30:38.354 --> 00:30:39.694
of lower level thing.

00:30:39.714 --> 00:30:39.954
And it was

00:30:40.434 --> 00:30:41.595
Oh, women are entering the workforce.

00:30:41.674 --> 00:30:41.714
Oh,

00:30:41.734 --> 00:30:42.915
we'll let women be the computers

00:30:42.935 --> 00:30:44.935
because it's not as high skilled.

00:30:45.556 --> 00:30:45.836
Right.

00:30:46.155 --> 00:30:47.076
And like, obviously,

00:30:47.876 --> 00:30:48.717
and that led to a lot of

00:30:49.497 --> 00:30:50.998
the founders of our

00:30:51.117 --> 00:30:52.798
software engineering industry being women,

00:30:52.817 --> 00:30:53.817
which is fantastic, right?

00:30:53.837 --> 00:30:56.199
Like Admiral Hopper or

00:30:56.278 --> 00:30:58.078
Margaret Hamilton on the Apollo program.

00:30:59.400 --> 00:31:01.480
But the reality is those folks said like,

00:31:01.559 --> 00:31:01.799
hey,

00:31:02.181 --> 00:31:03.020
we've got to treat this as an

00:31:03.060 --> 00:31:03.840
engineering skill.

00:31:04.894 --> 00:31:05.074
Right.

00:31:05.094 --> 00:31:05.894
Like Margaret Hamilton.

00:31:05.914 --> 00:31:07.237
I have a whole talk that I did on her.

00:31:07.277 --> 00:31:09.078
And when she was working on

00:31:09.098 --> 00:31:09.960
the Apollo program,

00:31:10.759 --> 00:31:11.862
people thought of it as like, oh, yeah,

00:31:11.882 --> 00:31:12.782
that's just the software.

00:31:12.863 --> 00:31:14.605
We're rocket scientists.

00:31:14.625 --> 00:31:15.605
That's the real engineering.

00:31:16.926 --> 00:31:17.748
And her point was like,

00:31:17.928 --> 00:31:19.089
this thing that I'm

00:31:19.130 --> 00:31:19.869
building is the Apollo

00:31:19.890 --> 00:31:21.432
guidance computer is going

00:31:21.451 --> 00:31:22.673
to be all of the telemetry

00:31:22.692 --> 00:31:23.994
for guiding us to the moon and back.

00:31:24.740 --> 00:31:24.901
Right.

00:31:25.020 --> 00:31:25.902
And guess what?

00:31:25.922 --> 00:31:26.622
If you miss, you know,

00:31:26.642 --> 00:31:27.643
if anybody's seen Apollo thirteen,

00:31:27.663 --> 00:31:29.463
if you miss the Earth's the

00:31:29.584 --> 00:31:31.305
angle of attack on the Earth's atmosphere,

00:31:31.325 --> 00:31:32.865
you either skip out or you burn up.

00:31:32.905 --> 00:31:34.727
So like it's kind of a life

00:31:34.747 --> 00:31:36.407
safety issue to the software.

00:31:36.567 --> 00:31:37.828
Yeah.

00:31:37.949 --> 00:31:39.410
And so she's she coined the

00:31:39.430 --> 00:31:40.390
term software engineer in a

00:31:40.411 --> 00:31:41.171
lot of ways because she's like,

00:31:41.330 --> 00:31:41.971
we're going to treat this

00:31:41.990 --> 00:31:42.872
like an engineering discipline.

00:31:44.028 --> 00:31:45.329
way off the rails of your question.

00:31:45.369 --> 00:31:45.790
But like,

00:31:46.171 --> 00:31:47.432
what I think that just means to

00:31:47.491 --> 00:31:47.892
us is like,

00:31:47.991 --> 00:31:49.173
it's just a different skill

00:31:49.212 --> 00:31:50.034
set that evolves.

00:31:50.693 --> 00:31:53.256
And I think it's naive to say, again,

00:31:53.296 --> 00:31:54.196
could be naive of me to say this,

00:31:54.217 --> 00:31:55.998
but I think it's naive to say, oh,

00:31:56.038 --> 00:31:57.219
that'll just go away and it

00:31:57.239 --> 00:31:58.118
won't be a thing anymore.

00:31:58.159 --> 00:31:59.921
That's the same naivety of,

00:32:00.601 --> 00:32:01.780
oh computers yeah that's

00:32:01.820 --> 00:32:03.382
just yeah let the let the

00:32:03.582 --> 00:32:04.481
you know lower skilled

00:32:04.541 --> 00:32:06.182
folks program because that

00:32:06.202 --> 00:32:07.103
won't be a thing anymore

00:32:07.123 --> 00:32:08.383
it's like obviously that

00:32:08.403 --> 00:32:10.604
was wrong the the auto

00:32:10.644 --> 00:32:11.864
manufacturing analogy is a

00:32:11.884 --> 00:32:14.765
really good one um uh I was

00:32:14.865 --> 00:32:16.285
recently looking at my uh

00:32:16.826 --> 00:32:17.865
like college programs with

00:32:17.905 --> 00:32:18.705
my son and he's really

00:32:18.746 --> 00:32:19.365
interested he was

00:32:19.425 --> 00:32:20.246
interested in the idea of

00:32:20.287 --> 00:32:21.007
robotics and so I was

00:32:21.027 --> 00:32:21.846
looking at a robotics

00:32:22.626 --> 00:32:24.627
actual program and what it involved.

00:32:24.708 --> 00:32:26.008
And most of what the

00:32:26.028 --> 00:32:27.428
robotics program involved

00:32:27.827 --> 00:32:29.969
was working with and

00:32:30.028 --> 00:32:31.308
learning the software that

00:32:31.388 --> 00:32:32.710
programs the robots.

00:32:32.769 --> 00:32:34.410
Not doing anything with the

00:32:34.470 --> 00:32:36.369
robots or building the robotics yourself,

00:32:36.691 --> 00:32:38.691
but doing the stuff to tell

00:32:38.730 --> 00:32:40.431
the robot how to do the

00:32:40.471 --> 00:32:41.352
stuff that it does.

