Listen now | Recording of AURA - Juha Korpela & Shane Gibson - 28 August 2026

WEBVTT
Kind: captions
Language: en
00:00:00.880 --> 00:00:02.950
Welcome to another Aura session. Good to
00:00:02.960 --> 00:00:03.909
see you again.
00:00:03.919 --> 00:00:05.990
>> Hey Shane, good to be here.
00:00:06.000 --> 00:00:09.110
>> Hey. And uh yeah, what's top of mind
00:00:09.120 --> 00:00:10.870
over the last month?
00:00:10.880 --> 00:00:13.270
>> Oh dear. Data things of all sorts of
00:00:13.280 --> 00:00:17.029
kinds. Yeah. Yeah. And I mean lots of
00:00:17.039 --> 00:00:19.670
things again still going on with with
00:00:19.680 --> 00:00:22.230
context and knowledge crafts and you
00:00:22.240 --> 00:00:24.070
know all of that stuff. It's just you
00:00:24.080 --> 00:00:26.950
know my brain's buzzing all the time.
00:00:26.960 --> 00:00:29.589
>> Yeah. I've been kind of working less on
00:00:29.599 --> 00:00:31.269
the contact stuff at the moment and more
00:00:31.279 --> 00:00:33.670
on the last mile kind of build. So,
00:00:33.680 --> 00:00:35.990
we've got some delivery stuff. Um, so
00:00:36.000 --> 00:00:38.310
deep in the in the harness and building
00:00:38.320 --> 00:00:41.590
out design systems and watching Claude
00:00:41.600 --> 00:00:43.110
behave really really well for a while
00:00:43.120 --> 00:00:45.750
and for some unknown reason completely
00:00:45.760 --> 00:00:48.150
go off the rails and act like a
00:00:48.160 --> 00:00:50.069
2-year-old. So, uh, apparently we're
00:00:50.079 --> 00:00:54.150
getting I think a Opus 51's coming out
00:00:54.160 --> 00:00:56.869
this week or very soon the rumor is. And
00:00:56.879 --> 00:00:58.389
hopefully that'll fix it because I'm
00:00:58.399 --> 00:01:01.349
back to Opus 4.8 because uh it actually
00:01:01.359 --> 00:01:03.189
does what's on the box. Yeah. So, all
00:01:03.199 --> 00:01:04.950
right, let's kick off with context.
00:01:04.960 --> 00:01:06.390
Well, you want to we the one we kind of
00:01:06.400 --> 00:01:08.070
chatted about LinkedIn and all the other
00:01:08.080 --> 00:01:11.510
ones is this idea of uh big and small in
00:01:11.520 --> 00:01:13.429
terms of work and focus. So, which way
00:01:13.439 --> 00:01:15.109
do you want to go? Big and small first
00:01:15.119 --> 00:01:16.230
or context?
00:01:16.240 --> 00:01:17.670
>> Let's let's do the big and small thing
00:01:17.680 --> 00:01:19.510
because I think it also relates to this
00:01:19.520 --> 00:01:24.230
whole context discussion. So, uh yeah,
00:01:24.240 --> 00:01:26.550
>> let's start with that. Um, so shall I
00:01:26.560 --> 00:01:28.550
basically explain what my sort of thesis
00:01:28.560 --> 00:01:29.270
is? Yeah,
00:01:29.280 --> 00:01:29.910
>> do it.
00:01:29.920 --> 00:01:32.550
>> Uh, I think these were maybe some of the
00:01:32.560 --> 00:01:33.990
first articles that I wrote on my
00:01:34.000 --> 00:01:37.670
Substack last year and um I called them
00:01:37.680 --> 00:01:39.510
kind of the small problem and the big
00:01:39.520 --> 00:01:42.230
problem and this is a pattern as you are
00:01:42.240 --> 00:01:44.870
a fan of patterns, right? So, so, uh,
00:01:44.880 --> 00:01:48.389
I've come to notice something that kind
00:01:48.399 --> 00:01:50.550
of keeps repeating itself in various
00:01:50.560 --> 00:01:51.830
areas, you know, be it data
00:01:51.840 --> 00:01:54.310
architecture, uh, sort of delivery
00:01:54.320 --> 00:01:57.350
discussions or, you know, um, context
00:01:57.360 --> 00:01:58.550
and knowledge management and so on and
00:01:58.560 --> 00:02:01.910
so forth that we uh
00:02:01.920 --> 00:02:04.069
must operate on two different levels and
00:02:04.079 --> 00:02:06.230
and and the the levels are basically
00:02:06.240 --> 00:02:07.910
defined by the problem that we are
00:02:07.920 --> 00:02:11.029
trying to solve and I think the vast
00:02:11.039 --> 00:02:13.670
majority of data organizations are
00:02:13.680 --> 00:02:15.510
focused on solving what I like to call
00:02:15.520 --> 00:02:17.750
the small problem not because it's a
00:02:17.760 --> 00:02:19.670
small thing or in terms of you know
00:02:19.680 --> 00:02:21.510
importance or anything but because the
00:02:21.520 --> 00:02:23.990
scope of that solution that we are
00:02:24.000 --> 00:02:26.070
creating for the problem is very small
00:02:26.080 --> 00:02:28.470
and the small problem is that individual
00:02:28.480 --> 00:02:31.750
project individual you know initiative
00:02:31.760 --> 00:02:33.750
whatever that might be uh wants to
00:02:33.760 --> 00:02:36.229
deliver something valuable as fast and
00:02:36.239 --> 00:02:38.630
efficiently as possible. So this is the
00:02:38.640 --> 00:02:39.990
obvious problem that everyone's trying
00:02:40.000 --> 00:02:42.710
to figure out all the time, right? But
00:02:42.720 --> 00:02:45.670
then uh in an any any organization
00:02:45.680 --> 00:02:47.910
larger than I don't know 20 people, you
00:02:47.920 --> 00:02:50.470
will have multiple people, multiple
00:02:50.480 --> 00:02:53.270
groups of people solving their own small
00:02:53.280 --> 00:02:55.830
problems kind of in parallel across the
00:02:55.840 --> 00:02:58.309
organization. Right? So then as the
00:02:58.319 --> 00:03:00.949
scale goes up we have the big problem
00:03:00.959 --> 00:03:03.350
which is again big because of its scope
00:03:03.360 --> 00:03:05.670
which is how do we create circus
00:03:05.680 --> 00:03:08.630
circumstances in the organization across
00:03:08.640 --> 00:03:10.869
the whole organization where all of
00:03:10.879 --> 00:03:13.750
those teams can consistently and you
00:03:13.760 --> 00:03:17.110
know safely and and uh efficiently keep
00:03:17.120 --> 00:03:20.869
delivering valuable results and the the
00:03:20.879 --> 00:03:23.670
distinction is very very kind of minute.
00:03:23.680 --> 00:03:26.550
It's it's just about uh at at the at the
00:03:26.560 --> 00:03:28.149
small level it's the individual
00:03:28.159 --> 00:03:29.910
solution, the individual team doing its
00:03:29.920 --> 00:03:32.390
individual thing and then at the kind of
00:03:32.400 --> 00:03:34.149
big problem level we're thinking all
00:03:34.159 --> 00:03:36.070
right well now there's a hundred teams
00:03:36.080 --> 00:03:38.390
or you know 20 people or whatever how do
00:03:38.400 --> 00:03:40.390
we make sure that all of them can
00:03:40.400 --> 00:03:43.190
consistently keep delivering and and the
00:03:43.200 --> 00:03:46.470
the tasks and the questions the tasks
00:03:46.480 --> 00:03:48.470
that we must do the questions we must
00:03:48.480 --> 00:03:49.990
ask are actually completely different
00:03:50.000 --> 00:03:52.789
than these and I keep running into this
00:03:52.799 --> 00:03:56.869
problem where uh I see teams doing local
00:03:56.879 --> 00:03:58.869
optimization at the kind of small
00:03:58.879 --> 00:04:01.110
problem area and then there's no one
00:04:01.120 --> 00:04:03.190
doing kind of global optimization at the
00:04:03.200 --> 00:04:06.949
big problem area and traditionally uh
00:04:06.959 --> 00:04:08.229
we've been talking of course about
00:04:08.239 --> 00:04:09.990
enterprise architecture and whatnot but
00:04:10.000 --> 00:04:12.309
I think that there the issue has been
00:04:12.319 --> 00:04:14.630
that uh many people haven't really
00:04:14.640 --> 00:04:16.469
focused on the feedback loops between
00:04:16.479 --> 00:04:19.749
these things that we need to be able to
00:04:19.759 --> 00:04:22.550
you know uh first of all from the people
00:04:22.560 --> 00:04:24.390
solving the small problems we have to be
00:04:24.400 --> 00:04:26.950
able to take learnings out of there and
00:04:26.960 --> 00:04:29.110
make them reusable and you know update
00:04:29.120 --> 00:04:30.550
our patterns and templates and
00:04:30.560 --> 00:04:32.230
components and whatnot. But on the other
00:04:32.240 --> 00:04:34.550
hand, anything that we do on the kind of
00:04:34.560 --> 00:04:37.270
big big problem level if it doesn't
00:04:37.280 --> 00:04:39.590
practically help those individuals and
00:04:39.600 --> 00:04:41.110
teams working at the small small
00:04:41.120 --> 00:04:43.430
problems then it's useless. So so there
00:04:43.440 --> 00:04:45.110
kind of has to be a con continuous
00:04:45.120 --> 00:04:46.710
feedback loop going between those two
00:04:46.720 --> 00:04:48.629
levels of people solving different
00:04:48.639 --> 00:04:50.950
problems. And I'm just not seeing enough
00:04:50.960 --> 00:04:53.270
people solving the big problems. I'm
00:04:53.280 --> 00:04:54.790
seeing a lot of vendors, a lot of
00:04:54.800 --> 00:04:56.870
consultants, lots of advisors, a lot of
00:04:56.880 --> 00:04:59.990
data leaders just focusing on individual
00:05:00.000 --> 00:05:02.710
small problems one at a time. So that's
00:05:02.720 --> 00:05:03.590
the thesis.
00:05:03.600 --> 00:05:05.350
>> Oh, so that that yeah, we're not going
00:05:05.360 --> 00:05:08.629
to get into context, are we? So there's
00:05:08.639 --> 00:05:11.029
a ton in there. Let's uh let's
00:05:11.039 --> 00:05:13.430
unpack that a little bit. So um so one
00:05:13.440 --> 00:05:14.950
of the one of the ones was on LinkedIn.
00:05:14.960 --> 00:05:17.590
Um there was uh one of the guys that's
00:05:17.600 --> 00:05:19.670
in the data mesh data product community
00:05:19.680 --> 00:05:22.150
out of Europe. I think he was was at
00:05:22.160 --> 00:05:24.230
when we were there. Um she had a
00:05:24.240 --> 00:05:27.350
question of uh if you've got distributed
00:05:27.360 --> 00:05:31.430
teams uh do they even need a modeling
00:05:31.440 --> 00:05:33.830
pattern you know and and so we won't go
00:05:33.840 --> 00:05:35.670
into the whole you model whether you
00:05:35.680 --> 00:05:37.510
think you do or not. The question is not
00:05:37.520 --> 00:05:39.510
whether you model it's whether you
00:05:39.520 --> 00:05:42.070
consciously or unconsciously do it. But
00:05:42.080 --> 00:05:45.189
it comes back to this idea of if we have
00:05:45.199 --> 00:05:47.189
truly distributed and decentralized
00:05:47.199 --> 00:05:50.230
teams uh and they're responsible end to
00:05:50.240 --> 00:05:54.150
end uh then actually we probably don't
00:05:54.160 --> 00:05:57.430
want any big solutions. We want silos,
00:05:57.440 --> 00:05:59.990
right? We want them to be completely uh
00:06:00.000 --> 00:06:02.230
independent. Uh and what that means
00:06:02.240 --> 00:06:05.029
though is we get no reuse across teams.
00:06:05.039 --> 00:06:06.870
>> Yes. and and we have to be really clear
00:06:06.880 --> 00:06:08.390
that we're going to be on this
00:06:08.400 --> 00:06:09.830
completely distributed because what I
00:06:09.840 --> 00:06:11.189
see like you is we're going to be
00:06:11.199 --> 00:06:12.550
completely distributed but they have to
00:06:12.560 --> 00:06:14.070
use the same platform or we're going to
00:06:14.080 --> 00:06:16.070
be completely distributed but they have
00:06:16.080 --> 00:06:19.350
to talk to each other and so Spotify
00:06:19.360 --> 00:06:20.469
when they first came out with the
00:06:20.479 --> 00:06:23.189
Spotify model uh it was completely
00:06:23.199 --> 00:06:25.110
decentralized teams right distributed do
00:06:25.120 --> 00:06:28.150
what uh whatever you want and then what
00:06:28.160 --> 00:06:30.629
they found was over time humans
00:06:30.639 --> 00:06:34.629
naturally uh align themselves to reuse
00:06:34.639 --> 00:06:36.629
So new squad would stand up, new people
00:06:36.639 --> 00:06:38.629
come in, they had to manage some kind of
00:06:38.639 --> 00:06:40.550
ticket system and they go ask every
00:06:40.560 --> 00:06:42.469
other team what do you use and in their
00:06:42.479 --> 00:06:45.749
case it was Jira. Um but they naturally
00:06:45.759 --> 00:06:47.670
would go oh yeah that would work or they
00:06:47.680 --> 00:06:49.270
wouldn't right they go I would go put
00:06:49.280 --> 00:06:50.469
something completely different in and
00:06:50.479 --> 00:06:52.230
then the other thing they found is over
00:06:52.240 --> 00:06:55.670
time teams became product teams for
00:06:55.680 --> 00:07:00.070
those big solutions. So when 20 squads
00:07:00.080 --> 00:07:02.070
were using Jira,
00:07:02.080 --> 00:07:03.270
those squads were like, I don't want
00:07:03.280 --> 00:07:04.629
have to maintain this right? I got
00:07:04.639 --> 00:07:06.230
other things more important. So they
00:07:06.240 --> 00:07:08.629
became a squad which maintained that
00:07:08.639 --> 00:07:11.749
platform, that product, right? Um but I
00:07:11.759 --> 00:07:12.950
think what happens first of all is
00:07:12.960 --> 00:07:14.230
organizations say, "Yeah, we're going to
00:07:14.240 --> 00:07:15.749
be decentralized, distributed, and then
00:07:15.759 --> 00:07:18.790
want to behave in and we're still going
00:07:18.800 --> 00:07:20.070
to reuse, we're still going to control,
00:07:20.080 --> 00:07:22.390
we still got to manage." Second problem
00:07:22.400 --> 00:07:24.629
is the if we're not completely
00:07:24.639 --> 00:07:27.029
distributed the other alternative is big
00:07:27.039 --> 00:07:29.430
upfront you know enterprise architecture
00:07:29.440 --> 00:07:30.150
>> yes
00:07:30.160 --> 00:07:31.749
>> ontologies
00:07:31.759 --> 00:07:35.270
um data model enterprise data models uh
00:07:35.280 --> 00:07:37.990
build a plat data platform for two years
00:07:38.000 --> 00:07:41.110
as the only piece of value so that it's
00:07:41.120 --> 00:07:43.670
ready and we know that in the current
00:07:43.680 --> 00:07:47.749
world that time to market that speed um
00:07:47.759 --> 00:07:49.350
is no longer acceptable for most
00:07:49.360 --> 00:07:51.110
organizations.
