206160a8-ab37-434f-b504-a480ea398a54-0
00:00:00.200 --> 00:00:00.480
All right.

d58e5e1e-aedd-40e5-ab6b-579142bbbb93-0
00:00:00.480 --> 00:00:04.520
Welcome back to the Mostly Unstructured
podcast sponsored by KeyMark.

baf3239d-bcd5-4b15-acd9-53ab8713bc2d-0
00:00:04.920 --> 00:00:06.640
Ed and Clay in the chair again.

f4209c29-9223-4073-bbca-7fc3cb2a6d3a-0
00:00:06.840 --> 00:00:07.840
Yeah, exactly.

37553ce4-06be-4b8c-8665-235a79f9422a-0
00:00:07.880 --> 00:00:09.080
Yeah, I didn't know we sponsored it.

491fbec7-1872-4c78-b0e0-eecf0f68889b-0
00:00:10.200 --> 00:00:11.600
Maybe we should start over again.

dc5147d2-146c-4910-84d7-e87b5b2266cb-0
00:00:24.280 --> 00:00:24.680
Hey there.

9f901b4a-7c40-4680-b44e-6b526f892ea2-0
00:00:24.680 --> 00:00:27.200
Welcome back to the Mostly Unstructured
podcast.

346a63fd-f41b-4758-ad38-b18f5ca8a814-0
00:00:27.200 --> 00:00:29.040
Clay and Ed in the chair again.

998de89d-3e73-4070-9750-d3e99cf750ff-0
00:00:29.160 --> 00:00:31.926
Yep,
really excited about this topic we're

998de89d-3e73-4070-9750-d3e99cf750ff-1
00:00:31.926 --> 00:00:33.600
going to talk about today.

43e8900c-047d-4094-9535-9580ade58410-0
00:00:34.080 --> 00:00:39.594
That may seem a little odd when we're
talking about why data governance is the

43e8900c-047d-4094-9535-9580ade58410-1
00:00:39.594 --> 00:00:44.480
foundation of enterprise AI and you're
like, why is that so exciting?

02bd99a5-c14f-4900-9b6f-fb4329154923-0
00:00:44.640 --> 00:00:46.425
Well,
I think it's when you have really

02bd99a5-c14f-4900-9b6f-fb4329154923-1
00:00:46.425 --> 00:00:48.880
important information you want to share
with somebody.

b127ff32-1707-4519-9c79-9943d2b3b4e7-0
00:00:48.880 --> 00:00:49.840
Good news.

1663952a-8286-470f-a747-e224841a1afc-0
00:00:49.960 --> 00:00:55.021
Hey, you know, look out for the bus like,
and it is because this is a is a critical

1663952a-8286-470f-a747-e224841a1afc-1
00:00:55.021 --> 00:00:59.600
thing that if you don't get right, it's,
it's detrimental to your business.

f8a5801a-25a9-497c-bcdf-12e447b89782-0
00:00:59.600 --> 00:01:07.601
And so we've talked personally about this
AI conversation being so dominant in the

f8a5801a-25a9-497c-bcdf-12e447b89782-1
00:01:07.601 --> 00:01:09.240
boardroom, right?

68f00c4d-cfca-4494-a669-af0fd4208ffb-0
00:01:09.240 --> 00:01:13.906
And but what's interesting is that the
biggest conversation around AI right now

68f00c4d-cfca-4494-a669-af0fd4208ffb-1
00:01:13.906 --> 00:01:17.640
really isn't about what you might think
like model performance.

60bce48c-e4fa-43ea-823d-80acf21cafc9-0
00:01:17.640 --> 00:01:20.440
It's around this idea of of governance.

00749d6f-a9cf-47be-87b0-b16807a4d2dc-0
00:01:20.440 --> 00:01:23.806
And so in fact,
I think we mentioned in a previous

00749d6f-a9cf-47be-87b0-b16807a4d2dc-1
00:01:23.806 --> 00:01:29.284
episode that analysts are saying that the
data governance is a non negotiable when

00749d6f-a9cf-47be-87b0-b16807a4d2dc-2
00:01:29.284 --> 00:01:31.000
it comes to enterprise AI.

24c0205d-71bd-4a84-aab9-1fb72a256ab7-0
00:01:31.000 --> 00:01:33.520
So today we just want to unpack that a
little bit.

f2a46953-f3ac-4cbf-b343-f78efdae640f-0
00:01:33.520 --> 00:01:37.840
Talk about why is this a new thing?

8de9a15b-7147-48a0-ac7a-ad6db8461b0d-0
00:01:38.000 --> 00:01:41.043
I think I remember you saying recently,
it's been a while,

8de9a15b-7147-48a0-ac7a-ad6db8461b0d-1
00:01:41.043 --> 00:01:44.861
even when it's kind of got started,
people were already picking up on the

8de9a15b-7147-48a0-ac7a-ad6db8461b0d-2
00:01:44.861 --> 00:01:48.266
risk associated with AI,
something you're seeing regularly in the

8de9a15b-7147-48a0-ac7a-ad6db8461b0d-3
00:01:48.266 --> 00:01:49.040
C-Suite, right?

b297a938-5d45-4c63-9931-a38c1b1d1950-0
00:01:49.160 --> 00:01:50.680
Yeah, no, absolutely.