00:32:42.352 --> 00:32:44.231
And it's almost an exact

00:32:44.271 --> 00:32:46.272
parallel of the idea of, yeah,

00:32:46.292 --> 00:32:47.232
we have this tool,

00:32:47.633 --> 00:32:48.653
and then we have the people

00:32:48.673 --> 00:32:49.874
that operate the tool,

00:32:49.933 --> 00:32:51.054
and the tool does the work.

00:32:51.473 --> 00:32:53.256
But, like, we're just one step removed.

00:32:53.276 --> 00:32:56.200
But you need to have the human, like,

00:32:56.359 --> 00:32:57.401
doing the stuff that tells

00:32:57.480 --> 00:32:58.903
the tool how to do its job.

00:32:59.022 --> 00:33:01.906
Otherwise, the tool is going to, you know,

00:33:01.926 --> 00:33:02.748
blow up the car or something.

00:33:02.768 --> 00:33:04.750
Yeah, yeah, yeah, exactly.

00:33:04.769 --> 00:33:04.849
Yeah,

00:33:04.869 --> 00:33:06.612
so I think the worst case scenario

00:33:06.632 --> 00:33:07.854
that I see is we get to

00:33:07.894 --> 00:33:09.816
that same place where...

00:33:10.797 --> 00:33:13.159
Yes, there's a robot, ChatGPT,

00:33:13.640 --> 00:33:14.519
that's a lot better at

00:33:14.539 --> 00:33:18.644
writing export function.

00:33:19.084 --> 00:33:20.184
But the human still has to

00:33:20.224 --> 00:33:24.669
tell it where we're going.

00:33:25.148 --> 00:33:29.252
Moving all the way back, what sort of

00:33:29.833 --> 00:33:31.673
drew you into the world of

00:33:31.834 --> 00:33:33.013
developer relations?

00:33:34.295 --> 00:33:38.176
It seems like I was looking at your site,

00:33:38.217 --> 00:33:39.057
and you have been doing

00:33:39.076 --> 00:33:40.217
this for quite a bit.

00:33:41.617 --> 00:33:43.618
What pulled you into developer relations?

00:33:44.183 --> 00:33:44.544
Yeah.

00:33:44.703 --> 00:33:44.923
Yeah.

00:33:45.284 --> 00:33:46.005
Well, I think like you,

00:33:46.085 --> 00:33:47.605
I kind of been doing it for a long time,

00:33:47.644 --> 00:33:47.785
right?

00:33:47.825 --> 00:33:49.226
So I,

00:33:49.246 --> 00:33:51.207
I spent my early part of my career at

00:33:51.326 --> 00:33:52.807
like not venture backed startups,

00:33:52.866 --> 00:33:54.067
but at small companies that

00:33:54.087 --> 00:33:56.848
were kind of like bootstrap funded, uh,

00:33:56.868 --> 00:33:57.348
if you will.

00:33:57.429 --> 00:33:58.469
So in medical software,

00:33:58.489 --> 00:34:00.029
I spent ten years in medical software.

00:34:00.910 --> 00:34:02.349
Um, and then I,

00:34:02.911 --> 00:34:04.131
I live in on the east coast

00:34:04.230 --> 00:34:05.230
of the US in Maryland and

00:34:05.412 --> 00:34:05.971
everybody here is a

00:34:06.031 --> 00:34:06.751
government contractor.

00:34:06.771 --> 00:34:07.551
So I spent a little bit of

00:34:07.571 --> 00:34:08.292
time doing that.

00:34:09.073 --> 00:34:09.252
Um,

00:34:10.660 --> 00:34:10.980
Uh,

00:34:11.000 --> 00:34:12.842
but all of the time that I was doing that,

00:34:13.022 --> 00:34:14.944
I was managing engineering teams.

00:34:14.984 --> 00:34:16.046
I was managing product.

00:34:18.247 --> 00:34:20.608
And kind of a lot of that is, you know,

00:34:20.888 --> 00:34:21.648
developer relation, right?

00:34:21.669 --> 00:34:22.650
Like if you're an engineering manager,

00:34:22.690 --> 00:34:24.851
you do develop, like spoiler alert,

00:34:24.911 --> 00:34:25.672
you do some developer

00:34:25.711 --> 00:34:26.371
relations because you're

00:34:26.391 --> 00:34:27.172
relating with developers.

00:34:29.173 --> 00:34:29.574
And then I got an

00:34:29.594 --> 00:34:31.175
opportunity to work at a startup, GitLab.

00:34:33.117 --> 00:34:35.878
And my first job there, again,

00:34:35.898 --> 00:34:36.539
was in engineering

00:34:36.559 --> 00:34:37.559
management because I had,

00:34:37.820 --> 00:34:38.940
at this government contractor,

00:34:38.981 --> 00:34:40.702
been responsible for

00:34:40.782 --> 00:34:42.023
GitLab's deployment at our

00:34:42.043 --> 00:34:43.543
government customer and at

00:34:43.603 --> 00:34:44.704
our own environment.

00:34:45.344 --> 00:34:45.864
And so I was like, oh,

00:34:45.905 --> 00:34:46.686
I understand GitLab.

00:34:46.726 --> 00:34:47.525
I can build an engineering

00:34:47.565 --> 00:34:48.746
team to deploy GitLab.

00:34:48.766 --> 00:34:49.487
And I did.

00:34:51.068 --> 00:34:51.887
And then I got to spend a

00:34:51.927 --> 00:34:53.289
little bit of time in product there.

00:34:54.048 --> 00:34:54.909
We were growing a lot.

00:34:54.949 --> 00:34:55.409
So I was like,

00:34:55.630 --> 00:34:56.550
I was like employee like one

00:34:56.590 --> 00:34:58.090
hundred and fifty when I

00:34:58.130 --> 00:35:01.373
was hired in twenty seventeen, I think.