00:07:51.120 --> 00:07:53.430
So we get stuck in this messy middle
00:07:53.440 --> 00:07:56.550
right and then the question is what
00:07:56.560 --> 00:07:59.670
stance do you take for that messy middle
00:07:59.680 --> 00:08:01.589
and more and more so so I used to do
00:08:01.599 --> 00:08:02.869
when I worked with some teams a while
00:08:02.879 --> 00:08:05.189
ago we tried we the problem we had was
00:08:05.199 --> 00:08:08.950
um fast versus slow development so we're
00:08:08.960 --> 00:08:09.990
trying to say well can we have a
00:08:10.000 --> 00:08:11.589
contouge in the middle and if we get a
00:08:11.599 --> 00:08:13.029
request from a stakeholder and it looks
00:08:13.039 --> 00:08:14.950
like it's a nice small one or a bau
00:08:14.960 --> 00:08:17.189
change this is 10 years ago which is why
00:08:17.199 --> 00:08:19.830
I'm using the word bau um go through the
00:08:19.840 --> 00:08:22.950
fast path Right. Uh and then uh if it's
00:08:22.960 --> 00:08:24.390
a bigger piece of work, it's almost like
00:08:24.400 --> 00:08:25.990
a product, you know, we got to do some
00:08:26.000 --> 00:08:28.710
collection, some design, some build,
00:08:28.720 --> 00:08:30.629
then it goes through the the product or
00:08:30.639 --> 00:08:33.750
the slow path. Um and then what we found
00:08:33.760 --> 00:08:35.509
is people cheated, right? They go, "Oh,
00:08:35.519 --> 00:08:37.670
well, the slow path six months." So, no,
00:08:37.680 --> 00:08:41.029
this is a small problem. Um we had to
00:08:41.039 --> 00:08:43.190
say to a team, "Oh, you're on the Bau
00:08:43.200 --> 00:08:45.910
team for ever, you know, doing
00:08:45.920 --> 00:08:48.070
maintenance work." which meant when the
00:08:48.080 --> 00:08:49.509
people doing the product builds were
00:08:49.519 --> 00:08:50.949
building, they were handing over
00:08:50.959 --> 00:08:52.310
that never worked because they didn't
00:08:52.320 --> 00:08:53.910
have to maintain it. They were like,
00:08:53.920 --> 00:08:56.070
"No, no, you have to maintain it." Uh,
00:08:56.080 --> 00:08:57.430
and then people going, "I don't want to
00:08:57.440 --> 00:08:58.870
be on the BAU team for the rest of my
00:08:58.880 --> 00:09:00.710
life. I want to do cool shit." So then
00:09:00.720 --> 00:09:02.949
we started moving teams around, squads
00:09:02.959 --> 00:09:04.949
around, right? You b this week and all
00:09:04.959 --> 00:09:07.910
that. And and what we were really doing
00:09:07.920 --> 00:09:09.350
is kind of trying to patch over the
00:09:09.360 --> 00:09:11.030
problems. and and where I've got to and
00:09:11.040 --> 00:09:12.870
I'll be interested in your view now is
00:09:12.880 --> 00:09:15.509
is I think more about pioneers and town
00:09:15.519 --> 00:09:17.430
builders.
00:09:17.440 --> 00:09:19.110
If we want to do something that has a
00:09:19.120 --> 00:09:23.430
high area of uncertainty or a need to go
00:09:23.440 --> 00:09:25.829
really fast, then we just pioneer it and
00:09:25.839 --> 00:09:28.310
that's a set of very siloed behaviors.
00:09:28.320 --> 00:09:31.269
There are minimal principles, right? Um
00:09:31.279 --> 00:09:33.110
minimal policies, just do whatever you
00:09:33.120 --> 00:09:34.870
need to get it done. And then we want to
00:09:34.880 --> 00:09:35.910
figure out how we pick up that
00:09:35.920 --> 00:09:37.430
pioneering capability and move it back
00:09:37.440 --> 00:09:40.230
into town planning. Um, which means we
00:09:40.240 --> 00:09:42.949
may rebuild it, right? Or we may port it
00:09:42.959 --> 00:09:44.470
or we may go, "Shit, that was really
00:09:44.480 --> 00:09:46.870
good. Let's make that a product, right?"
00:09:46.880 --> 00:09:48.310
And and that's the last point I wanted
00:09:48.320 --> 00:09:49.670
to pick up because you kind of mentioned
00:09:49.680 --> 00:09:52.470
it. When we're if we're building big
00:09:52.480 --> 00:09:54.389
things, we have to build them for
00:09:54.399 --> 00:09:56.550
somebody else to use, not ourselves.
00:09:56.560 --> 00:09:58.550
Yeah. So, we're effectively building a a
00:09:58.560 --> 00:10:01.110
product to go, "This is going to make
00:10:01.120 --> 00:10:03.509
other people faster, right? Their their
00:10:03.519 --> 00:10:05.750
small solutions will actually just reuse
00:10:05.760 --> 00:10:08.630
the big one because it's so easy. Why
00:10:08.640 --> 00:10:11.350
wouldn't they? Yeah. Um that's kind of
00:10:11.360 --> 00:10:13.990
my thought, right? But it's that
00:10:14.000 --> 00:10:15.990
balance, right? And and I struggle with
00:10:16.000 --> 00:10:18.150
ontologies right now because I go
00:10:18.160 --> 00:10:20.150
>> I mean I just I just got hit with it to
00:10:20.160 --> 00:10:21.670
this week, right? Where I was like,
00:10:21.680 --> 00:10:24.069
"Yep, I made the mistake again and it's
00:10:24.079 --> 00:10:27.829
killing me." Um but it's this idea of
00:10:27.839 --> 00:10:29.990
big upfront you know big platform design
00:10:30.000 --> 00:10:32.389
but enterprise architecture big data
00:10:32.399 --> 00:10:33.990
modeling they're all really really
00:10:34.000 --> 00:10:37.030
valuable but in the new world they take
00:10:37.040 --> 00:10:39.350
too long so your thoughts
00:10:39.360 --> 00:10:42.870
>> yeah no no I agree on on the sentiment
00:10:42.880 --> 00:10:45.590
big upfront is not the right u solution
00:10:45.600 --> 00:10:47.990
here but think of uh team topologies you
00:10:48.000 --> 00:10:49.670
you've read the book right
00:10:49.680 --> 00:10:53.430
>> so so uh we have um what are they called
00:10:53.440 --> 00:10:55.190
streamlined teams
00:10:55.200 --> 00:10:56.389
>> which are basically what we are now
00:10:56.399 --> 00:10:58.069
talking about the individual teams
00:10:58.079 --> 00:10:59.990
working on individual small problems
00:11:00.000 --> 00:11:01.829
right and then we have all the other
00:11:01.839 --> 00:11:03.750
teams basically platform team now being
00:11:03.760 --> 00:11:06.710
the kind of key here that uh the idea in
00:11:06.720 --> 00:11:08.230
team topologies is that from a
00:11:08.240 --> 00:11:11.190
technology perspective the big problem
00:11:11.200 --> 00:11:13.350
of figuring out that everyone has the
00:11:13.360 --> 00:11:15.750
capability to build efficiently and you
00:11:15.760 --> 00:11:17.750
know reuse things and use shared
00:11:17.760 --> 00:11:19.670
components and whatnot there's separate
00:11:19.680 --> 00:11:21.590
team that takes care of that completely
00:11:21.600 --> 00:11:23.509
as you said in in your Spotify examples
00:11:23.519 --> 00:11:25.430
and so and so forth. So eventually
00:11:25.440 --> 00:11:27.350
someone will take start taking care of
00:11:27.360 --> 00:11:28.710
development tools or whatnot. This is
00:11:28.720 --> 00:11:30.470
basically organically moving towards
00:11:30.480 --> 00:11:32.870
kind of platform team versus streamlined
00:11:32.880 --> 00:11:36.230
teams structure, right? And I think it's
00:11:36.240 --> 00:11:38.550
not just technology that needs that. I
00:11:38.560 --> 00:11:40.710
mean it already if you do that kind of
00:11:40.720 --> 00:11:42.710
actual division between kind of product
00:11:42.720 --> 00:11:44.389
teams and platform teams that's already
00:11:44.399 --> 00:11:46.069
a huge improvement for many kind of
00:11:46.079 --> 00:11:48.069
traditional IT organizations where
00:11:48.079 --> 00:11:49.430
everyone is basically just doing
00:11:49.440 --> 00:11:52.230
solutions all the time. But uh actual
00:11:52.240 --> 00:11:54.230
platform services are basically products
00:11:54.240 --> 00:11:57.509
as you said to the other teams.
00:11:57.519 --> 00:11:59.910
But uh I think this goes wider than just
00:11:59.920 --> 00:12:01.430
technology architecture and platforms.
00:12:01.440 --> 00:12:02.949
And now I'm getting to the ontology
00:12:02.959 --> 00:12:05.590
point. Uh for a few years now when I've
00:12:05.600 --> 00:12:08.629
been uh in various organizations doing
00:12:08.639 --> 00:12:10.389
data modeling and setting up kind of
00:12:10.399 --> 00:12:12.949
data modeling practices and ways of
00:12:12.959 --> 00:12:14.470
working and roles, you know, teaching
00:12:14.480 --> 00:12:17.350
people how to do it. I've used basically
00:12:17.360 --> 00:12:19.269
the same pattern in in terms of data
00:12:19.279 --> 00:12:21.829
models that there should be a capability
00:12:21.839 --> 00:12:24.710
or or you know um enough understanding
00:12:24.720 --> 00:12:27.350
and knowledge that individual teams can
00:12:27.360 --> 00:12:29.750
model their own things as they see fit
00:12:29.760 --> 00:12:31.590
you know in their own schedule you know
00:12:31.600 --> 00:12:33.750
whatever that might be but at the same
00:12:33.760 --> 00:12:36.230
time there must be someone looking at
00:12:36.240 --> 00:12:39.509
the you know overall picture of models
00:12:39.519 --> 00:12:42.550
uh in the organization who is able to a
00:12:42.560 --> 00:12:45.350
uh if necessary uh tell other teams that
00:12:45.360 --> 00:12:46.949
Okay, you're actually uh doing something
00:12:46.959 --> 00:12:48.629
here that we have checked this earlier
00:12:48.639 --> 00:12:51.030
that this is not so uh you must go in
00:12:51.040 --> 00:12:52.230
this direction this is the right
00:12:52.240 --> 00:12:55.829
direction for you i.e kind of uh taking
00:12:55.839 --> 00:12:57.910
information from the top the enterprise
00:12:57.920 --> 00:13:00.710
level back down to individual teams but
00:13:00.720 --> 00:13:03.509
at the same time we must build a system
00:13:03.519 --> 00:13:05.430
which allows a process which allows
00:13:05.440 --> 00:13:08.150
these individual teams to basically keep
00:13:08.160 --> 00:13:11.269
expanding the enterprise scope. It's not
00:13:11.279 --> 00:13:12.870
so that there's separate enterprise
00:13:12.880 --> 00:13:13.910
modeling that models the whole
00:13:13.920 --> 00:13:15.509
enterprise and then separate project
00:13:15.519 --> 00:13:16.870
modeling that models individual
00:13:16.880 --> 00:13:19.110
projects. It's a feedback loop again in
00:13:19.120 --> 00:13:21.829
both directions. Projects must increase
00:13:21.839 --> 00:13:23.590
the amount of knowledge in the
00:13:23.600 --> 00:13:27.190
enterprise level. Now I think that with
00:13:27.200 --> 00:13:29.110
you know forget about data models part
00:13:29.120 --> 00:13:31.269
in particular for a moment and this
00:13:31.279 --> 00:13:33.190
whole thing applies also to ontologies
00:13:33.200 --> 00:13:35.269
and all this knowledge management work.
00:13:35.279 --> 00:13:37.829
If a project discovers a piece of
00:13:37.839 --> 00:13:39.910
information, you know, semantics about
00:13:39.920 --> 00:13:41.670
whatever,
00:13:41.680 --> 00:13:43.990
how do we make that piece of information
00:13:44.000 --> 00:13:45.829
available to everyone else, how do we
00:13:45.839 --> 00:13:48.230
pull it from the project level up to the
00:13:48.240 --> 00:13:49.590
enterprise level so that the next
00:13:49.600 --> 00:13:51.670
project can again reuse it? And I think
00:13:51.680 --> 00:13:54.310
that kind of process is is what we
00:13:54.320 --> 00:13:56.069
really need.
00:13:56.079 --> 00:13:58.470
>> Decision tracing, right? Can't explain.
00:13:58.480 --> 00:14:00.069
Um,
00:14:00.079 --> 00:14:02.790
>> so but human nature, right? So by
00:14:02.800 --> 00:14:05.829
default, most humans are relatively lazy
00:14:05.839 --> 00:14:07.350
in a good way.
00:14:07.360 --> 00:14:07.670
>> True.
00:14:07.680 --> 00:14:09.670
>> Yeah. Yeah. So it's like, well, if I
00:14:09.680 --> 00:14:12.629
don't have to do the work, then I'm not
00:14:12.639 --> 00:14:14.069
going to do the work, right? I'll just
00:14:14.079 --> 00:14:17.110
use something that's fit for purpose.
00:14:17.120 --> 00:14:19.829
>> But data people and technologists
00:14:19.839 --> 00:14:23.110
are builders by default and tend to not
00:14:23.120 --> 00:14:24.629
reuse something else if they can build
00:14:24.639 --> 00:14:25.750
it themselves. Yeah.
00:14:25.760 --> 00:14:28.550
>> Right. So, we kind of got this this this
00:14:28.560 --> 00:14:33.110
problem of two behaviors. And then if we
00:14:33.120 --> 00:14:35.750
treat the the big piece of work, right,
00:14:35.760 --> 00:14:39.509
as a review cycle, it's too late,
00:14:39.519 --> 00:14:40.949
>> right? It's like, well, I'm going to go
00:14:40.959 --> 00:14:42.790
review your work until you you kind of
00:14:42.800 --> 00:14:44.629
got it wrong. But actually, you're
00:14:44.639 --> 00:14:45.990
you're almost near the end. Your
00:14:46.000 --> 00:14:47.430
stakeholder is not going to give you
00:14:47.440 --> 00:14:49.430
permission to spend another two weeks to
00:14:49.440 --> 00:14:53.269
to refactor it, right? Uh and so that's
00:14:53.279 --> 00:14:55.509
one of the problems we have to solve.