00:35:02.353 --> 00:35:03.014
And I was there for five

00:35:03.054 --> 00:35:03.833
years and we grew to three

00:35:03.873 --> 00:35:04.474
thousand people.

00:35:04.554 --> 00:35:04.715
So.

00:35:05.614 --> 00:35:07.117
Very fast, crazy growth.

00:35:07.137 --> 00:35:07.978
This is like Zerp.

00:35:09.259 --> 00:35:10.161
This is a crazy time.

00:35:11.661 --> 00:35:12.362
And so we were growing all

00:35:12.382 --> 00:35:12.943
the time and we were

00:35:12.963 --> 00:35:13.804
growing the product team

00:35:13.844 --> 00:35:15.286
and I got a chance to step

00:35:15.347 --> 00:35:17.509
in and own the CI product

00:35:17.528 --> 00:35:18.710
for GitLab for about a year,

00:35:18.731 --> 00:35:19.612
which was crazy to me

00:35:19.632 --> 00:35:20.192
because it was the thing

00:35:20.211 --> 00:35:21.474
that made me fall in love with GitLab,

00:35:21.494 --> 00:35:22.835
the product to begin with.

00:35:22.875 --> 00:35:24.077
And now they're handing me keys to it.

00:35:25.610 --> 00:35:27.070
And all this time there's this woman,

00:35:27.090 --> 00:35:28.210
her name is Priyanka Sharma.

00:35:28.251 --> 00:35:30.032
She actually is now the

00:35:30.132 --> 00:35:31.152
executive director of the

00:35:31.192 --> 00:35:32.873
CNCF where like Kubernetes sits.

00:35:33.693 --> 00:35:34.733
By the time she was at

00:35:34.773 --> 00:35:36.534
GitLab and she kept asking me like, hey,

00:35:36.894 --> 00:35:37.735
come do developer relations.

00:35:37.815 --> 00:35:38.655
I kept saying, no, no,

00:35:38.675 --> 00:35:39.536
I don't want to like travel

00:35:39.556 --> 00:35:40.456
that much and I don't want

00:35:40.496 --> 00:35:41.757
to do all this.

00:35:42.724 --> 00:35:43.664
And the story I tell here is

00:35:43.684 --> 00:35:45.405
I was at an AWS reInvent,

00:35:45.686 --> 00:35:46.447
which is a massive

00:35:46.507 --> 00:35:47.827
conference for anyone that's never been.

00:35:48.728 --> 00:35:49.969
And I was working the GitLab booth.

00:35:51.030 --> 00:35:53.512
And in those early days of GitLab, this is,

00:35:54.014 --> 00:35:56.635
let's say you are coming up.

00:35:57.396 --> 00:35:58.157
You're working the booth.

00:35:58.177 --> 00:35:59.257
You're going to work the GitLab booth.

00:35:59.277 --> 00:35:59.998
This is the early days

00:36:00.018 --> 00:36:00.838
working in the GitLab booth.

00:36:00.878 --> 00:36:03.460
And I'm going to walk in.

00:36:03.521 --> 00:36:04.121
I'm looking at you.

00:36:05.782 --> 00:36:06.423
I'm looking at your booth.

00:36:07.590 --> 00:36:08.371
Are you GitHub?

00:36:09.072 --> 00:36:09.231
Right?

00:36:09.251 --> 00:36:11.673
Like this is what happened early days,

00:36:11.695 --> 00:36:11.835
right?

00:36:11.855 --> 00:36:14.077
I was like, no, we're not GitHub.

00:36:14.996 --> 00:36:16.257
So this person came up and

00:36:16.277 --> 00:36:17.780
they were confused, similar confusion,

00:36:17.820 --> 00:36:18.800
but even more, I realized, oh,

00:36:18.820 --> 00:36:19.581
they don't understand like

00:36:19.601 --> 00:36:22.443
what Git is and like source

00:36:22.483 --> 00:36:23.864
code management maybe even,

00:36:23.905 --> 00:36:24.945
or like definitely not what Git is.

00:36:24.965 --> 00:36:26.586
And I was like, oh, okay.

00:36:26.606 --> 00:36:28.108
Like forget GitLab, like whatever.

00:36:28.168 --> 00:36:28.728
Like, let me show you.

00:36:28.748 --> 00:36:29.489
And so I grabbed this piece

00:36:29.510 --> 00:36:30.070
of paper and I started

00:36:30.090 --> 00:36:31.871
drawing like local repository,

00:36:31.891 --> 00:36:33.072
like your laptop and like

00:36:33.172 --> 00:36:35.675
remote repository, like GitHub or GitLab,

00:36:35.695 --> 00:36:35.914
you know?

00:36:37.451 --> 00:36:39.331
GitLab, really, right?

00:36:39.492 --> 00:36:40.592
And I'm like, explain all this.

00:36:40.632 --> 00:36:42.554
And I get done, and they're like, okay,

00:36:42.574 --> 00:36:42.773
thanks.

00:36:43.454 --> 00:36:44.215
Now they understand a lot more.

00:36:44.675 --> 00:36:45.416
And I turned to a friend of mine,

00:36:45.436 --> 00:36:45.996
and I was like, man,

00:36:46.016 --> 00:36:47.217
that's the thing I would do for free.

00:36:48.117 --> 00:36:50.179
And that friend, they say to me,

00:36:50.199 --> 00:36:50.500
they're like,

00:36:50.599 --> 00:36:51.900
why won't you go work for Priyanka then?

00:36:51.940 --> 00:36:53.862
And I was like, you know what?

00:36:54.163 --> 00:36:54.822
You're not wrong.

00:36:54.862 --> 00:36:58.184
And so that is what then I

00:36:58.204 --> 00:36:59.126
spent most of my career at

00:36:59.166 --> 00:37:00.467
GitLab in developer relations,

00:37:00.507 --> 00:37:01.628
and I've been doing it ever since.

00:37:02.608 --> 00:37:02.688
Yeah,

00:37:02.708 --> 00:37:04.128
it's just this concept of being able

00:37:04.148 --> 00:37:06.731
to like teach and like get

00:37:06.751 --> 00:37:08.112
people excited and be

00:37:08.152 --> 00:37:09.092
excited about the products.

00:37:09.112 --> 00:37:11.094
Like I've always worked with

00:37:11.153 --> 00:37:11.835
products that I've been

00:37:11.855 --> 00:37:12.775
excited about for one

00:37:12.815 --> 00:37:14.996
reason or another ever

00:37:15.016 --> 00:37:17.619
since I worked at GitLab for sure,

00:37:18.139 --> 00:37:19.159
if not even before that.

00:37:19.239 --> 00:37:21.442
And so, yeah, that's what got me into it.

00:37:21.481 --> 00:37:22.902
And now I've been doing it ever since.

00:37:23.822 --> 00:37:25.664
And so, yeah, that was, I don't know,

00:37:25.684 --> 00:37:26.585
that was like probably five

00:37:26.605 --> 00:37:27.485
years ago now at this point.

00:37:27.505 --> 00:37:29.007
So it's been a long time.

00:37:30.818 --> 00:37:34.521
Yeah.

00:37:35.282 --> 00:37:37.284
I feel like that's very similar.

00:37:37.704 --> 00:37:38.905
I don't know that I had the exact,

00:37:39.206 --> 00:37:40.027
I would do this for free,

00:37:40.067 --> 00:37:43.809
but I was already doing it

00:37:44.490 --> 00:37:46.391
in engineering roles even

00:37:46.411 --> 00:37:49.175
before I spent a couple of

00:37:49.195 --> 00:37:52.036
years doing online

00:37:52.056 --> 00:37:53.338
developer training classes,

00:37:54.518 --> 00:37:56.201
making them for Pluralsight.

00:37:58.159 --> 00:37:59.460
which was really big for a

00:37:59.501 --> 00:38:01.061
minute and then kind of got

00:38:01.101 --> 00:38:02.501
eaten by LinkedIn Learning.

00:38:04.061 --> 00:38:06.083
But yeah,

00:38:06.202 --> 00:38:09.105
I was always already sort of

00:38:09.125 --> 00:38:10.644
speaking at conferences and

00:38:10.664 --> 00:38:11.565
meetups and stuff,

00:38:13.365 --> 00:38:14.746
and this is the stuff that

00:38:14.786 --> 00:38:15.527
I really like doing.

00:38:18.068 --> 00:38:22.577
Do you, what is sort of your, what's,

00:38:23.097 --> 00:38:23.699
I'll ask this,

00:38:24.679 --> 00:38:28.527
what is your default presentation,

00:38:28.768 --> 00:38:30.150
your default conference talk?

00:38:31.871 --> 00:38:32.951
my default conference.

00:38:32.971 --> 00:38:34.652
So I always, I started to give it,

00:38:34.672 --> 00:38:35.793
it was my Margaret Hamilton one.

00:38:36.393 --> 00:38:38.693
Uh, yeah, yeah, no, I, so I had a,

00:38:38.893 --> 00:38:41.534
I did a whole thing on, uh,

00:38:41.554 --> 00:38:42.434
I did a lot of deep talk

00:38:42.474 --> 00:38:43.436
research and Margaret Hamilton,

00:38:43.476 --> 00:38:45.936
she's an incredible, like again,

00:38:45.996 --> 00:38:46.597
mother of our,

00:38:46.757 --> 00:38:47.978
our industry coined the

00:38:48.018 --> 00:38:48.898
term software engineer.

00:38:49.947 --> 00:38:51.188
um, you know, really,

00:38:51.867 --> 00:38:53.429
really pushed or like, uh, there's,

00:38:53.528 --> 00:38:55.210
there's all, and then I have like, I like,

00:38:55.250 --> 00:38:56.311
so I have a lot of Margaret

00:38:56.331 --> 00:38:57.251
Hamilton stories that I can

00:38:57.291 --> 00:38:59.072
work into anything, right?

00:38:59.092 --> 00:38:59.432
Like if you,

00:38:59.452 --> 00:39:01.293
if people ask you the question, like,

00:39:01.313 --> 00:39:02.554
what are the things you could talk with,

00:39:02.695 --> 00:39:04.155
talk about without, you know, prepping,

00:39:04.175 --> 00:39:04.376
like,

00:39:06.023 --> 00:39:06.603
for a long time.

00:39:06.623 --> 00:39:07.704
And I don't know if you know

00:39:07.724 --> 00:39:08.885
the Abraham Lincoln quote,

00:39:08.907 --> 00:39:10.768
but like he people, he said, well,

00:39:10.789 --> 00:39:11.730
if you want me to talk for

00:39:11.789 --> 00:39:12.931
like forty five minutes or an hour,

00:39:12.951 --> 00:39:13.612
I'm ready now.

00:39:14.152 --> 00:39:14.833
If you want me to talk for

00:39:14.873 --> 00:39:15.494
five or ten minutes,

00:39:15.514 --> 00:39:16.576
I'm going to need a couple of weeks.

00:39:16.916 --> 00:39:19.940
Right.

00:39:20.059 --> 00:39:20.380
And so, yeah,

00:39:20.400 --> 00:39:22.322
I can talk about her and like

00:39:22.342 --> 00:39:23.764
a lot of great stories

00:39:25.686 --> 00:39:27.889
that she had, she had her, she,

00:39:28.431 --> 00:39:29.371
not surprisingly for the

00:39:29.391 --> 00:39:31.434
time of the sixties was she was a mom.

00:39:31.594 --> 00:39:32.755
And so she had to end up

00:39:32.775 --> 00:39:33.978
bringing her daughter there

00:39:34.077 --> 00:39:35.460
a lot to work because the

00:39:35.519 --> 00:39:37.782
men would not have to think.