00:14:55.519 --> 00:14:58.069
And so my my view is with the GNAI
00:14:58.079 --> 00:15:00.389
toolings now we start to get technology
00:15:00.399 --> 00:15:02.710
that can help us solve that socio
00:15:02.720 --> 00:15:04.310
problem
00:15:04.320 --> 00:15:08.069
because everybody's using LLMs well not
00:15:08.079 --> 00:15:11.030
everybody sorry lots of people are using
00:15:11.040 --> 00:15:14.389
LLM and GI in their work now and they're
00:15:14.399 --> 00:15:16.870
using it in a ask AI pattern right I ask
00:15:16.880 --> 00:15:18.470
a question I get a response can you
00:15:18.480 --> 00:15:20.389
check this for me hey I'm going to chat
00:15:20.399 --> 00:15:23.269
to you and just get some feedback if
00:15:23.279 --> 00:15:27.110
those tools are actually bound to our
00:15:27.120 --> 00:15:29.269
context plane, right? To the knowledge
00:15:29.279 --> 00:15:32.949
and and um uh content that we have in
00:15:32.959 --> 00:15:34.870
our organization,
00:15:34.880 --> 00:15:36.790
then it's going to help the person doing
00:15:36.800 --> 00:15:38.949
the work go, "Oh, you do realize that
00:15:38.959 --> 00:15:41.110
actually over there it's already been
00:15:41.120 --> 00:15:43.030
done. Like, why don't you reuse that?"
00:15:43.040 --> 00:15:44.629
versus Claude going, "Yeah, I'll write
00:15:44.639 --> 00:15:47.110
that for you." Um, it's going to
00:15:47.120 --> 00:15:48.870
hopefully say, "Well, you're just
00:15:48.880 --> 00:15:51.110
defining marketing active customer, but
00:15:51.120 --> 00:15:53.269
actually it's over there. It's been
00:15:53.279 --> 00:15:55.749
reused." Um, sorry, it's over there.
00:15:55.759 --> 00:15:58.069
It's being used. It's been validated.
00:15:58.079 --> 00:16:00.389
Why don't you use that? And then
00:16:00.399 --> 00:16:03.829
potentially policies that order it.
00:16:03.839 --> 00:16:06.781
Okay. You know, hey, you Shane got
00:16:06.791 --> 00:16:07.749
[clears throat] told that marketing
00:16:07.759 --> 00:16:09.430
active customer was over there and then
00:16:09.440 --> 00:16:11.350
he went and created his own one with the
00:16:11.360 --> 00:16:14.069
same term. Um,
00:16:14.079 --> 00:16:15.590
but the problem with that is we're
00:16:15.600 --> 00:16:16.870
building these little pieces and then
00:16:16.880 --> 00:16:18.389
we're having to fix things. And and I
00:16:18.399 --> 00:16:19.749
got a classic example, right? And this
00:16:19.759 --> 00:16:22.949
is the one that nailed me last week and
00:16:22.959 --> 00:16:25.110
and I knew it and that's the worst part
00:16:25.120 --> 00:16:26.870
right where you go I knew it. So uh
00:16:26.880 --> 00:16:29.910
doing some work around uh payments. Um
00:16:29.920 --> 00:16:31.829
so I went through grabbed a glossery
00:16:31.839 --> 00:16:33.670
right like a payment arrangement,
00:16:33.680 --> 00:16:36.310
payment installment, that kind of thing.
00:16:36.320 --> 00:16:40.230
Um, and because of the iterate iterative
00:16:40.240 --> 00:16:42.870
way we work, it's kind of like get an
00:16:42.880 --> 00:16:44.870
early cut of it, test it with the
00:16:44.880 --> 00:16:46.949
stakeholders, subject matter experts,
00:16:46.959 --> 00:16:49.350
looks good, get the data, test it with
00:16:49.360 --> 00:16:50.790
that, give them something back to
00:16:50.800 --> 00:16:53.829
actually see, and then, you know, go,
00:16:53.839 --> 00:16:55.350
yeah, you know, that state change where
00:16:55.360 --> 00:16:56.629
we said we went from the end to the air,
00:16:56.639 --> 00:16:59.749
it's not quite true, right? Like, um,
00:16:59.759 --> 00:17:01.670
and so one of the ones is is I use the
00:17:01.680 --> 00:17:04.470
term broken arrangement, right? uh and I
00:17:04.480 --> 00:17:07.510
use the term payment and then I went
00:17:07.520 --> 00:17:09.189
actually there is no payment's an
00:17:09.199 --> 00:17:10.870
overarching term. There is no term
00:17:10.880 --> 00:17:12.150
payment. Yeah, there's a payment
00:17:12.160 --> 00:17:13.110
arrangement, there's a payment
00:17:13.120 --> 00:17:15.110
arrangement installment, there's a
00:17:15.120 --> 00:17:16.870
payment attempt, there's a decline
00:17:16.880 --> 00:17:20.069
payment, there's a um yeah, a whole
00:17:20.079 --> 00:17:21.510
bunch of payment states, right? But
00:17:21.520 --> 00:17:23.189
payment is never a term I should have
00:17:23.199 --> 00:17:25.829
used and broken arrangement is never a
00:17:25.839 --> 00:17:27.590
term I should have used because it's all
00:17:27.600 --> 00:17:29.270
around a payment arrangement. The
00:17:29.280 --> 00:17:30.470
problem was I'd already built some
00:17:30.480 --> 00:17:31.669
stuff.
00:17:31.679 --> 00:17:34.710
>> Yeah. And I'm like, "What do I do?"
00:17:34.720 --> 00:17:37.190
Right? And and you go, "Yeah, I've got a
00:17:37.200 --> 00:17:38.549
delivery. I've got to get something
00:17:38.559 --> 00:17:41.430
back, right? We want another iteration."
00:17:41.440 --> 00:17:43.990
Claude naturally says to me, "Oh, we'll
00:17:44.000 --> 00:17:45.750
just change it in the front thing they
00:17:45.760 --> 00:17:47.909
see, right? We'll go broken arrangement
00:17:47.919 --> 00:17:50.470
becomes broken payment arrangement."
00:17:50.480 --> 00:17:53.350
And I'm sitting there going, "Well, no."
00:17:53.360 --> 00:17:56.470
because I know that if the field in the
00:17:56.480 --> 00:17:57.990
table is not called broken payment
00:17:58.000 --> 00:18:01.590
arrangement at some stage you an MCP
00:18:01.600 --> 00:18:03.110
something an agent is going to go look
00:18:03.120 --> 00:18:04.630
at that and it's going to get confused
00:18:04.640 --> 00:18:07.270
right uh so I ended up paying that tax
00:18:07.280 --> 00:18:09.430
right I ended up spending time in tokens
00:18:09.440 --> 00:18:12.150
going and rebuilding it but the reason
00:18:12.160 --> 00:18:15.029
is because we're a twoerson company and
00:18:15.039 --> 00:18:19.430
I know my incentive is I can't have that
00:18:19.440 --> 00:18:21.350
effort and work turn up in a month's
00:18:21.360 --> 00:18:24.310
time yeah I can't spend a cycle where it
00:18:24.320 --> 00:18:27.110
gets it wrong and I don't know why. So
00:18:27.120 --> 00:18:29.029
I'm incentive to go back and do my
00:18:29.039 --> 00:18:33.270
rework. Um if we've got separate teams
00:18:33.280 --> 00:18:36.230
is the you know is incentive.
00:18:36.240 --> 00:18:37.350
>> Yeah. Is the platform team really
00:18:37.360 --> 00:18:39.350
incentive? Is the building team the
00:18:39.360 --> 00:18:40.950
pioneering? Yeah. So again I think
00:18:40.960 --> 00:18:42.390
that's part of the problem. What's your
00:18:42.400 --> 00:18:43.350
thoughts?
00:18:43.360 --> 00:18:45.590
>> Yeah. Yeah. I I was using actually an
00:18:45.600 --> 00:18:47.909
example because uh I use Google
00:18:47.919 --> 00:18:50.150
Workspace for my all sorts of my running
00:18:50.160 --> 00:18:52.789
my entire business basically and uh
00:18:52.799 --> 00:18:55.590
Google Workspace has the Google contacts
00:18:55.600 --> 00:18:58.230
thing that you know runs your email
00:18:58.240 --> 00:19:00.070
addresses and everything and I
00:19:00.080 --> 00:19:02.789
religiously use that. So I I have all my
00:19:02.799 --> 00:19:05.029
contacts in there. All my you know all
00:19:05.039 --> 00:19:06.950
my phone contacts are synchronized and
00:19:06.960 --> 00:19:09.669
emails and whatnot. And uh every time I
00:19:09.679 --> 00:19:12.390
send a new email the um basically it
00:19:12.400 --> 00:19:15.510
goes directly as a new contact. Google
00:19:15.520 --> 00:19:17.909
contacts has this merge and fix
00:19:17.919 --> 00:19:20.310
functionality that when I open the
00:19:20.320 --> 00:19:22.150
contacts kind of UI which actually I
00:19:22.160 --> 00:19:25.190
rarely need to open. The first thing I
00:19:25.200 --> 00:19:27.669
face there is a suggestion uh that okay
00:19:27.679 --> 00:19:29.350
you could merge and fix the following
00:19:29.360 --> 00:19:30.870
contacts and it basically goes through
00:19:30.880 --> 00:19:32.390
my contact list and says okay you know
00:19:32.400 --> 00:19:34.310
here's the same name with different uh
00:19:34.320 --> 00:19:35.909
two different emails two different
00:19:35.919 --> 00:19:38.310
contacts you have the it's just the uh
00:19:38.320 --> 00:19:39.990
you know the domain in the email is is
00:19:40.000 --> 00:19:41.590
different and this guy has phone number
00:19:41.600 --> 00:19:43.270
that doesn't are these the same person
00:19:43.280 --> 00:19:44.789
would we just you know merge them I say
00:19:44.799 --> 00:19:47.510
yeah let's merge them that's the kind of
00:19:47.520 --> 00:19:49.750
work that happens I think at the sort of
00:19:49.760 --> 00:19:51.830
or should happen at the enterprise level
00:19:51.840 --> 00:19:54.710
basically that uh we have we need to
00:19:54.720 --> 00:19:56.390
have people who are able to do sort of
00:19:56.400 --> 00:19:59.029
janitorial management work uh
00:19:59.039 --> 00:20:02.549
consolidating things at enterprise level
00:20:02.559 --> 00:20:05.350
but I uh I agree with you that uh it
00:20:05.360 --> 00:20:07.510
can't be a process where you need to
00:20:07.520 --> 00:20:09.990
kind of wait for a review I mean the
00:20:10.000 --> 00:20:12.870
review can't be a uh showstopper for the
00:20:12.880 --> 00:20:14.789
project that's that's impossible we we
00:20:14.799 --> 00:20:17.190
all know that those EA gates or whatever
00:20:17.200 --> 00:20:20.150
they they will never work uh and also uh
00:20:20.160 --> 00:20:23.270
it can't be the only method of of kind
00:20:23.280 --> 00:20:24.870
injecting information from the
00:20:24.880 --> 00:20:27.350
enterprise level into a project. But uh
00:20:27.360 --> 00:20:29.510
if if this merge and fix operation
00:20:29.520 --> 00:20:31.110
that's kind of continuous work that
00:20:31.120 --> 00:20:32.870
happens at the enterprise level and as
00:20:32.880 --> 00:20:35.430
you said maybe if we are then able to
00:20:35.440 --> 00:20:38.149
feed that information to you know LLMs
00:20:38.159 --> 00:20:40.470
and so on so forth at the project level
00:20:40.480 --> 00:20:42.390
then you know when you initiate a
00:20:42.400 --> 00:20:44.390
initiate a new project you say oh I'm
00:20:44.400 --> 00:20:46.230
going to be working now on you know
00:20:46.240 --> 00:20:48.070
payments agreements whatever that might
00:20:48.080 --> 00:20:51.990
be then your fancy AI friend helper
00:20:52.000 --> 00:20:54.470
might uh come down and say oh you know
00:20:54.480 --> 00:20:56.630
well that's cool here's the domain main
00:20:56.640 --> 00:21:00.070
model for for payments or adjacent items
00:21:00.080 --> 00:21:01.909
to already discovered. You know, here's
00:21:01.919 --> 00:21:03.750
the documentation that has been created.
00:21:03.760 --> 00:21:05.750
Here's what we know already. These are
00:21:05.760 --> 00:21:07.270
the patterns. These are the dos. These
00:21:07.280 --> 00:21:10.470
are the don'ts. And uh that could be
00:21:10.480 --> 00:21:14.549
kind of a pull from the project rather
00:21:14.559 --> 00:21:16.950
than push from the above.
00:21:16.960 --> 00:21:18.549
>> I think you're right. I think part of it
00:21:18.559 --> 00:21:20.870
is before we had technology that could
00:21:20.880 --> 00:21:22.950
help us with this, it was always humans
00:21:22.960 --> 00:21:23.590
only.
00:21:23.600 --> 00:21:27.590
>> Yeah. it was always too late. You know,
00:21:27.600 --> 00:21:29.669
data quality committee or a data
00:21:29.679 --> 00:21:32.870
stewards committee where I go and do the
00:21:32.880 --> 00:21:34.950
work. I define the payment arrange
00:21:34.960 --> 00:21:37.029
payment terms. I then write a document.
00:21:37.039 --> 00:21:39.110
I then get it on to the next meeting.
00:21:39.120 --> 00:21:40.710
That meeting's busy because it's been
00:21:40.720 --> 00:21:43.350
full. So, I'm two meetings out. Um they
00:21:43.360 --> 00:21:45.110
look at it, they have a question. Oh,
00:21:45.120 --> 00:21:48.310
what about this one? You know, um then
00:21:48.320 --> 00:21:49.270
we have to go back and do some more
00:21:49.280 --> 00:21:51.990
work. and the feedback loop, the time to
00:21:52.000 --> 00:21:54.549
go from I need to be able to define and
00:21:54.559 --> 00:21:57.830
use this to this is good enough to go
00:21:57.840 --> 00:22:01.430
forward um is too long, right? And and
00:22:01.440 --> 00:22:04.789
it and so we just kind of go fast. We do
00:22:04.799 --> 00:22:08.149
point solutions uh because we can't wait
00:22:08.159 --> 00:22:10.470
for the proper process. Um
00:22:10.480 --> 00:22:12.870
>> yeah but but but let me just challenge
00:22:12.880 --> 00:22:14.870
the the the weight and and everything
00:22:14.880 --> 00:22:17.590
because uh what I find weird here is
00:22:17.600 --> 00:22:21.270
that obviously the the argument for um
00:22:21.280 --> 00:22:23.750
for rejecting the kind of enterprise
00:22:23.760 --> 00:22:26.870
level uh architecture or or knowledge
00:22:26.880 --> 00:22:28.549
management or anything is that the
00:22:28.559 --> 00:22:30.789
individual projects can't wait. But
00:22:30.799 --> 00:22:33.990
isn't the point the point is not just to
00:22:34.000 --> 00:22:36.070
um drop out every kind of external
00:22:36.080 --> 00:22:37.510
influence so that the project goes
00:22:37.520 --> 00:22:40.470
faster. it is to inject beneficial
00:22:40.480 --> 00:22:43.029
information into the project that it
00:22:43.039 --> 00:22:45.350
goes faster. So, so the feedback loop
00:22:45.360 --> 00:22:46.870
coming down from the enterprise level
00:22:46.880 --> 00:22:48.870
into the project. The point of that is
00:22:48.880 --> 00:22:51.350
not to slow down the project. It is to
00:22:51.360 --> 00:22:54.070
avoid the project reinventing the wheel,
00:22:54.080 --> 00:22:55.590
doing the same things again in many
00:22:55.600 --> 00:22:57.990
different ways and and uh having to
00:22:58.000 --> 00:23:00.630
learn uh the hard way the lessons that
00:23:00.640 --> 00:23:03.190
someone else has already learned. So I
00:23:03.200 --> 00:23:04.710
think you know when it comes to data
00:23:04.720 --> 00:23:06.870
governance ontologies knowledge
00:23:06.880 --> 00:23:08.549
management data models architecture
00:23:08.559 --> 00:23:10.310
whatever I think if we we we just need
00:23:10.320 --> 00:23:12.390
to turn the whole question around how
00:23:12.400 --> 00:23:14.390
can the enterprise level information
00:23:14.400 --> 00:23:16.950
that we've gathered help the projects go
00:23:16.960 --> 00:23:19.909
faster not just how do we minimize
00:23:19.919 --> 00:23:21.590
enterprise level involvement in
00:23:21.600 --> 00:23:23.510
individual projects
00:23:23.520 --> 00:23:25.029
>> and so for me as soon as we say
00:23:25.039 --> 00:23:26.549
enterprise we do exactly what you did
00:23:26.559 --> 00:23:28.310
right we go enterprise up here and
00:23:28.320 --> 00:23:30.470
people down projects down here
00:23:30.480 --> 00:23:32.630
>> I see it more as principles policies and
00:23:32.640 --> 00:23:33.830
patents.
00:23:33.840 --> 00:23:35.430
>> Yeah. Because they are what an
00:23:35.440 --> 00:23:37.510
enterprise is made up of if you kind of
00:23:37.520 --> 00:23:41.110
take it back to its core patents. Um and
00:23:41.120 --> 00:23:44.789
then uh Jorgen upper took team
00:23:44.799 --> 00:23:47.110
topologies and he took Spotify and a
00:23:47.120 --> 00:23:49.190
bunch of those other Spotify model and
00:23:49.200 --> 00:23:50.870
he created a thing called unfix which
00:23:50.880 --> 00:23:53.110
was more detailed and I really liked it.
00:23:53.120 --> 00:23:55.830
It's kind of gone now um because he he
00:23:55.840 --> 00:23:58.149
kind of went to the next level of patent
00:23:58.159 --> 00:24:00.630
behavior amongst humans which resonated
00:24:00.640 --> 00:24:01.990
with me. So one of the things you talk
00:24:02.000 --> 00:24:03.669
about and maybe it was in technology I
00:24:03.679 --> 00:24:06.310
can't remember um is when you
00:24:06.320 --> 00:24:08.950
effectively parachute an expert into a
00:24:08.960 --> 00:24:10.390
squad for a period of time to bring
00:24:10.400 --> 00:24:12.870
their expertise that's a really valuable
00:24:12.880 --> 00:24:14.630
pattern right they get access to that
00:24:14.640 --> 00:24:16.310
knowledge that expertise that way of
00:24:16.320 --> 00:24:18.710
working those sets of patterns but that
00:24:18.720 --> 00:24:21.430
person's going in to make that squad
00:24:21.440 --> 00:24:23.350
more mature and more more you know
00:24:23.360 --> 00:24:25.269
increase their expertise uh in that
00:24:25.279 --> 00:24:26.789
thing and then they go and help somebody
00:24:26.799 --> 00:24:30.070
else and so for me and and that's the
00:24:30.080 --> 00:24:31.110
feedback loop because they're going to
00:24:31.120 --> 00:24:32.549
be watching what their team are doing
00:24:32.559 --> 00:24:34.070
and go, "Oh, actually that's really
00:24:34.080 --> 00:24:35.110
really useful. I should have thought
00:24:35.120 --> 00:24:36.070
about that." And then they're going to
00:24:36.080 --> 00:24:37.350
bring it back into the toolkit
00:24:37.360 --> 00:24:40.149
effectively. And so that feedback loop,
00:24:40.159 --> 00:24:42.070
right, where we take policies, patterns,
00:24:42.080 --> 00:24:44.230
and um principles from an
00:24:44.240 --> 00:24:45.990
organizationwide
00:24:46.000 --> 00:24:47.990
and then bring those into those point
00:24:48.000 --> 00:24:50.070
solutions, I think that's really
00:24:50.080 --> 00:24:51.029
valuable. But there's got to be a
00:24:51.039 --> 00:24:52.149
feedback loop
00:24:52.159 --> 00:24:53.830
>> and it's got to be fast, right? That
00:24:53.840 --> 00:24:55.350
process should give us agility. It
00:24:55.360 --> 00:24:58.549
should allow us to make change at faster
00:24:58.559 --> 00:25:03.430
pace with less risk. Yeah. Um so that's
00:25:03.440 --> 00:25:04.789
what we should be aiming for. So I'm
00:25:04.799 --> 00:25:07.669
with you. It just I hardly ever see it,
00:25:07.679 --> 00:25:09.350
right? I I mean you see it with a two
00:25:09.360 --> 00:25:11.669
person team because you have no choice.
00:25:11.679 --> 00:25:15.590
You have a 100 person organization then
00:25:15.600 --> 00:25:17.510
we naturally get silos and we get silos
00:25:17.520 --> 00:25:19.029
in the organization. There's the finance
00:25:19.039 --> 00:25:21.430
team, the marketing team. Yeah. We don't
00:25:21.440 --> 00:25:23.350
break up the teams and reform them every
00:25:23.360 --> 00:25:24.630
six months.
00:25:24.640 --> 00:25:25.909
um which would be a really interesting
00:25:25.919 --> 00:25:26.870
idea.
00:25:26.880 --> 00:25:29.510
>> Yeah. I mean I did a little exercise
00:25:29.520 --> 00:25:31.909
with the client some time ago where uh
00:25:31.919 --> 00:25:36.230
we were discussing you know kind of data
00:25:36.240 --> 00:25:38.390
product management approach right that
00:25:38.400 --> 00:25:40.789
um what if we were to introduce kind of
00:25:40.799 --> 00:25:42.870
data product approach. Now without going
00:25:42.880 --> 00:25:44.470
too deeply into what a data product is
00:25:44.480 --> 00:25:46.630
but the idea was that we would kind of
00:25:46.640 --> 00:25:49.110
uh figure out what are sensible
00:25:49.120 --> 00:25:51.269
collections of things that are related
00:25:51.279 --> 00:25:53.909
to some specific user group doing a
00:25:53.919 --> 00:25:55.430
specific thing you know bunch of
00:25:55.440 --> 00:25:57.110
dashboards or you know files or whatever
00:25:57.120 --> 00:25:59.190
and then we would basically document
00:25:59.200 --> 00:26:00.789
that using a very very simple kind of
00:26:00.799 --> 00:26:03.510
metadata wrapper which would be the same
00:26:03.520 --> 00:26:05.750
metadata for every single data product
00:26:05.760 --> 00:26:07.669
and then there would be just one thing
00:26:07.679 --> 00:26:09.269
to document and it would be very easy
00:26:09.279 --> 00:26:11.029
and we could all list them in a catalog
00:26:11.039 --> 00:26:13.669
somewhere. list and by catalog I mean a
00:26:13.679 --> 00:26:15.430
catalog of these data products not a
00:26:15.440 --> 00:26:17.269
catalog of every single component within
00:26:17.279 --> 00:26:19.110
them making everything that much easier
00:26:19.120 --> 00:26:21.110
because the granularity level level is
00:26:21.120 --> 00:26:23.590
is much easier to handle. So uh there
00:26:23.600 --> 00:26:24.789
was a lot of push back from the
00:26:24.799 --> 00:26:26.630
technical side because they said oh no
00:26:26.640 --> 00:26:28.789
you know oh we can't we can't possibly
00:26:28.799 --> 00:26:30.950
document anything in this organization
00:26:30.960 --> 00:26:32.470
or you know it's it's very difficult and
00:26:32.480 --> 00:26:33.990
it slows us down we just want to code
00:26:34.000 --> 00:26:36.710
and everything and uh when we discussed
00:26:36.720 --> 00:26:39.029
with the project managers and the the
00:26:39.039 --> 00:26:40.870
kind of business counterparts and and
00:26:40.880 --> 00:26:43.110
team leads in the technical teams and
00:26:43.120 --> 00:26:45.669
and they were also kind of very sort of
00:26:45.679 --> 00:26:48.310
not very excited at first but then then
00:26:48.320 --> 00:26:49.990
they tried it out. We said, "Okay, you
00:26:50.000 --> 00:26:52.950
know, here's the 10 little cells in an
00:26:52.960 --> 00:26:55.269
Excel file that you you would fill to
00:26:55.279 --> 00:26:57.269
document your data product. Just try
00:26:57.279 --> 00:26:59.110
it." And they spent like, you know, 30
00:26:59.120 --> 00:27:00.549
minutes on each product. And they said,
00:27:00.559 --> 00:27:02.149
"You know, actually, well, it was quite
00:27:02.159 --> 00:27:04.549
easy." And and uh we went through the
00:27:04.559 --> 00:27:07.269
motions and and and tried everything uh
00:27:07.279 --> 00:27:08.390
you know, where do we put them, how do
00:27:08.400 --> 00:27:10.390
we catalog things? And the end result we
00:27:10.400 --> 00:27:12.310
then after some time we asked for
00:27:12.320 --> 00:27:15.830
feedback and these uh new newly minted
00:27:15.840 --> 00:27:17.590
product team themes they said there's
00:27:17.600 --> 00:27:19.669
simply no other way which would make
00:27:19.679 --> 00:27:21.510
things any easier that this is like the
00:27:21.520 --> 00:27:23.990
only way that makes any sense. It this
00:27:24.000 --> 00:27:27.430
is simply so kind of practical and such
00:27:27.440 --> 00:27:30.070
an easy way to make things coherent make
00:27:30.080 --> 00:27:32.390
us our products visible to understand
00:27:32.400 --> 00:27:34.070
other people's data products. Why would
00:27:34.080 --> 00:27:36.789
we do anything else?
00:27:36.799 --> 00:27:40.230
And that was just a sort of um little
00:27:40.240 --> 00:27:42.630
kind of twist in their minds that they
00:27:42.640 --> 00:27:45.590
had to go through that it's not that a
00:27:45.600 --> 00:27:48.070
shared pattern would slow you down. It's
00:27:48.080 --> 00:27:49.590
something that might actually speed you
00:27:49.600 --> 00:27:50.630
up.
00:27:50.640 --> 00:27:52.879
>> And that's the idea, right? Is reduce
00:27:52.889 --> 00:27:54.310
[clears throat] waste. So
00:27:54.320 --> 00:27:54.630
>> yeah,
00:27:54.640 --> 00:27:56.630
>> reduce waste and time, reduce waste and
00:27:56.640 --> 00:27:59.430
waiting, reduce waste and rework, reduce
00:27:59.440 --> 00:28:01.909
waste and handoff. Um I mean it's
00:28:01.919 --> 00:28:03.909
effectively bringing lean thinking into
00:28:03.919 --> 00:28:05.190
knowledge work, right? And
00:28:05.200 --> 00:28:07.350
>> it is. It is. And we often don't do that
00:28:07.360 --> 00:28:09.590
because we can't we can't see the
00:28:09.600 --> 00:28:11.750
knowledge work. We can watch a factory
00:28:11.760 --> 00:28:13.269
floor and we can see the machines. We
00:28:13.279 --> 00:28:15.510
can see the movement of parts and goods,
00:28:15.520 --> 00:28:17.190
right? We can see the humans. We can see
00:28:17.200 --> 00:28:19.590
the stoppage time. We can see the waste
00:28:19.600 --> 00:28:22.310
because that's what lean processing
00:28:22.320 --> 00:28:24.549
gives us, right? And knowledge working
00:28:24.559 --> 00:28:27.909
and data working. Often we can't. And
00:28:27.919 --> 00:28:29.590
again, it's not the people doing the
00:28:29.600 --> 00:28:32.549
work's fault because they're incented to
00:28:32.559 --> 00:28:34.950
take another juro ticket and get it out
00:28:34.960 --> 00:28:36.710
to the stakeholder, right? They're not
00:28:36.720 --> 00:28:38.870
incented to look at their way of
00:28:38.880 --> 00:28:41.029
working, to look at their technology and
00:28:41.039 --> 00:28:42.870
say, "Actually, I'm just going to reduce
00:28:42.880 --> 00:28:44.789
one piece of waste this week
00:28:44.799 --> 00:28:47.590
>> because it will pay us back." Um, and
00:28:47.600 --> 00:28:49.430
they're not incented to experiment,
00:28:49.440 --> 00:28:50.870
right? They're not experiment to say,
00:28:50.880 --> 00:28:53.830
"Try try a spreadsheet with 10 fields.
00:28:53.840 --> 00:28:55.669
>> See if it improves the process or makes
00:28:55.679 --> 00:28:57.590
it worse. If it improves it, keep it. If
00:28:57.600 --> 00:29:00.310
it doesn't, uh, throw it away, right?
00:29:00.320 --> 00:29:02.470
Try something else. But, you know, it's
00:29:02.480 --> 00:29:05.269
a wasteful part of the process. So, uh,
00:29:05.279 --> 00:29:06.710
go fix it.
00:29:06.720 --> 00:29:09.750
>> Yeah. But but I think and I've worked in
00:29:09.760 --> 00:29:11.990
corporate IT for many many years and you
00:29:12.000 --> 00:29:14.310
know as part of it and consulting in
00:29:14.320 --> 00:29:16.149
there and you know all sorts of things
00:29:16.159 --> 00:29:18.630
and uh I have come to the conclusion
00:29:18.640 --> 00:29:20.710
that the traditional sort of IT service
00:29:20.720 --> 00:29:24.230
management mindset is really really bad
00:29:24.240 --> 00:29:26.549
for data analytics work in particular
00:29:26.559 --> 00:29:28.549
probably for other kinds of work as well
00:29:28.559 --> 00:29:31.590
but uh in particular for data analytics
00:29:31.600 --> 00:29:34.950
work because uh it turns data teams into
00:29:34.960 --> 00:29:37.350
ticket factories as you you just grab
00:29:37.360 --> 00:29:39.110
the next ticket, you do the thing, you
00:29:39.120 --> 00:29:40.789
grab the next ticket and the only thing
00:29:40.799 --> 00:29:42.549
that the kind of IT service management
00:29:42.559 --> 00:29:45.350
mindset seems to be uh trying to
00:29:45.360 --> 00:29:47.669
optimize there is the speed of
00:29:47.679 --> 00:29:49.830
individual tickets uh kind of going
00:29:49.840 --> 00:29:51.990
through the system and that might
00:29:52.000 --> 00:29:54.710
involve oh let's switch from data bricks
00:29:54.720 --> 00:29:56.710
to snowflake or you know whatever that
00:29:56.720 --> 00:29:58.389
might be where you're just trying to get
00:29:58.399 --> 00:30:00.389
individual digs going through faster
00:30:00.399 --> 00:30:02.310
without paying attention to kind of
00:30:02.320 --> 00:30:04.230
overall data architecture or anything
00:30:04.240 --> 00:30:06.630
because that's not part of service
00:30:06.640 --> 00:30:09.750
management. Your service is the service
00:30:09.760 --> 00:30:13.029
where you uh you offer these data
00:30:13.039 --> 00:30:14.549
engineers or whatever to your business
00:30:14.559 --> 00:30:16.870
units so that they can send in tickets
00:30:16.880 --> 00:30:18.310
and you are just promising that we'll
00:30:18.320 --> 00:30:19.669
have lots of engineers and they will
00:30:19.679 --> 00:30:22.070
respond very quickly to each ticket.