00:39:37.822 --> 00:39:38.423
And they would just work

00:39:38.463 --> 00:39:40.106
late into the evening.

00:39:40.166 --> 00:39:41.668
And like, it was just expected of them.

00:39:42.554 --> 00:39:43.215
but she would have to have

00:39:43.235 --> 00:39:43.996
her daughter there with her.

00:39:45.835 --> 00:39:47.637
And her daughter was like six at the time,

00:39:47.657 --> 00:39:49.197
I want to say, and was in the like lunar,

00:39:49.398 --> 00:39:54.278
like lander, like simulator and was,

00:39:54.298 --> 00:39:54.478
you know,

00:39:54.539 --> 00:39:55.860
messing with stuff and

00:39:56.280 --> 00:39:57.721
programmed the Apollo,

00:39:57.740 --> 00:39:58.300
the Apollo guidance

00:39:58.320 --> 00:39:59.121
computer that she worked on

00:39:59.161 --> 00:40:00.402
is like the one that you can think of,

00:40:00.442 --> 00:40:00.782
like, right.

00:40:00.802 --> 00:40:02.262
Like the green screen and

00:40:02.282 --> 00:40:03.702
the big buttons that were made for,

00:40:04.603 --> 00:40:05.563
And the daughter just like

00:40:05.842 --> 00:40:06.963
is haphazardly kind of

00:40:07.003 --> 00:40:08.204
playing with it and puts in

00:40:08.244 --> 00:40:09.724
like the they're like mid

00:40:09.784 --> 00:40:10.644
flight in the simulator,

00:40:10.804 --> 00:40:11.724
like halfway to the moon.

00:40:12.385 --> 00:40:13.206
And she puts in like the

00:40:13.286 --> 00:40:15.726
launch like program zero zero zero,

00:40:15.945 --> 00:40:16.126
right?

00:40:16.166 --> 00:40:17.487
Like like restart.

00:40:18.367 --> 00:40:19.827
And it was like a massive

00:40:19.867 --> 00:40:20.668
problem because like

00:40:20.708 --> 00:40:22.407
there's all this very little memory,

00:40:22.427 --> 00:40:23.168
as you can imagine.

00:40:24.588 --> 00:40:27.309
And and like it cleared out

00:40:27.349 --> 00:40:28.690
all of the telemetry data.

00:40:29.913 --> 00:40:32.773
on where they were in space.

00:40:33.894 --> 00:40:36.436
And Margaret went up and said, oh,

00:40:37.356 --> 00:40:38.436
we've really got to put

00:40:38.476 --> 00:40:40.916
some protections in here.

00:40:40.996 --> 00:40:42.358
And everyone was like, no, no, no.

00:40:42.378 --> 00:40:44.378
These are professional astronauts.

00:40:44.418 --> 00:40:45.139
They don't need.

00:40:46.023 --> 00:40:47.164
And guess what happened?

00:40:47.304 --> 00:40:48.666
Jim Lovell's first trip to the moon,

00:40:48.706 --> 00:40:50.987
not Apollo-XIII, but Apollo-VIII.

00:40:51.086 --> 00:40:52.467
Guess what happened?

00:40:52.688 --> 00:40:53.648
Someone literally did that.

00:40:54.068 --> 00:40:55.409
It took three hours of

00:40:55.710 --> 00:40:56.949
relaying telemetry data

00:40:56.989 --> 00:40:59.012
back up from NASA to reprogram.

00:41:00.452 --> 00:41:02.092
And Margaret was like, see?

00:41:02.172 --> 00:41:02.614
Told you.

00:41:02.753 --> 00:41:03.954
Yeah.

00:41:03.974 --> 00:41:05.655
Her daughter just invented QA.

00:41:06.175 --> 00:41:06.695
Exactly.

00:41:06.715 --> 00:41:07.135
Yeah, QA.

00:41:07.817 --> 00:41:08.356
So in this talk,

00:41:08.376 --> 00:41:09.177
I talk about how she

00:41:09.217 --> 00:41:10.318
basically invented QA.

00:41:10.657 --> 00:41:11.438
She basically invented

00:41:11.478 --> 00:41:13.119
branching because...

00:41:13.780 --> 00:41:15.179
she would show up I mean it

00:41:15.199 --> 00:41:15.880
wasn't like really

00:41:15.920 --> 00:41:16.541
branching like we think

00:41:16.561 --> 00:41:18.181
about it today but like she

00:41:18.201 --> 00:41:19.420
showed up to her office one

00:41:19.440 --> 00:41:20.661
morning and had broken it

00:41:20.981 --> 00:41:21.782
and there was like a line

00:41:21.802 --> 00:41:22.661
of people that couldn't

00:41:22.722 --> 00:41:23.963
work because like she had

00:41:23.983 --> 00:41:25.242
broken the bill basically

00:41:25.282 --> 00:41:26.822
the night before again I'm

00:41:26.842 --> 00:41:27.643
using terms that are like

00:41:27.682 --> 00:41:28.523
modern that like don't

00:41:28.623 --> 00:41:29.523
really apply to how they

00:41:29.563 --> 00:41:30.364
did it so like if anyone

00:41:30.963 --> 00:41:31.724
understands how the

00:41:31.744 --> 00:41:32.465
apologizing computer or

00:41:32.485 --> 00:41:33.744
software works yes I'm

00:41:33.784 --> 00:41:34.684
oversimplifying it but like

00:41:34.704 --> 00:41:35.485
she had broken the bill the

00:41:35.505 --> 00:41:36.606
night before and had a line

00:41:36.626 --> 00:41:38.005
at her at her office the next day

00:41:38.666 --> 00:41:39.367
And she was like, oh,

00:41:39.387 --> 00:41:39.989
we should probably have a

00:41:40.048 --> 00:41:41.190
test environment where we

00:41:41.210 --> 00:41:42.733
don't just ship all the

00:41:42.753 --> 00:41:43.815
code right to production.