00:30:22.080 --> 00:30:24.470
Your service actually is not what it
00:30:24.480 --> 00:30:26.230
should be. The what service that should
00:30:26.240 --> 00:30:28.149
be is we will maintain a sensible data
00:30:28.159 --> 00:30:30.389
architecture for this organization as a
00:30:30.399 --> 00:30:32.389
whole. meaning the kind of big problem
00:30:32.399 --> 00:30:35.269
and not just uh trying to optimize the
00:30:35.279 --> 00:30:37.350
inflow of these small problems all the
00:30:37.360 --> 00:30:41.029
time and I'm not sure it might not be u
00:30:41.039 --> 00:30:43.590
an inherent problem of it or something
00:30:43.600 --> 00:30:46.630
like that but I kind of feel like it is
00:30:46.640 --> 00:30:48.549
that the organizations who operate in
00:30:48.559 --> 00:30:51.029
this way are somehow almost
00:30:51.039 --> 00:30:54.710
fundamentally incapable of of uh kind of
00:30:54.720 --> 00:30:57.029
thinking about data work as a bigger
00:30:57.039 --> 00:30:59.269
picture it's just you know individual
00:30:59.279 --> 00:31:00.470
tickets
00:31:00.480 --> 00:31:01.750
>> I I think one of the the most
00:31:01.760 --> 00:31:03.510
embarrassing parts of that whole process
00:31:03.520 --> 00:31:06.470
though is um nobody can tell you how
00:31:06.480 --> 00:31:07.750
long it takes.
00:31:07.760 --> 00:31:08.230
>> Yeah, true.
00:31:08.240 --> 00:31:10.549
>> Like nobody actually checks the ITIL
00:31:10.559 --> 00:31:12.950
based ticketing system to see uh how
00:31:12.960 --> 00:31:14.789
long it takes from a request landing to
00:31:14.799 --> 00:31:17.990
being served. Um and I don't have a
00:31:18.000 --> 00:31:20.310
problem if it's a true factory, but
00:31:20.320 --> 00:31:23.190
bring in lean and that kind of thinking,
00:31:23.200 --> 00:31:25.190
right? Because that's what you're
00:31:25.200 --> 00:31:27.590
operating, but we don't, right? We don't
00:31:27.600 --> 00:31:29.669
we don't manage cycle times. Yeah. We
00:31:29.679 --> 00:31:31.669
don't manage down times. We don't track
00:31:31.679 --> 00:31:33.750
the handoffs. Those kind of things. But
00:31:33.760 --> 00:31:35.830
it kind of reminds me there was a post
00:31:35.840 --> 00:31:38.310
um on LinkedIn and and I kind of made a
00:31:38.320 --> 00:31:41.350
comment I think or I rep anyway. Um and
00:31:41.360 --> 00:31:44.710
it came back to uh the chief data and
00:31:44.720 --> 00:31:46.630
analytics officer. Let's just say chief
00:31:46.640 --> 00:31:48.789
data officer, right? Where should they
00:31:48.799 --> 00:31:51.029
sit in an organization? Should they sit
00:31:51.039 --> 00:31:53.430
under the CIO? Oh, it was A&Z, right?
00:31:53.440 --> 00:31:54.549
Had a role. That's right. In New
00:31:54.559 --> 00:31:57.350
Zealand. Uh and it was Yeah. the head of
00:31:57.360 --> 00:31:59.509
data sat under the CIO, but they had a
00:31:59.519 --> 00:32:02.070
group head of data and analytics. Um,
00:32:02.080 --> 00:32:04.230
and it made me think, right, I don't I
00:32:04.240 --> 00:32:07.110
don't think the the data I think the
00:32:07.120 --> 00:32:09.269
data and analytics should be at the top
00:32:09.279 --> 00:32:10.789
table, right? I think actually it's that
00:32:10.799 --> 00:32:13.430
part of the strategy of an organization
00:32:13.440 --> 00:32:16.789
is how's data and um AI used to achieve
00:32:16.799 --> 00:32:18.470
the corporate strategy. I think that's a
00:32:18.480 --> 00:32:20.630
top table conversation right now. Um,
00:32:20.640 --> 00:32:22.950
but the second one, the more important
00:32:22.960 --> 00:32:25.029
for now is is where should they sit,
00:32:25.039 --> 00:32:28.310
right? Should the head of data sit under
00:32:28.320 --> 00:32:31.257
the CIO, the CTO? What do you think?
00:32:31.267 --> 00:32:31.269
[snorts]
00:32:31.279 --> 00:32:34.310
>> Yeah, I I I Yeah, I mean this is an
00:32:34.320 --> 00:32:36.470
age-old question, isn't it? And this
00:32:36.480 --> 00:32:38.470
this has been probably a question ever
00:32:38.480 --> 00:32:40.470
since the first person came up with the
00:32:40.480 --> 00:32:42.389
title of CDO or head of data or
00:32:42.399 --> 00:32:43.509
anything.
00:32:43.519 --> 00:32:46.230
uh I don't think it should be in under
00:32:46.240 --> 00:32:48.870
CIO but not necessarily for the reason
00:32:48.880 --> 00:32:51.350
that that uh other people often use that
00:32:51.360 --> 00:32:53.430
you know you want to have the data
00:32:53.440 --> 00:32:55.190
analyst to have a seat in the top table
00:32:55.200 --> 00:32:56.230
or something like that. I don't think
00:32:56.240 --> 00:32:57.669
that's the reason. I I want to kind of
00:32:57.679 --> 00:33:00.389
go back to what I said about ITSM
00:33:00.399 --> 00:33:03.190
mentality and so on. I think the culture
00:33:03.200 --> 00:33:05.990
under CIO is wrong for data analytics
00:33:06.000 --> 00:33:08.230
work. The culture, the processes, the
00:33:08.240 --> 00:33:10.070
roles, the ways of working are simply
00:33:10.080 --> 00:33:12.950
wrong for data analytics work. And that
00:33:12.960 --> 00:33:16.630
is kind of big enterprises have sort of
00:33:16.640 --> 00:33:19.190
standardized how everything works under
00:33:19.200 --> 00:33:21.750
CIO. It's IT service management and
00:33:21.760 --> 00:33:24.710
that's it. And and data analytics if it
00:33:24.720 --> 00:33:27.190
is organically kind of uh sorry from
00:33:27.200 --> 00:33:30.549
organizationally uh put under CIO means
00:33:30.559 --> 00:33:33.509
that data analytics will also operate in
00:33:33.519 --> 00:33:35.110
either service management manner by
00:33:35.120 --> 00:33:36.950
default and then you have your kind of
00:33:36.960 --> 00:33:38.630
brave heroes who try to break out of
00:33:38.640 --> 00:33:40.389
that and change things. So they will
00:33:40.399 --> 00:33:42.870
constantly kind of swim against current.
00:33:42.880 --> 00:33:45.190
So I think that culturally it's
00:33:45.200 --> 00:33:47.990
important to have data analytics outside
00:33:48.000 --> 00:33:50.389
corporate IT. It's not because of
00:33:50.399 --> 00:33:53.269
importance of you know titles or or
00:33:53.279 --> 00:33:55.269
anything like that a pecking order. It's
00:33:55.279 --> 00:33:57.509
a question of organizational culture and
00:33:57.519 --> 00:33:59.110
I I don't say this because I would
00:33:59.120 --> 00:34:01.269
somehow hate corporate IT as I said I've
00:34:01.279 --> 00:34:02.789
been working there for many many years
00:34:02.799 --> 00:34:04.789
now in different roles. But I'm just
00:34:04.799 --> 00:34:06.789
thinking that this is not the way for
00:34:06.799 --> 00:34:09.589
data analytics because our job is to
00:34:09.599 --> 00:34:11.990
figure out together with the business
00:34:12.000 --> 00:34:14.629
how can they use data to um you know
00:34:14.639 --> 00:34:16.310
improve whatever the things are in
00:34:16.320 --> 00:34:18.389
business that they need to improve and
00:34:18.399 --> 00:34:20.790
that is contrary to the kind of ticket
00:34:20.800 --> 00:34:22.710
taking mentality where you just expect
00:34:22.720 --> 00:34:24.790
that someone will throw at me a ticket
00:34:24.800 --> 00:34:26.550
explaining what they will need to have
00:34:26.560 --> 00:34:28.790
done and then I'll just implement and
00:34:28.800 --> 00:34:30.550
then business value magically appears on
00:34:30.560 --> 00:34:31.990
the other end. That's not how anything
00:34:32.000 --> 00:34:33.589
works and this is why we can't deliver
00:34:33.599 --> 00:34:35.669
business value in data.
00:34:35.679 --> 00:34:37.270
>> I mean the interesting thing is CIO
00:34:37.280 --> 00:34:38.790
stands for chief information officer,
00:34:38.800 --> 00:34:39.030
right?
00:34:39.040 --> 00:34:40.069
>> Well, I mean that's that's the big
00:34:40.079 --> 00:34:40.869
irony, right?
00:34:40.879 --> 00:34:42.710
>> Not not chief technology officer. Um
00:34:42.720 --> 00:34:44.629
>> the whole it is information technology.
00:34:44.639 --> 00:34:45.990
Honestly, it's just T. There's very
00:34:46.000 --> 00:34:46.950
little eye in there.
00:34:46.960 --> 00:34:50.550
>> Yep. True. Uh I I I'll go back to why I
00:34:50.560 --> 00:34:51.909
think they should have a top table right
00:34:51.919 --> 00:34:54.550
at uh seat. And
00:34:54.560 --> 00:34:56.550
for me it's because they need to be as
00:34:56.560 --> 00:34:58.710
close to the conversation around the
00:34:58.720 --> 00:35:00.390
operation of the organization and the
00:35:00.400 --> 00:35:02.230
strategy of the organization that they
00:35:02.240 --> 00:35:05.030
can be. So chief financial officer sits
00:35:05.040 --> 00:35:06.230
there to understand where the
00:35:06.240 --> 00:35:08.150
organization's going and saying what's
00:35:08.160 --> 00:35:09.990
that going to do to our revenue expenses
00:35:10.000 --> 00:35:11.109
right like are we going to be
00:35:11.119 --> 00:35:13.270
financially viable what's our risk um
00:35:13.280 --> 00:35:15.430
from that chief operating officer sits
00:35:15.440 --> 00:35:17.109
there and says what do I need to change
00:35:17.119 --> 00:35:20.550
from a process point of view yeah uh
00:35:20.560 --> 00:35:22.710
head of HR you know what does this mean
00:35:22.720 --> 00:35:25.750
to our our resources no sorry our people
00:35:25.760 --> 00:35:27.190
right um
00:35:27.200 --> 00:35:31.589
>> and I think data and analytics now is is
00:35:31.599 --> 00:35:33.109
something that's important enough to sit
00:35:33.119 --> 00:35:35.270
there and go, okay, these conversations,
00:35:35.280 --> 00:35:36.710
what does that mean? How can data and
00:35:36.720 --> 00:35:39.589
analytics reduce waste or support that
00:35:39.599 --> 00:35:41.829
strategy, right? How can we make it uh
00:35:41.839 --> 00:35:44.710
achieve it? Um, and I think reading a
00:35:44.720 --> 00:35:46.230
strategy document or having it pass
00:35:46.240 --> 00:35:49.910
through a bunch of whispers um isn't
00:35:49.920 --> 00:35:51.750
isn't the right way to do it anymore.
00:35:51.760 --> 00:35:54.470
Um, that's why I believe it's it's it
00:35:54.480 --> 00:35:56.710
should be part of it. Um, but yeah, I
00:35:56.720 --> 00:35:57.750
agree with you. I think there's that
00:35:57.760 --> 00:36:00.550
culture thing. Um, if you're in a team
00:36:00.560 --> 00:36:01.990
that works in a certain way or part of
00:36:02.000 --> 00:36:03.510
an organization that works a certain
00:36:03.520 --> 00:36:05.910
way, it's incredibly difficult to to
00:36:05.920 --> 00:36:07.430
swim against the tide as you said, you
00:36:07.440 --> 00:36:09.349
know, the corporate I always called it
00:36:09.359 --> 00:36:11.190
the the corporate virus, right? It just
00:36:11.200 --> 00:36:13.430
comes and attacks you so often that you
00:36:13.440 --> 00:36:16.870
end up leaving or or complaining. Um, so
00:36:16.880 --> 00:36:18.390
yeah, I think that's that's a that's a
00:36:18.400 --> 00:36:20.310
good point. Uh let me take this
00:36:20.320 --> 00:36:21.910
discussion back towards kind of context
00:36:21.920 --> 00:36:23.270
and knowledge because this is also
00:36:23.280 --> 00:36:27.109
relevant to this topic that um now that
00:36:27.119 --> 00:36:29.510
increasingly I was just having a
00:36:29.520 --> 00:36:31.910
discussion with one CIO a few days ago
00:36:31.920 --> 00:36:34.069
about this and and uh she was asking
00:36:34.079 --> 00:36:36.390
about uh you know library sciences she
00:36:36.400 --> 00:36:38.870
heard that library scientists scientists
00:36:38.880 --> 00:36:41.349
are kind of now involved in these things
00:36:41.359 --> 00:36:43.670
and and uh had lots of knowledge about
00:36:43.680 --> 00:36:45.589
that topic actually and and I said yeah
00:36:45.599 --> 00:36:48.150
they're basically now the hot potato of
00:36:48.160 --> 00:36:49.430
this era.
00:36:49.440 --> 00:36:51.910
Yeah, data scientist was the sexiest job
00:36:51.920 --> 00:36:54.790
of of uh what 2000s or something.
00:36:54.800 --> 00:36:56.550
Anyways,
00:36:56.560 --> 00:36:59.430
and uh I started thinking after that
00:36:59.440 --> 00:37:01.750
discussion that oh my god you know what
00:37:01.760 --> 00:37:03.510
is going to happen in many organizations
00:37:03.520 --> 00:37:05.270
is that now that context and knowledge
00:37:05.280 --> 00:37:07.589
management are big topics IT
00:37:07.599 --> 00:37:09.750
organizations will try to start working
00:37:09.760 --> 00:37:12.230
on that and what IT organization does
00:37:12.240 --> 00:37:13.670
when it starts working on knowledge
00:37:13.680 --> 00:37:15.670
management or something is it buys a
00:37:15.680 --> 00:37:16.550
technology.
00:37:16.560 --> 00:37:19.190
>> Mhm. It sets up a technology and then it
00:37:19.200 --> 00:37:20.950
says here is your technology now you can
00:37:20.960 --> 00:37:23.030
now you can do knowledge management and
00:37:23.040 --> 00:37:27.190
and uh the management of content has
00:37:27.200 --> 00:37:31.190
never been it kind of you know work or
00:37:31.200 --> 00:37:33.510
or any any kind of even interest to lots
00:37:33.520 --> 00:37:37.190
of it I one of my old friends Sad who
00:37:37.200 --> 00:37:38.950
also does Helsinki data week with with
00:37:38.960 --> 00:37:41.190
me uh Sades said and I will always
00:37:41.200 --> 00:37:44.310
remember and use this that it and data
00:37:44.320 --> 00:37:46.150
things are like like you have pipes
00:37:46.160 --> 00:37:48.870
where you have running water in that it
00:37:48.880 --> 00:37:50.790
is interested in the pipes themselves
00:37:50.800 --> 00:37:53.990
but they are not interested in the water
00:37:54.000 --> 00:37:55.990
and it's very weird that we're kind of
00:37:56.000 --> 00:37:57.670
framing this whole data analytics and
00:37:57.680 --> 00:37:58.950
knowledge management issue as sort of
00:37:58.960 --> 00:38:01.270
water management issue as it were but we
00:38:01.280 --> 00:38:02.790
are actually only focusing on the pipes
00:38:02.800 --> 00:38:04.870
and that's because it is running things
00:38:04.880 --> 00:38:07.430
so n context management and knowledge
00:38:07.440 --> 00:38:09.589
management and so on I don't think that
00:38:09.599 --> 00:38:13.030
uh culturally organizationally or kind
00:38:13.040 --> 00:38:17.030
of competence wise IT units will be able
00:38:17.040 --> 00:38:19.430
to handle the needs for context and
00:38:19.440 --> 00:38:21.589
knowledge management that we now need
00:38:21.599 --> 00:38:24.790
due to AI. It's not part of what they've
00:38:24.800 --> 00:38:26.870
done part of, you know, the competence
00:38:26.880 --> 00:38:28.790
that exists and trying to kind of
00:38:28.800 --> 00:38:31.750
shoehorn the uh sort of knowledge
00:38:31.760 --> 00:38:33.670
management management compet uh
00:38:33.680 --> 00:38:36.470
competence into existing corporate IT
00:38:36.480 --> 00:38:37.990
organizations is going to be super
00:38:38.000 --> 00:38:39.589
difficult. We'll see a lot of people
00:38:39.599 --> 00:38:41.910
burnt out and you know switching jobs
00:38:41.920 --> 00:38:44.310
and failed projects and you know what
00:38:44.320 --> 00:38:45.990
what what not.
00:38:46.000 --> 00:38:47.750
Yeah, I just I just want to bounce back
00:38:47.760 --> 00:38:50.870
to you. I'm old enough now that I
00:38:50.880 --> 00:38:53.430
remember when mainframes were around and
00:38:53.440 --> 00:38:57.270
PCs turned up and the dream back then
00:38:57.280 --> 00:38:59.910
was technology enabled the organization
00:38:59.920 --> 00:39:02.069
to work in a better way.
00:39:02.079 --> 00:39:03.829
>> It it wasn't about just owning the
00:39:03.839 --> 00:39:05.349
plumbing and the infrastructure, right?
00:39:05.359 --> 00:39:07.030
It was about, hey, we got the strategy
00:39:07.040 --> 00:39:10.870
and how does a PC affect it, right? Does
00:39:10.880 --> 00:39:12.470
it make it better? That's where it kind
00:39:12.480 --> 00:39:14.630
of started. But yeah, I agree now that
00:39:14.640 --> 00:39:16.950
generalizing it, technology has now been
00:39:16.960 --> 00:39:19.990
relegated to plumbing and maintaining
00:39:20.000 --> 00:39:23.589
hardware, cloud hardware. Um, I have
00:39:23.599 --> 00:39:26.150
worked with CTO's and CIOS that don't
00:39:26.160 --> 00:39:28.470
behave that way that are enablers, you
00:39:28.480 --> 00:39:29.910
know, that sit there and go, you know,
00:39:29.920 --> 00:39:31.270
what is the strategy? What is, you know,
00:39:31.280 --> 00:39:33.190
where we going, right? How can we help
00:39:33.200 --> 00:39:36.230
do that? But nine times out of 10 I'll
00:39:36.240 --> 00:39:39.270
see that uh that type of role be in a
00:39:39.280 --> 00:39:41.109
when I used to work in enterprise
00:39:41.119 --> 00:39:42.470
organizations you know digital
00:39:42.480 --> 00:39:43.990
transformation. Yeah. And and the
00:39:44.000 --> 00:39:45.670
problem was they treated it as a program
00:39:45.680 --> 00:39:47.990
right a start and an end not a way of
00:39:48.000 --> 00:39:51.829
working. Um and it got relegated uh to
00:39:51.839 --> 00:39:53.510
being hey we're going to tell you what
00:39:53.520 --> 00:39:54.950
hardware or technology we need you to
00:39:54.960 --> 00:39:57.990
implement late uh with no notice and
00:39:58.000 --> 00:39:59.109
expect you to do it really really
00:39:59.119 --> 00:40:01.589
quickly. Um, so I think that's where we
00:40:01.599 --> 00:40:03.190
kind of got to and I don't think it's
00:40:03.200 --> 00:40:04.390
right and I think it's, you're right,
00:40:04.400 --> 00:40:06.230
it's going to happen with context. I
00:40:06.240 --> 00:40:08.069
think the other thing is when we talk
00:40:08.079 --> 00:40:11.510
about context technology, we're at the
00:40:11.520 --> 00:40:13.750
beginning of the Hadoop era.
00:40:13.760 --> 00:40:14.630
>> Yes.
00:40:14.640 --> 00:40:16.390
>> Well, we're experimenting, right? Nobody
00:40:16.400 --> 00:40:18.470
has an answer. There is no proven
00:40:18.480 --> 00:40:21.190
patterns, I don't think. Everybody's
00:40:21.200 --> 00:40:23.109
learning really fast. And therefore, if
00:40:23.119 --> 00:40:25.829
you're going to go buy technology first,
00:40:25.839 --> 00:40:27.030
you're probably going to buy something
00:40:27.040 --> 00:40:29.109
that won't survive or something that's
00:40:29.119 --> 00:40:31.510
wrong. Um, and we all know that, you
00:40:31.520 --> 00:40:33.750
know, standing joke is you buy SAP to
00:40:33.760 --> 00:40:36.470
implement best practice, best practice,
00:40:36.480 --> 00:40:38.790
and then you tailor it so it becomes the
00:40:38.800 --> 00:40:40.630
most expensive, unusable product in the
00:40:40.640 --> 00:40:42.630
world that can't be maintained. That's
00:40:42.640 --> 00:40:44.790
not SAP's fault, right? Their technology
00:40:44.800 --> 00:40:47.270
says, "This is best practice. Put it in
00:40:47.280 --> 00:40:49.190
like this and you'll be as good as
00:40:49.200 --> 00:40:51.589
everybody else like you." it's the
00:40:51.599 --> 00:40:53.510
organization that decides to break that
00:40:53.520 --> 00:40:55.990
rule uh and do something different. And
00:40:56.000 --> 00:41:00.230
so yeah, I I don't know um like we just
00:41:00.240 --> 00:41:01.829
everything's moving so fast right now.
00:41:01.839 --> 00:41:04.390
You go buy a context technology uh I
00:41:04.400 --> 00:41:06.790
think you'll be lucky if it survives in
00:41:06.800 --> 00:41:09.990
its form that you buy it uh in a year's
00:41:10.000 --> 00:41:11.430
time.
00:41:11.440 --> 00:41:13.430
>> Yeah. Well, the thing I think the
00:41:13.440 --> 00:41:15.030
comparison to Hadoop era is really
00:41:15.040 --> 00:41:17.349
really good because you know what is
00:41:17.359 --> 00:41:21.109
left of those uh Hadoop based enterprise
00:41:21.119 --> 00:41:25.270
data lakes and so on so forth basically
00:41:25.280 --> 00:41:27.109
nothing and
00:41:27.119 --> 00:41:29.109
>> yeah but we got snowflake and data
00:41:29.119 --> 00:41:29.750
bricks right
00:41:29.760 --> 00:41:31.510
>> well yeah I mean yeah that's kind of
00:41:31.520 --> 00:41:32.550
where I was getting to
00:41:32.560 --> 00:41:34.309
>> yeah the next generation came along and
00:41:34.319 --> 00:41:36.150
did it yeah
00:41:36.160 --> 00:41:38.230
>> yeah we got a comment on the chat no
00:41:38.240 --> 00:41:41.670
real reuse at any level I mean yes So
00:41:41.680 --> 00:41:43.589
I'm guessing that's John O Gorman maybe.
00:41:43.599 --> 00:41:45.349
Uh and I see Christopher Chin as well.
00:41:45.359 --> 00:41:46.470
Feel free to give us a question
00:41:46.480 --> 00:41:47.990
Christopher. Um
00:41:48.000 --> 00:41:50.550
>> so no real reuse at any level. I think
00:41:50.560 --> 00:41:52.870
that's a choice. Uh, like I said, I
00:41:52.880 --> 00:41:55.349
think you go we are primarily focused on
00:41:55.359 --> 00:41:57.829
no reuse and we're going to optimize for
00:41:57.839 --> 00:42:00.870
a no reuse way of working or we're
00:42:00.880 --> 00:42:04.309
primarily focused on reuse and then
00:42:04.319 --> 00:42:06.309
we're going to, you know, we're going to
00:42:06.319 --> 00:42:09.109
see how we can get agility and speed um
00:42:09.119 --> 00:42:10.790
without having to do big design up
00:42:10.800 --> 00:42:11.349
front.
00:42:11.359 --> 00:42:13.670
>> Yeah. And big design I mean six months,
00:42:13.680 --> 00:42:16.230
year, two years kind of design. Yeah.
00:42:16.240 --> 00:42:17.990
Although I want to point out something
00:42:18.000 --> 00:42:20.550
that I've often ranted against as well
00:42:20.560 --> 00:42:23.430
which which is the endless enablement
00:42:23.440 --> 00:42:27.349
projects that uh that you know if no one
00:42:27.359 --> 00:42:29.030
is building actually anything valuable
00:42:29.040 --> 00:42:30.870
and everyone's just building tools for
00:42:30.880 --> 00:42:32.950
each other to enable them to deliver
00:42:32.960 --> 00:42:34.630
value then that's completely pointless
00:42:34.640 --> 00:42:37.109
and I think that's one of the uh big
00:42:37.119 --> 00:42:41.030
sort of basic instincts of of us sort of
00:42:41.040 --> 00:42:43.430
more holistically minded data people
00:42:43.440 --> 00:42:46.069
that we want to create framework works
00:42:46.079 --> 00:42:47.829
and tools and methods that are basically
00:42:47.839 --> 00:42:50.630
enabling someone to do something. But
00:42:50.640 --> 00:42:53.109
you know where's the actual value? It's
00:42:53.119 --> 00:42:54.630
quite often lost.
00:42:54.640 --> 00:42:57.109
>> Uh you know there are people out there
00:42:57.119 --> 00:42:59.589
putting putting my hand up where
00:42:59.599 --> 00:43:00.470
>> Yeah. Yeah.
00:43:00.480 --> 00:43:03.270
>> I have probably spent far too long over
00:43:03.280 --> 00:43:05.270
the last couple of months working on my
00:43:05.280 --> 00:43:06.950
harnesses.
00:43:06.960 --> 00:43:07.670
>> Uh
00:43:07.680 --> 00:43:08.950
>> there we go.
00:43:08.960 --> 00:43:11.589
where you know I could question am I
00:43:11.599 --> 00:43:14.230
really getting a reuse return on value
00:43:14.240 --> 00:43:15.829
uh for some of those things. I
00:43:15.839 --> 00:43:18.069
definitely am right. So there is I'm
00:43:18.079 --> 00:43:20.950
seeing the benefit of reuse now but
00:43:20.960 --> 00:43:24.390
rabbit holes I go down where I go with
00:43:24.400 --> 00:43:25.990
that and and again one of the
00:43:26.000 --> 00:43:27.349
interesting things for me is is
00:43:27.359 --> 00:43:29.349
incentive right because I'm on a clawed
00:43:29.359 --> 00:43:31.990
subscription because I don't pay per
00:43:32.000 --> 00:43:36.230
tokens at the moment. uh when I get to
00:43:36.240 --> 00:43:38.230
the near the end of my subscription
00:43:38.240 --> 00:43:42.470
week, I will go down a rabbit hole of
00:43:42.480 --> 00:43:44.710
what would what about, right? Because I
00:43:44.720 --> 00:43:46.550
kind of feel like I have to use it or
00:43:46.560 --> 00:43:50.150
lose it. I kind of feel like it's free.
00:43:50.160 --> 00:43:51.510
But the interesting thing is that
00:43:51.520 --> 00:43:55.109
experimentation time is often where
00:43:55.119 --> 00:43:56.390
something happens and I go, "Well,
00:43:56.400 --> 00:43:57.910
that's interesting." And then I I don't
00:43:57.920 --> 00:43:59.349
do anything about then I think about it
00:43:59.359 --> 00:44:01.030
and and a week or two later I'll I'll
00:44:01.040 --> 00:44:03.270
reuse it. So again, this is the other
00:44:03.280 --> 00:44:05.109
part is this idea of experimentation.
00:44:05.119 --> 00:44:06.950
Often experimentation is treated like
00:44:06.960 --> 00:44:08.710
the toy shop. You know, you got the the
00:44:08.720 --> 00:44:10.790
the digital or the experimental the
00:44:10.800 --> 00:44:12.150
innovation team. Yeah, that's my
00:44:12.160 --> 00:44:13.750
favorite one. You know, they're the ones
00:44:13.760 --> 00:44:15.670
over back in the days we had buildings,
00:44:15.680 --> 00:44:16.870
you know, they were on the floor that
00:44:16.880 --> 00:44:19.190
had the cool green carpet and the and
00:44:19.200 --> 00:44:21.430
the park bench. Seriously, you know, and
00:44:21.440 --> 00:44:23.990
and it one organization, one government
00:44:24.000 --> 00:44:26.470
agency had a tree inside the wall in the
00:44:26.480 --> 00:44:28.230
room, right? Because they were the
00:44:28.240 --> 00:44:31.030
innovation thing. Um, they didn't last
00:44:31.040 --> 00:44:33.190
too long cuz they delivered jack
00:44:33.200 --> 00:44:35.349
Um, so I'm not talking about that, but
00:44:35.359 --> 00:44:37.430
I'm talking about giving people time to
00:44:37.440 --> 00:44:39.190
experiment, learn, and and play with
00:44:39.200 --> 00:44:40.710
things. You know, the what is it? The
00:44:40.720 --> 00:44:42.950
Amazon 20% or or whatever they talk
00:44:42.960 --> 00:44:46.630
about. Um, so incentive to pay the tax
00:44:46.640 --> 00:44:48.630
that you know you're applying because
00:44:48.640 --> 00:44:52.390
you're going too fast. Uh, I still think
00:44:52.400 --> 00:44:54.550
there's massive value and reuse.
00:44:54.560 --> 00:44:57.990
answered John's view. I think reuse is
00:44:58.000 --> 00:45:00.390
is highly important and that's just part
00:45:00.400 --> 00:45:02.150
of my persona,
00:45:02.160 --> 00:45:04.710
but I try and figure out how we can go
00:45:04.720 --> 00:45:07.430
faster, more point solutions that then
00:45:07.440 --> 00:45:09.430
become reusable
00:45:09.440 --> 00:45:13.270
um rather than big big upfront. Um, and
00:45:13.280 --> 00:45:14.630
don't get me wrong, you know, when I was
00:45:14.640 --> 00:45:17.589
in it, in in the agile community, um,
00:45:17.599 --> 00:45:20.390
everybody used to rant about waterfall.