00:41:44.094 --> 00:41:46.199
Again, oversimplifying how it worked,

00:41:46.259 --> 00:41:47.762
but basically invented that

00:41:47.782 --> 00:41:48.483
concept as well.

00:41:49.460 --> 00:41:51.121
um the thing that she spent

00:41:51.141 --> 00:41:52.041
her whole career on after

00:41:52.061 --> 00:41:52.963
was really like systems

00:41:53.003 --> 00:41:54.844
programming and like this

00:41:54.884 --> 00:41:56.344
kind of like high level

00:41:56.364 --> 00:41:57.545
concept of error catching

00:41:58.425 --> 00:42:00.148
um because again she she

00:42:00.248 --> 00:42:01.429
really pushed them hard for

00:42:01.469 --> 00:42:03.309
like hey like things could

00:42:03.369 --> 00:42:05.831
happen in out of order the

00:42:05.871 --> 00:42:06.931
way the cpu doesn't expect

00:42:06.952 --> 00:42:07.713
it like things are going to

00:42:07.733 --> 00:42:10.054
happen um and actually the

00:42:10.255 --> 00:42:11.615
if you know from the apollo

00:42:11.635 --> 00:42:12.896
eleven landing there was

00:42:12.916 --> 00:42:13.976
like a program alarm

00:42:14.056 --> 00:42:15.257
happening at twelve oh one

00:42:15.778 --> 00:42:17.179
like if you like that right

00:42:17.239 --> 00:42:18.681
up until the moment they landed

00:42:19.936 --> 00:42:20.856
That was actually an alarm

00:42:20.876 --> 00:42:25.181
that she put in that they

00:42:25.282 --> 00:42:26.603
happened to have the radar

00:42:26.903 --> 00:42:27.784
configured incorrectly.

00:42:27.804 --> 00:42:29.025
Again, I don't know all the details,

00:42:29.045 --> 00:42:30.387
but the radar configured incorrectly,

00:42:30.847 --> 00:42:32.048
and it was bombarding the

00:42:32.128 --> 00:42:34.110
CPU with interrupts.

00:42:34.672 --> 00:42:35.813
But the CPU knew,

00:42:36.092 --> 00:42:38.596
because of this hierarchy

00:42:38.615 --> 00:42:39.295
that she put in,

00:42:40.097 --> 00:42:41.836
knew this is less important

00:42:41.856 --> 00:42:42.958
than the fact that we're landing,

00:42:43.018 --> 00:42:44.318
basically, again, oversimplifying,

00:42:44.998 --> 00:42:46.539
and was throwing this alarm like, hey,

00:42:46.659 --> 00:42:47.898
you're giving me too much data,

00:42:48.418 --> 00:42:49.838
but I'm going to ignore it,

00:42:50.380 --> 00:42:51.380
which was great because it

00:42:51.659 --> 00:42:52.719
actually allowed them to land.

00:42:54.041 --> 00:42:55.760
And so this program alarm

00:42:55.780 --> 00:42:56.561
that was going off was

00:42:56.581 --> 00:42:58.181
actually an early concept

00:42:58.242 --> 00:43:02.682
of being able to have

00:43:02.862 --> 00:43:04.342
prioritization to the CPU

00:43:04.643 --> 00:43:06.543
cycles and stuff that she

00:43:06.563 --> 00:43:07.143
basically invented.

00:43:07.923 --> 00:43:08.704
Yeah, it was an early page.

00:43:08.784 --> 00:43:10.246
It was just like, and they knew,

00:43:10.266 --> 00:43:10.766
they like got it.

00:43:10.786 --> 00:43:11.927
They're like, oh yeah, ignore that page.

00:43:11.947 --> 00:43:12.327
It's fine.

00:43:12.407 --> 00:43:14.909
Just keep playing.

00:43:16.148 --> 00:43:17.230
So yeah.

00:43:17.250 --> 00:43:19.130
So I could talk about her for a lot of,

00:43:19.490 --> 00:43:20.032
a long time.

00:43:20.052 --> 00:43:21.072
Awesome.

00:43:21.393 --> 00:43:21.693
Yeah.

00:43:22.472 --> 00:43:23.014
Yeah, that's fun.

00:43:24.153 --> 00:43:24.434
Well,

00:43:24.954 --> 00:43:27.856
do you have any sort of thoughts that

00:43:27.876 --> 00:43:29.117
you would like to leave

00:43:29.557 --> 00:43:32.699
folks with or any links or

00:43:32.739 --> 00:43:33.639
projects to share that

00:43:33.659 --> 00:43:34.420
you're excited about?

00:43:35.577 --> 00:43:37.398
Yeah, I mean, I'd love folks to like,

00:43:37.498 --> 00:43:38.438
if you go to my website,

00:43:38.478 --> 00:43:39.878
it's got a lot of things

00:43:39.898 --> 00:43:41.059
that I've been thinking

00:43:41.079 --> 00:43:42.380
about and writing about.

00:43:42.460 --> 00:43:46.402
It's boleary.dev.

00:43:46.442 --> 00:43:47.541
But yeah, I mean,

00:43:47.681 --> 00:43:49.043
I think generally the thing

00:43:49.083 --> 00:43:49.902
that I'm trying to get

00:43:50.103 --> 00:43:52.403
folks to do is understand that, you know,

00:43:52.463 --> 00:43:54.405
if you have tried, you know,

00:43:54.545 --> 00:43:55.324
AI in the past,

00:43:55.344 --> 00:43:56.706
you're a software engineer, you tried AI,

00:43:56.746 --> 00:43:58.246
you know, six, twelve,

00:43:58.266 --> 00:44:00.847
eighteen months ago and kind of, again,

00:44:00.867 --> 00:44:02.447
are just tabbing in cursor

00:44:02.487 --> 00:44:04.009
and not really leveraging it.

00:44:04.789 --> 00:44:05.971
I'd encourage you to try it again.