00:45:20.400 --> 00:45:23.349
And, you know, one of the things I I I I
00:45:23.359 --> 00:45:27.430
believed was Waterfall works. Yeah, it
00:45:27.440 --> 00:45:28.950
has a way of working. If you follow that
00:45:28.960 --> 00:45:31.750
way of working, it's successful.
00:45:31.760 --> 00:45:33.430
It's not my way of working, but that's
00:45:33.440 --> 00:45:35.829
okay. I just wouldn't walk work on a a
00:45:35.839 --> 00:45:38.150
waterfall project. Um, but you can make
00:45:38.160 --> 00:45:39.430
it, it's like anything. You can make it
00:45:39.440 --> 00:45:41.270
work, you know. It's like data vault
00:45:41.280 --> 00:45:43.109
dimensional activity schema you can make
00:45:43.119 --> 00:45:47.190
them work like just pick one do it or
00:45:47.200 --> 00:45:49.829
you know completely distributed don't
00:45:49.839 --> 00:45:52.309
just whatever that looks like because I
00:45:52.319 --> 00:45:55.270
don't know so yeah um I I think it's a
00:45:55.280 --> 00:45:57.190
conscious decision and then you should
00:45:57.200 --> 00:45:58.790
make sure you everything else lines up
00:45:58.800 --> 00:46:00.870
with that conscious decision.
00:46:00.880 --> 00:46:03.270
>> Yeah. Yeah. altogether. Yeah, again kind
00:46:03.280 --> 00:46:04.776
of contradicting my earlier earlier
00:46:04.786 --> 00:46:07.109
[laughter] comment on I like
00:46:07.119 --> 00:46:10.150
contradicting myself but um on kind of
00:46:10.160 --> 00:46:11.750
enablement and useless enablement
00:46:11.760 --> 00:46:13.270
projects and so on so forth but but
00:46:13.280 --> 00:46:15.829
honestly I I also agree and I think that
00:46:15.839 --> 00:46:18.150
reuse is actually like super super
00:46:18.160 --> 00:46:20.630
important key word because reusing
00:46:20.640 --> 00:46:23.589
things reduces waste. We want to reduce
00:46:23.599 --> 00:46:26.230
waste. I'm I'm sort of lean purist in
00:46:26.240 --> 00:46:27.910
the sense that I think in everything we
00:46:27.920 --> 00:46:29.750
do we should aim to constantly kind of
00:46:29.760 --> 00:46:31.750
reduce waste. figure out what's the
00:46:31.760 --> 00:46:33.670
bottleneck, fix the bottleneck and then
00:46:33.680 --> 00:46:35.430
just do the next thing. Never fix
00:46:35.440 --> 00:46:37.510
anything which uh isn't the bottleneck
00:46:37.520 --> 00:46:40.630
and so on so forth. Okay. But uh what I
00:46:40.640 --> 00:46:42.470
like in product thinking and this
00:46:42.480 --> 00:46:44.150
approach to products and this is why I
00:46:44.160 --> 00:46:46.069
keep yelling about data products all the
00:46:46.079 --> 00:46:48.230
time. But I also think that uh well like
00:46:48.240 --> 00:46:50.230
Anna Burin has been writing a lot about
00:46:50.240 --> 00:46:52.470
kind of data platforms and capabilities
00:46:52.480 --> 00:46:54.309
of the platform as a product. I also
00:46:54.319 --> 00:46:55.910
think that it applies there very very
00:46:55.920 --> 00:46:59.190
much indeed that if we just managed to
00:46:59.200 --> 00:47:01.910
think about not just the valuable things
00:47:01.920 --> 00:47:03.750
that we produce for end users as
00:47:03.760 --> 00:47:05.910
products you know apply actual product
00:47:05.920 --> 00:47:08.309
thinking there but also apply actual
00:47:08.319 --> 00:47:10.470
product thinking in the harnesses and
00:47:10.480 --> 00:47:13.190
and capabilities and and components and
00:47:13.200 --> 00:47:15.829
templates and whatnot that we produce to
00:47:15.839 --> 00:47:18.150
enable others to produce valuable things
00:47:18.160 --> 00:47:19.990
then I I think that's also a product
00:47:20.000 --> 00:47:21.910
management discussion you know there's
00:47:21.920 --> 00:47:23.910
demand for certain kinds of products We
00:47:23.920 --> 00:47:25.829
have to do some experimentation to find
00:47:25.839 --> 00:47:27.430
what product actually fits. There's kind
00:47:27.440 --> 00:47:29.589
of a go to market strategy with an
00:47:29.599 --> 00:47:31.270
internal tool that you build for your
00:47:31.280 --> 00:47:33.270
own developers and all of that is
00:47:33.280 --> 00:47:35.190
product management work. We just don't
00:47:35.200 --> 00:47:37.510
think of it that way often.
00:47:37.520 --> 00:47:39.270
>> Yeah. I I think you know we should treat
00:47:39.280 --> 00:47:41.510
our tooling as if it's a product.
00:47:41.520 --> 00:47:41.910
>> Yeah.
00:47:41.920 --> 00:47:44.390
>> But the problem is as soon as you go
00:47:44.400 --> 00:47:45.990
back to team topologies, right? As soon
00:47:46.000 --> 00:47:48.150
as you have a team that are focused on
00:47:48.160 --> 00:47:49.990
that platform and that tooling as their
00:47:50.000 --> 00:47:52.230
product that serves internally, we now
00:47:52.240 --> 00:47:54.470
have a massive sequencing problem
00:47:54.480 --> 00:47:56.150
because the people doing the rest of the
00:47:56.160 --> 00:47:57.510
work that need that tooling or that
00:47:57.520 --> 00:48:00.870
product, right? That reuse have to
00:48:00.880 --> 00:48:02.950
actually telegraph that they're going to
00:48:02.960 --> 00:48:04.470
need it and they probably don't know
00:48:04.480 --> 00:48:07.190
yet. They have to tell them when they
00:48:07.200 --> 00:48:08.870
need it by and it actually has to meet
00:48:08.880 --> 00:48:10.550
that time frame which often doesn't
00:48:10.560 --> 00:48:12.630
happen. They have to be available for
00:48:12.640 --> 00:48:14.230
feedback loops, right? because you don't
00:48:14.240 --> 00:48:16.790
want to give them the spec and then um
00:48:16.800 --> 00:48:18.550
just wait for it to turn up and not be
00:48:18.560 --> 00:48:19.829
what you want, right? That's what we do
00:48:19.839 --> 00:48:21.589
to our our stakeholders, right? Don't
00:48:21.599 --> 00:48:23.430
don't do it to ourselves and that
00:48:23.440 --> 00:48:26.230
sequencing becomes really really uh
00:48:26.240 --> 00:48:27.030
thing.
00:48:27.040 --> 00:48:29.109
>> But but how is that different from what
00:48:29.119 --> 00:48:31.349
kind of the end user facing data teams
00:48:31.359 --> 00:48:33.190
do with their business counterparts?
00:48:33.200 --> 00:48:35.190
It's it's the exact same thing, right?
00:48:35.200 --> 00:48:37.190
The business could have deadlines for
00:48:37.200 --> 00:48:39.510
you know whatever and and the business
00:48:39.520 --> 00:48:41.990
could have requests coming in that
00:48:42.000 --> 00:48:43.589
they're quite not sure what they should
00:48:43.599 --> 00:48:45.829
actually need and they don't necessarily
00:48:45.839 --> 00:48:47.670
even know what they need and and all
00:48:47.680 --> 00:48:49.349
that discussion is product management in
00:48:49.359 --> 00:48:51.510
the kind of enduserfacing
00:48:51.520 --> 00:48:53.670
uh interface but it also should be the
00:48:53.680 --> 00:48:55.589
exact same discussion the exact same
00:48:55.599 --> 00:48:56.870
feedback loops the exact same
00:48:56.880 --> 00:48:59.430
connections between the kind of internal
00:48:59.440 --> 00:49:01.750
capability building platform team and
00:49:01.760 --> 00:49:04.470
and and their consumers which are then
00:49:04.480 --> 00:49:06.710
other development teams. I I think it's
00:49:06.720 --> 00:49:08.870
the exact same process in both interface
00:49:08.880 --> 00:49:11.190
interfaces, userf facing and and kind of
00:49:11.200 --> 00:49:12.150
developer facing.
00:49:12.160 --> 00:49:13.510
>> I I think it's the same principles,
00:49:13.520 --> 00:49:15.109
right? And potentially the same
00:49:15.119 --> 00:49:16.950
patterns, although I think we do need to
00:49:16.960 --> 00:49:18.710
change the patterns slightly if we've
00:49:18.720 --> 00:49:21.109
got a platform or tooling versus data
00:49:21.119 --> 00:49:23.910
versus uh things that there's some
00:49:23.920 --> 00:49:25.190
things we can tweak that make it a
00:49:25.200 --> 00:49:27.589
little bit less wasteful or more useful
00:49:27.599 --> 00:49:31.030
for us. Um but that comes back to the
00:49:31.040 --> 00:49:34.230
should, right? like you know CIO should
00:49:34.240 --> 00:49:35.670
be looking after information for the
00:49:35.680 --> 00:49:37.190
organization and how it enables the
00:49:37.200 --> 00:49:39.510
strategy but a lot of the times they
00:49:39.520 --> 00:49:42.470
become technology plumbing maintainers.
00:49:42.480 --> 00:49:43.990
We should be able to apply product
00:49:44.000 --> 00:49:47.030
patterns to the way we do this but often
00:49:47.040 --> 00:49:50.150
we get stuck in a stupid 90-day loop
00:49:50.160 --> 00:49:51.510
right tell me what you want for the next
00:49:51.520 --> 00:49:53.190
90 days and we'll go away and build it
00:49:53.200 --> 00:49:55.510
and then nothing sequences. So yeah,
00:49:55.520 --> 00:49:57.109
it's it's the implementation, the
00:49:57.119 --> 00:49:59.829
execution of the patterns that we get
00:49:59.839 --> 00:50:01.589
wrong, not the patterns themselves. But
00:50:01.599 --> 00:50:03.990
yeah, no, I agree with you. Uh I come
00:50:04.000 --> 00:50:06.630
back to I still and and it's because I'm
00:50:06.640 --> 00:50:08.150
kind of working on and thinking about it
00:50:08.160 --> 00:50:11.109
at the moment again, which always kind
00:50:11.119 --> 00:50:13.349
of um
00:50:13.359 --> 00:50:15.829
uh changes, you know, not changes, kind
00:50:15.839 --> 00:50:18.470
of um colors the way I think about
00:50:18.480 --> 00:50:20.549
things. And so right now this idea of
00:50:20.559 --> 00:50:23.750
pioneers and towns
00:50:23.760 --> 00:50:26.150
I I you know it's still product thinking
00:50:26.160 --> 00:50:30.069
right. Um and I'm like yeah it kind of
00:50:30.079 --> 00:50:31.990
makes sense to me. I can see how we can
00:50:32.000 --> 00:50:33.910
focus on some of those areas of waste
00:50:33.920 --> 00:50:35.829
and see whether we can apply patents to
00:50:35.839 --> 00:50:38.150
solve it or not. So it's why coming back
00:50:38.160 --> 00:50:40.390
to it on this on this one. Um yeah, the
00:50:40.400 --> 00:50:41.829
interesting thing is Anna is part of one
00:50:41.839 --> 00:50:44.870
of the Aura Forsome, right? And um
00:50:44.880 --> 00:50:46.150
>> she's in this really interesting
00:50:46.160 --> 00:50:48.710
situation where she's gone into a
00:50:48.720 --> 00:50:50.710
children's hospital effectively or
00:50:50.720 --> 00:50:51.990
organization because they got lot of
00:50:52.000 --> 00:50:54.230
hospital buildings depending on your
00:50:54.240 --> 00:50:55.829
ontological definition of what
00:50:55.839 --> 00:50:58.230
hospitals. Um she'll ask her that one.
00:50:58.240 --> 00:51:00.630
We do the next thing. Um and they've
00:51:00.640 --> 00:51:03.349
just replaced their their um core
00:51:03.359 --> 00:51:04.950
system, right? their electronic health
00:51:04.960 --> 00:51:06.870
record system, which again, funny thing
00:51:06.880 --> 00:51:09.589
is she keeps saying EHR, but with her
00:51:09.599 --> 00:51:11.589
accent and my accent, I keep hearing HR,
00:51:11.599 --> 00:51:12.630
and I'm like, why are we talking about
00:51:12.640 --> 00:51:15.190
HR and you're in a in a Yeah. Anyway,
00:51:15.200 --> 00:51:18.470
um, and so she's getting to kind of
00:51:18.480 --> 00:51:20.230
build things from scratch, right? Uh and
00:51:20.240 --> 00:51:22.230
so we talked about ontologies and I said
00:51:22.240 --> 00:51:24.710
well wouldn't you know if I think about
00:51:24.720 --> 00:51:26.390
ontologies
00:51:26.400 --> 00:51:29.109
if I look at health and bioscience
00:51:29.119 --> 00:51:31.670
like it seems to be the place that's had
00:51:31.680 --> 00:51:33.270
the most work done and had the most
00:51:33.280 --> 00:51:35.030
standards in the outer box and and that
00:51:35.040 --> 00:51:37.030
was a really interesting conversation.
00:51:37.040 --> 00:51:38.950
Um and the other thing she's thinking
00:51:38.960 --> 00:51:41.750
about is waste. So what she's saying is
00:51:41.760 --> 00:51:43.190
uh when they were going through all the
00:51:43.200 --> 00:51:46.790
workshops to define how the EI the
00:51:46.800 --> 00:51:48.309
electronic health record system was
00:51:48.319 --> 00:51:50.549
working, they were capturing a hell of a
00:51:50.559 --> 00:51:52.870
lot of subject matter expertise.
00:51:52.880 --> 00:51:56.549
Um and then the data team had to go and
00:51:56.559 --> 00:51:58.870
try and recapture it after the fact. And
00:51:58.880 --> 00:52:00.790
I'm like oh you and I talked about this
00:52:00.800 --> 00:52:02.630
when we were talking about the book and
00:52:02.640 --> 00:52:04.390
we're like why don't we just store the
00:52:04.400 --> 00:52:06.230
transcripts?
00:52:06.240 --> 00:52:08.390
and she said, "Yeah, but the problem is
00:52:08.400 --> 00:52:10.470
with the tooling they're using, the
00:52:10.480 --> 00:52:12.230
transcripts only get held for a certain
00:52:12.240 --> 00:52:14.150
period of time and then they get wiped
00:52:14.160 --> 00:52:16.630
by the platform vendor."
00:52:16.640 --> 00:52:17.990
>> And one of the things she'd mentioned
00:52:18.000 --> 00:52:19.829
was in those subject matter expert
00:52:19.839 --> 00:52:22.230
workshops, it was really expensive
00:52:22.240 --> 00:52:24.230
because you've got specialist medical
00:52:24.240 --> 00:52:25.829
practitioners in those workshops, right?
00:52:25.839 --> 00:52:27.349
you've got the the really really
00:52:27.359 --> 00:52:29.510
expensive neurosurgeons
00:52:29.520 --> 00:52:32.549
and so the meeting to capture that
00:52:32.559 --> 00:52:35.030
knowledge is really expensive and
00:52:35.040 --> 00:52:36.309
they're really busy people so they're
00:52:36.319 --> 00:52:38.630
never coming back in to relitigate
00:52:38.640 --> 00:52:41.589
definition of a term or how things work.