00:44:06.030 --> 00:44:07.353
Again, it doesn't have to be Kilo.

00:44:07.373 --> 00:44:10.396
There's a lot of options out there.

00:44:11.597 --> 00:44:13.880
I just, I think that, again,

00:44:13.920 --> 00:44:14.621
I don't think we're going

00:44:14.641 --> 00:44:15.782
to get replaced by AI,

00:44:15.802 --> 00:44:18.684
but I do think people that get good at

00:44:19.211 --> 00:44:20.193
programming and pair

00:44:20.213 --> 00:44:21.333
programming with ai are

00:44:21.353 --> 00:44:23.074
going to replace uh those

00:44:23.094 --> 00:44:24.315
of us that have been around

00:44:24.335 --> 00:44:25.396
you know since the early

00:44:25.416 --> 00:44:27.898
days um and so I just

00:44:27.938 --> 00:44:28.719
really encourage people to

00:44:28.760 --> 00:44:30.922
try it there's lots of um

00:44:31.262 --> 00:44:32.182
you know free tiers out

00:44:32.202 --> 00:44:34.123
there it's not you know

00:44:34.143 --> 00:44:34.925
vibe I'm trying to make

00:44:35.025 --> 00:44:37.126
vibe engineering a term uh

00:44:37.487 --> 00:44:38.708
to combat you know vibe

00:44:38.748 --> 00:44:39.528
coding you know like yes

00:44:39.568 --> 00:44:41.590
vibe coding is is really it

00:44:41.710 --> 00:44:42.590
actually it has a lot of

00:44:42.630 --> 00:44:43.791
purposes I think it's great

00:44:43.811 --> 00:44:44.492
at prototyping

00:44:45.777 --> 00:44:46.358
It's great when you just

00:44:46.398 --> 00:44:46.978
need something that you

00:44:46.998 --> 00:44:47.940
know you're going to throw away.

00:44:48.860 --> 00:44:50.141
But I think vibe engineering

00:44:50.181 --> 00:44:51.181
is going to be here to stay.

00:44:51.222 --> 00:44:51.722
And so I would really

00:44:51.782 --> 00:44:52.583
encourage people to think

00:44:52.603 --> 00:44:54.724
about how can I vibe with

00:44:54.804 --> 00:44:56.684
AI to do real engineering.

00:44:58.286 --> 00:45:00.907
And that's kilocode.ai?

00:45:00.927 --> 00:45:01.407
That's correct.

00:45:01.447 --> 00:45:01.668
Yeah.

00:45:01.807 --> 00:45:04.130
K-I-L-O.

00:45:04.429 --> 00:45:06.070
As in like kilos of code.

00:45:07.487 --> 00:45:08.748
idea yeah so it actually

00:45:08.789 --> 00:45:11.550
like is yeah yeah yeah I

00:45:11.590 --> 00:45:13.132
was trying my our ceo

00:45:13.152 --> 00:45:15.353
doesn't love it he he uh I

00:45:15.373 --> 00:45:16.114
was trying to make don't

00:45:16.135 --> 00:45:17.054
code by the line code by

00:45:17.076 --> 00:45:17.976
the kilo happen and he

00:45:17.996 --> 00:45:19.898
doesn't love it anymore so

00:45:20.918 --> 00:45:21.759
but that's that's like the

00:45:21.798 --> 00:45:23.820
concept right um and he's a

00:45:23.840 --> 00:45:24.882
really energetic guy the

00:45:24.922 --> 00:45:26.463
last thing he did was um I

00:45:26.643 --> 00:45:27.302
don't know if you saw that

00:45:27.364 --> 00:45:29.505
vesuvius challenge it was

00:45:29.545 --> 00:45:30.626
where there was the scroll

00:45:30.666 --> 00:45:32.467
from mount vesuvius that was rolled up

00:45:33.648 --> 00:45:34.608
and they couldn't unroll it

00:45:34.628 --> 00:45:35.110
because it would just

00:45:35.349 --> 00:45:36.951
disintegrate so they

00:45:37.190 --> 00:45:39.313
x-rayed it um like kind of

00:45:39.333 --> 00:45:40.273
cross sections of it and

00:45:40.293 --> 00:45:41.134
then it was like can you

00:45:41.173 --> 00:45:43.255
from a rolled up scroll and

00:45:43.436 --> 00:45:45.476
x-rays and densities like

00:45:45.697 --> 00:45:46.878
read it while it's still

00:45:46.918 --> 00:45:47.778
rolled up that was that was

00:45:47.798 --> 00:45:48.480
the thing he did before

00:45:48.519 --> 00:45:49.581
kilo was he ran that

00:45:49.740 --> 00:45:50.740
challenge it was like you

00:45:50.760 --> 00:45:51.541
know one of those like oh

00:45:51.561 --> 00:45:52.262
it's a million dollar I

00:45:52.282 --> 00:45:52.882
can't remember what the

00:45:52.902 --> 00:45:53.744
prize was but like oh

00:45:53.764 --> 00:45:54.324
there's a million dollar

00:45:54.344 --> 00:45:55.125
challenge if you can do it

00:45:55.885 --> 00:45:57.085
um and he ran that so he's

00:45:57.847 --> 00:45:59.148
he's a fun guy he's got a lot of energy

00:46:01.286 --> 00:46:02.288
Is there a free tier?

00:46:02.307 --> 00:46:02.907
Is there a paid tier?

00:46:02.927 --> 00:46:03.449
How does that work?

00:46:03.469 --> 00:46:04.789
Because I'm going to go

00:46:04.809 --> 00:46:06.452
install it right after this.

00:46:06.512 --> 00:46:07.612
Definitely install it.

00:46:08.353 --> 00:46:09.094
It's open source.

00:46:09.315 --> 00:46:10.556
The extension is open source

00:46:10.576 --> 00:46:11.237
and free and going to be

00:46:11.277 --> 00:46:12.498
free forever because it's open source.

00:46:13.719 --> 00:46:18.202
We have with it our Kilo Code API provider,

00:46:18.603 --> 00:46:19.103
where again,

00:46:19.143 --> 00:46:21.806
we're just connecting you to the models,

00:46:21.847 --> 00:46:23.628
to Claude, to Gemini.

00:46:24.688 --> 00:46:26.030
And so there is a free tier

00:46:26.050 --> 00:46:27.050
with that where we give you

00:46:27.070 --> 00:46:28.130
twenty dollars to get started.

00:46:29.231 --> 00:46:30.512
And then we don't take any

00:46:30.552 --> 00:46:31.793
commission on any top up.

00:46:31.853 --> 00:46:34.333
So this is a difference that, again,

00:46:34.393 --> 00:46:35.954
once you start to dive into this world,

00:46:35.974 --> 00:46:36.554
you start to find there's

00:46:36.574 --> 00:46:37.476
like there's a lot of great

00:46:37.496 --> 00:46:38.335
providers out there like

00:46:38.376 --> 00:46:39.577
Requestly and Open Router

00:46:39.597 --> 00:46:40.277
and these other things that

00:46:40.317 --> 00:46:41.498
can kind of like

00:46:42.605 --> 00:46:44.525
give you all of the models, right?

00:46:44.565 --> 00:46:46.768
Like T three chat is a very common one.

00:46:46.788 --> 00:46:48.449
Like we're not T three chat,

00:46:48.469 --> 00:46:49.989
but like we're giving you

00:46:50.070 --> 00:46:50.771
access to all the models

00:46:50.831 --> 00:46:52.150
and we don't take any commission on it.

00:46:52.190 --> 00:46:53.913
So that's where folks end up

00:46:53.952 --> 00:46:56.313
staying is if they start to

00:46:56.353 --> 00:46:57.074
get used to it and they

00:46:57.094 --> 00:46:58.094
don't want to pay

00:46:58.114 --> 00:46:59.175
commission to somebody else.

00:47:00.496 --> 00:47:02.137
But that's through our provider.

00:47:02.577 --> 00:47:03.139
When you install it,

00:47:03.159 --> 00:47:04.719
you also see there's a

00:47:04.760 --> 00:47:05.621
bunch of other providers.

00:47:05.760 --> 00:47:07.442
You can use a local, right?

00:47:07.461 --> 00:47:08.922
You can use like Olam locally.

00:47:09.702 --> 00:47:10.744
Like I've got a max two year here.

00:47:10.804 --> 00:47:11.985
Someday there'll be content

00:47:12.005 --> 00:47:12.804
for me about like

00:47:13.186 --> 00:47:15.867
completely off grid vibe engineering.

00:47:18.608 --> 00:47:19.429
Or you can use, you know,

00:47:19.469 --> 00:47:20.251
like if you have an

00:47:20.291 --> 00:47:21.371
existing anthropic key or

00:47:21.391 --> 00:47:22.913
you have Claude code already,

00:47:22.952 --> 00:47:24.152
like you can just use those

00:47:25.974 --> 00:47:27.376
and then just connect that to Kilo code.

00:47:27.815 --> 00:47:29.036
Like our plan to make money

00:47:29.077 --> 00:47:32.920
is with like enterprise and LDAP and SSO,

00:47:33.119 --> 00:47:33.840
those kinds of things that

00:47:33.860 --> 00:47:34.501
enterprises wants,

00:47:34.541 --> 00:47:36.322
we want it to be free for

00:47:36.342 --> 00:47:37.623
developers forever, so.

00:47:38.143 --> 00:47:38.663
Awesome.

00:47:38.862 --> 00:47:39.384
And that's the model I'm

00:47:39.463 --> 00:47:40.284
used to from GitLab.

00:47:40.304 --> 00:47:41.684
So I'm really excited about.

00:47:41.965 --> 00:47:42.664
Yeah.

00:47:42.704 --> 00:47:44.585
Well, thanks for coming on, Brendan.

00:47:45.306 --> 00:47:46.786
And where can people find

00:47:46.806 --> 00:47:47.646
you online if they have

00:47:47.706 --> 00:47:49.168
questions about you or

00:47:49.327 --> 00:47:50.128
anything you've done?

00:47:50.748 --> 00:47:50.929
Yeah.

00:47:50.949 --> 00:47:53.349
So I'm O'Leary Crew most places,

00:47:53.510 --> 00:47:55.490
O-L-E-A-R-Y-C-R-E-W.

00:47:56.170 --> 00:47:58.291
You can find me on basically

00:47:58.331 --> 00:47:59.211
any social media there.

00:47:59.251 --> 00:48:00.432
And then, yeah, boleary.dev,

00:48:01.032 --> 00:48:02.773
B-O-L-E-A-R-Y.dev.

00:48:03.409 --> 00:48:04.570
No apostrophe in any of that.

00:48:05.391 --> 00:48:06.010
Next time I'm on,

00:48:06.030 --> 00:48:06.650
I can tell you all about

00:48:06.771 --> 00:48:08.132
living life in the age of

00:48:08.152 --> 00:48:08.632
computers with an

00:48:08.652 --> 00:48:09.472
apostrophe in your name.

00:48:09.733 --> 00:48:10.353
You can imagine.

00:48:10.472 --> 00:48:11.532
Bobby Tables tells you a lot

00:48:11.552 --> 00:48:12.393
about what that looks like.

00:48:12.414 --> 00:48:12.554
Yeah.

00:48:12.574 --> 00:48:15.315
Awesome.

00:48:15.735 --> 00:48:16.755
Well, thanks again.

00:48:17.036 --> 00:48:19.117
And yeah,

00:48:20.197 --> 00:48:21.157
thank you to everyone on the

00:48:21.177 --> 00:48:22.157
internet who's listening to

00:48:22.197 --> 00:48:23.077
this or watching this.

00:48:23.338 --> 00:48:24.378
And we'll see you next time.