00:52:41.599 --> 00:52:44.870
Um but yet the cost of storing that
00:52:44.880 --> 00:52:47.030
knowledge which is incredibly expensive
00:52:47.040 --> 00:52:49.030
to capture from a technology point of
00:52:49.040 --> 00:52:52.230
view was seen as too high and I'm like
00:52:52.240 --> 00:52:55.670
yeah if we talk about oil that's the oil
00:52:55.680 --> 00:52:57.829
right not the data that that's not oil.
00:52:57.839 --> 00:52:58.230
Yeah,
00:52:58.240 --> 00:53:01.270
>> I mean exactly and this is at the very
00:53:01.280 --> 00:53:03.109
very core of the kind of feedback loops
00:53:03.119 --> 00:53:04.390
that I was talking about in the
00:53:04.400 --> 00:53:06.390
beginning that uh as you said you know
00:53:06.400 --> 00:53:09.190
you have your extremely expensive
00:53:09.200 --> 00:53:11.109
doctors or whatn not coming down to talk
00:53:11.119 --> 00:53:13.349
about some individual project perhaps
00:53:13.359 --> 00:53:15.829
you know and they in that discussion
00:53:15.839 --> 00:53:17.750
they bring up all sorts of information
00:53:17.760 --> 00:53:20.630
that never before has existed kind of in
00:53:20.640 --> 00:53:23.190
a digital or even on paper or anything
00:53:23.200 --> 00:53:27.109
form and and uh we are just using that
00:53:27.119 --> 00:53:28.710
information to produce that single
00:53:28.720 --> 00:53:31.030
artifact at the end as delivery of that
00:53:31.040 --> 00:53:33.270
single project. We're not making that
00:53:33.280 --> 00:53:35.430
information reusable. That is the
00:53:35.440 --> 00:53:36.950
feedback loop from the kind of small
00:53:36.960 --> 00:53:39.349
problem, the individual workflows up to
00:53:39.359 --> 00:53:41.510
some sort of enterprise knowledge
00:53:41.520 --> 00:53:44.549
repository graph of everything which
00:53:44.559 --> 00:53:47.030
should actually capture those things.
00:53:47.040 --> 00:53:50.230
But this is kind of super I understand
00:53:50.240 --> 00:53:52.710
it's technologically diff difficult but
00:53:52.720 --> 00:53:55.510
I think this is where we have a lot to
00:53:55.520 --> 00:53:58.150
kind of improve on because you know we
00:53:58.160 --> 00:54:00.549
are wasting so much information. We're
00:54:00.559 --> 00:54:02.790
not taking advantage of the information
00:54:02.800 --> 00:54:05.030
we get in those situation because we
00:54:05.040 --> 00:54:08.390
consider it to be kind of siloed in this
00:54:08.400 --> 00:54:10.710
little solution specific delivery
00:54:10.720 --> 00:54:12.870
pipeline and then we just forget it and
00:54:12.880 --> 00:54:14.630
start figuring out the next sol solution
00:54:14.640 --> 00:54:15.109
delivery
00:54:15.119 --> 00:54:16.710
>> and we don't treat it as reusable. So,
00:54:16.720 --> 00:54:18.069
one of the things I used to do was, you
00:54:18.079 --> 00:54:19.990
know, I'd talk to a stakeholder, go
00:54:20.000 --> 00:54:21.430
through the uh information product
00:54:21.440 --> 00:54:23.270
canvas
00:54:23.280 --> 00:54:25.270
>> and you I'd summarize the transcript
00:54:25.280 --> 00:54:26.790
from that conversation because I found
00:54:26.800 --> 00:54:28.630
it useful, but then I kind of forget
00:54:28.640 --> 00:54:30.870
about the transcript itself, right? And
00:54:30.880 --> 00:54:33.510
then now what I've got is a summary of
00:54:33.520 --> 00:54:36.950
what it thought or I thought happened.
00:54:36.960 --> 00:54:38.950
What I found is if I now go back to the
00:54:38.960 --> 00:54:41.030
transcript every time I want to clarify
00:54:41.040 --> 00:54:43.750
something, I I focus on different
00:54:43.760 --> 00:54:45.109
things. And again I was writing a
00:54:45.119 --> 00:54:46.870
workshop with a stakeholder this this
00:54:46.880 --> 00:54:49.589
payment one and the purpose of the
00:54:49.599 --> 00:54:51.589
workshop was for something else and we
00:54:51.599 --> 00:54:54.230
got onto the subject of payments
00:54:54.240 --> 00:54:57.030
and they just started talking right and
00:54:57.040 --> 00:54:58.790
my natural reaction is no hold on that's
00:54:58.800 --> 00:55:00.549
a that's a different subject right we'll
00:55:00.559 --> 00:55:02.309
come back to that and I'm like no
00:55:02.319 --> 00:55:04.309
right like just let them talk I don't
00:55:04.319 --> 00:55:05.670
actually have to ask him any questions
00:55:05.680 --> 00:55:06.950
because we don't have time for me to
00:55:06.960 --> 00:55:08.790
deep dive and understand but because
00:55:08.800 --> 00:55:10.309
we've got agreement that I can use the
00:55:10.319 --> 00:55:11.910
transcripts that's now going in there
00:55:11.920 --> 00:55:13.430
right now I can go back and look at it
00:55:13.440 --> 00:55:17.109
and I and go ah so payment arrangement
00:55:17.119 --> 00:55:18.950
right or broken payment arrangement not
00:55:18.960 --> 00:55:20.150
broken arrangement that's going to be
00:55:20.160 --> 00:55:23.109
really important and so that that's the
00:55:23.119 --> 00:55:25.349
goal right and and to capture that spend
00:55:25.359 --> 00:55:27.510
the time capturing it quickly store it
00:55:27.520 --> 00:55:29.670
uh is really really important and the
00:55:29.680 --> 00:55:30.950
other thing I found and I kind of wrote
00:55:30.960 --> 00:55:32.549
about that in the in the last article I
00:55:32.559 --> 00:55:35.510
did is when I'm working with the GI
00:55:35.520 --> 00:55:38.309
tooling I'm finding I was finding that
00:55:38.319 --> 00:55:40.069
uh stakeholder or the person I'm talking
00:55:40.079 --> 00:55:41.990
to was speaking German effectively their
00:55:42.000 --> 00:55:43.910
language I was converting getting it
00:55:43.920 --> 00:55:46.150
into English to my patterns or what I
00:55:46.160 --> 00:55:47.349
understood and then I was trying to
00:55:47.359 --> 00:55:49.030
convert it from that into something I
00:55:49.040 --> 00:55:51.829
could give Claude or Gemini uh which is
00:55:51.839 --> 00:55:54.390
French, right? And and that translation
00:55:54.400 --> 00:55:56.630
meant it was nowhere near as effective
00:55:56.640 --> 00:55:59.670
as me just giving their German to the
00:55:59.680 --> 00:56:03.030
LLM and it translating it into French.
00:56:03.040 --> 00:56:04.789
And as I wrote in the article, I'm still
00:56:04.799 --> 00:56:06.150
in the loop, right? It's not fully
00:56:06.160 --> 00:56:07.990
autonomous, right? I'm a human in the
00:56:08.000 --> 00:56:09.270
loop, but it's assisting me, right? And
00:56:09.280 --> 00:56:11.270
I'm getting out of the way for the thing
00:56:11.280 --> 00:56:14.150
that I'm bad at. um and letting it use
00:56:14.160 --> 00:56:15.589
it and then I'm doing the things I'm
00:56:15.599 --> 00:56:19.430
good at. Um so yeah, I'm I'm like the
00:56:19.440 --> 00:56:21.910
technology, you know, if I'm a
00:56:21.920 --> 00:56:23.430
technologist
00:56:23.440 --> 00:56:25.670
almost a data lake, right? And I hate to
00:56:25.680 --> 00:56:28.150
say that, but actually these knowledge,
00:56:28.160 --> 00:56:31.270
this this transcripts, this text
00:56:31.280 --> 00:56:33.109
actually is the stuff that we should be
00:56:33.119 --> 00:56:34.549
capturing and storing because we're
00:56:34.559 --> 00:56:35.910
going to need it, but we still don't
00:56:35.920 --> 00:56:38.470
know how to use it effectively yet.
00:56:38.480 --> 00:56:40.150
>> Yep. Yep. Fully agreed. Fully agreed.
00:56:40.160 --> 00:56:42.789
The world of knowledge is so much wider
00:56:42.799 --> 00:56:44.549
than just the data that we actually
00:56:44.559 --> 00:56:46.390
have. It's it's just
00:56:46.400 --> 00:56:48.710
>> insane. Now that I've got my eyes have
00:56:48.720 --> 00:56:51.109
opened to this thing, it's just insane
00:56:51.119 --> 00:56:53.349
how much information is out there that
00:56:53.359 --> 00:56:56.230
is not actually anyway related to
00:56:56.240 --> 00:56:59.270
whatever tables we have or you know
00:56:59.280 --> 00:57:01.349
parket files somewhere.
00:57:01.359 --> 00:57:03.349
>> Well, it's back to the big the three V's
00:57:03.359 --> 00:57:04.630
of big data, isn't it? You know what was
00:57:04.640 --> 00:57:09.109
that? Velocity, volume, and veracity. No
00:57:09.119 --> 00:57:11.510
variety. Was that it? Maybe.
00:57:11.520 --> 00:57:13.510
>> Yeah. Yeah. God, you know, but back to
00:57:13.520 --> 00:57:16.230
the idea that audio and video are
00:57:16.240 --> 00:57:19.109
valuable bits of of data, you know,
00:57:19.119 --> 00:57:19.670
information.
00:57:19.680 --> 00:57:21.910
>> Yeah. I still think that okay, you know,
00:57:21.920 --> 00:57:24.870
storing the raw raw kind of data, raw
00:57:24.880 --> 00:57:26.630
recordings and so on. Yep. That's value
00:57:26.640 --> 00:57:28.710
in there. But I still do think that
00:57:28.720 --> 00:57:32.150
there is a way to distill from that kind
00:57:32.160 --> 00:57:34.870
of facts that are universal to the
00:57:34.880 --> 00:57:37.030
organization which is basically what our
00:57:37.040 --> 00:57:39.030
book with the conceptual models and and
00:57:39.040 --> 00:57:40.710
the modeling business concepts does
00:57:40.720 --> 00:57:43.270
right we are distilling from the story
00:57:43.280 --> 00:57:45.910
some sort of easily usable parts
00:57:45.920 --> 00:57:48.150
>> and I think that sort of a process is is
00:57:48.160 --> 00:57:50.150
where it's going to be at
00:57:50.160 --> 00:57:52.309
>> but again it comes back to and I suppose
00:57:52.319 --> 00:57:54.230
it comes back to John's point right
00:57:54.240 --> 00:57:56.549
question if we think about the process
00:57:56.559 --> 00:57:58.390
we're writing about you know mapping
00:57:58.400 --> 00:58:00.069
business concepts or or modeling
00:58:00.079 --> 00:58:03.430
business concepts. The concept map is a
00:58:03.440 --> 00:58:06.230
way of articulating complexity with
00:58:06.240 --> 00:58:08.150
simplicity through a diagram so people
00:58:08.160 --> 00:58:10.230
can understand and reuse
00:58:10.240 --> 00:58:13.750
>> or decide to reuse or not. Um
00:58:13.760 --> 00:58:17.190
does an LLM need the map? I don't know.
00:58:17.200 --> 00:58:19.270
That's a really good question. It it
00:58:19.280 --> 00:58:21.670
probably gives it hints on what we've
00:58:21.680 --> 00:58:24.230
agreed if it can't see those agreements.
00:58:24.240 --> 00:58:25.910
stops it scanning a whole lot of stuff
00:58:25.920 --> 00:58:27.589
because we've given it the agreement,
00:58:27.599 --> 00:58:28.069
the answer.
00:58:28.079 --> 00:58:29.990
>> From a token economics point of view,
00:58:30.000 --> 00:58:31.190
seems sensible.
00:58:31.200 --> 00:58:34.390
>> Yeah. Um but yeah, it it's that map
00:58:34.400 --> 00:58:37.510
really is is a shared language via a
00:58:37.520 --> 00:58:39.510
diagram. Uh that means somebody can
00:58:39.520 --> 00:58:41.670
understand it quickly and decide to
00:58:41.680 --> 00:58:44.950
reuse it because that makes sense. Um
00:58:44.960 --> 00:58:46.069
you know, if we had completely
00:58:46.079 --> 00:58:47.589
distributed teams going back to that
00:58:47.599 --> 00:58:50.950
model, would we would we even do it?
00:58:50.960 --> 00:58:53.910
Like I don't know. like if we're truly
00:58:53.920 --> 00:58:56.789
distributed then having that shared map
00:58:56.799 --> 00:58:59.030
and that reuse of that map would be
00:58:59.040 --> 00:59:00.950
waste, right? Because we'd slow that
00:59:00.960 --> 00:59:02.230
team down.
00:59:02.240 --> 00:59:04.150
>> Um but but I don't believe any
00:59:04.160 --> 00:59:06.549
organization is truly decentralized.
00:59:06.559 --> 00:59:08.390
>> Exactly. Exactly. Exactly. Because why
00:59:08.400 --> 00:59:09.510
would they be part of the same
00:59:09.520 --> 00:59:10.870
organization if they have nothing to do
00:59:10.880 --> 00:59:11.750
with each other?
00:59:11.760 --> 00:59:13.030
>> Good point. Yeah. There's no shared
00:59:13.040 --> 00:59:14.390
objective, no shared strength. Yeah.
00:59:14.400 --> 00:59:17.670
Yeah. Um so yeah, interesting times
00:59:17.680 --> 00:59:18.309
though, right?
00:59:18.319 --> 00:59:20.549
>> Oh yeah. Yeah. Yeah. Yeah.
00:59:20.559 --> 00:59:21.910
>> Excellent. All right. Well, that was
00:59:21.920 --> 00:59:23.990
another good uh hour. It's amazing how
00:59:24.000 --> 00:59:25.670
fast time flies. Thank you for everybody
00:59:25.680 --> 00:59:28.150
that came on and uh anybody that's going
00:59:28.160 --> 00:59:30.309
to listen to it afterwards again. Uh
00:59:30.319 --> 00:59:32.230
we're still looking to be driven by the
00:59:32.240 --> 00:59:34.230
community, right, by our audience. So,
00:59:34.240 --> 00:59:37.589
uh and I keep forgetting to to uh to do
00:59:37.599 --> 00:59:40.150
that at the end. So, uh yep, hop hop on
00:59:40.160 --> 00:59:42.870
Substack, ask us questions on any of the
00:59:42.880 --> 00:59:44.630
ones we've done before or any ones we're
00:59:44.640 --> 00:59:47.270
going to do and um we will go from
00:59:47.280 --> 00:59:49.750
there.
00:59:49.760 --> 00:59:51.510
>> Yep. Oh, thank you.
00:59:51.520 --> 00:59:53.030
>> My screen just timed itself out, so
00:59:53.040 --> 00:59:55.589
that's right. [laughter]
00:59:55.599 --> 00:59:57.190
>> Excellent. All right, we'll see you
00:59:57.200 --> 00:59:58.470
again in a month.
00:59:58.480 --> 01:00:02.200
>> See you. Thanks.