11350d89-93f2-4c66-b7f8-1ee9c035887a-0
00:01:50.680 --> 00:01:54.931
And, and yeah, I mean, I think,
I think to your point, you don't,

11350d89-93f2-4c66-b7f8-1ee9c035887a-1
00:01:54.931 --> 00:01:59.440
you don't often necessarily correlate
like, Oh my God, I'm so pumped.

17b67b09-d47f-43cf-b694-142c47a35de3-0
00:01:59.440 --> 00:02:00.680
We're going to talk about governance.

f015d043-d1f5-46c2-b2cc-f17a0777e732-0
00:02:02.000 --> 00:02:04.843
But I think, you know,
what you're saying is,

f015d043-d1f5-46c2-b2cc-f17a0777e732-1
00:02:04.843 --> 00:02:06.080
is absolutely right.

1f202aec-2301-49fc-b57d-b7de5e150e33-0
00:02:06.160 --> 00:02:09.448
I, I think this,
the cycle that's happened, of course,

1f202aec-2301-49fc-b57d-b7de5e150e33-1
00:02:09.448 --> 00:02:11.720
is people began experimenting with AI.

dfec3754-0900-40c8-bcba-d00b3689ff77-0
00:02:12.280 --> 00:02:17.658
And as they've been experimenting,
they've begun to run into some of the

dfec3754-0900-40c8-bcba-d00b3689ff77-1
00:02:17.658 --> 00:02:22.080
realities that come with ungoverned AI
and what that means.

eb534478-8c2f-41fe-a72a-26e7a7346a00-0
00:02:22.440 --> 00:02:28.338
And so those risk factors of course make
it up to the C-Suite because what

eb534478-8c2f-41fe-a72a-26e7a7346a00-1
00:02:28.338 --> 00:02:32.742
transpires is you go from this
productivity efficiency,

eb534478-8c2f-41fe-a72a-26e7a7346a00-2
00:02:32.742 --> 00:02:38.562
exciting innovation that we're trying to
accomplish as an organization to

eb534478-8c2f-41fe-a72a-26e7a7346a00-3
00:02:38.562 --> 00:02:44.382
something with potentially significant
liability if you don't address the

eb534478-8c2f-41fe-a72a-26e7a7346a00-4
00:02:44.382 --> 00:02:45.719
governance piece.

7fe410bd-3bb1-4165-94c3-7d6068fbad7c-0
00:02:45.720 --> 00:02:48.379
And so, you know, I think that it's,
it's clearly,

7fe410bd-3bb1-4165-94c3-7d6068fbad7c-1
00:02:48.379 --> 00:02:50.987
it's top of mind for for me and our
organization,

7fe410bd-3bb1-4165-94c3-7d6068fbad7c-2
00:02:50.987 --> 00:02:53.959
it's top of mind to the the customers
that we talked to.

ca6e5ba0-eda9-43ed-9341-bcd844b123ee-0
00:02:54.120 --> 00:02:54.440
Yeah.

2d99ae71-549c-4332-8b5c-29e237b175cf-0
00:02:54.880 --> 00:03:00.384
So let's start out by kind of defining it
a little bit and what we mean by

2d99ae71-549c-4332-8b5c-29e237b175cf-1
00:03:00.384 --> 00:03:02.440
enterprise grade governance.

870df75a-d6ee-4b57-a380-8180ebd3e991-0
00:03:03.080 --> 00:03:07.400
First of all,
governance isn't just IT security.

93cc2b14-4441-4be2-acfa-c0ac62a1ecea-0
00:03:07.400 --> 00:03:11.940
I think it's,
it's a framework that ensures AI systems

93cc2b14-4441-4be2-acfa-c0ac62a1ecea-1
00:03:11.940 --> 00:03:18.380
are developed, that they're deployed,
maintained and used in a safe way that's

93cc2b14-4441-4be2-acfa-c0ac62a1ecea-2
00:03:18.380 --> 00:03:24.736
in alignment with the privacy laws,
with your ethical standards and business

93cc2b14-4441-4be2-acfa-c0ac62a1ecea-3
00:03:24.736 --> 00:03:25.480
strategy.

1cd115fa-7b6a-486a-b457-51a897f26ef1-0
00:03:25.920 --> 00:03:30.659
So when we talk about data governance,
what are some of the things that that

1cd115fa-7b6a-486a-b457-51a897f26ef1-1
00:03:30.659 --> 00:03:35.337
might include an organization that you,
you know, you would talk about and,

1cd115fa-7b6a-486a-b457-51a897f26ef1-2
00:03:35.337 --> 00:03:38.600
and emphasize what,
what areas are we talking about?

1b33f464-47c6-49b0-b784-d10f518a3a48-0
00:03:39.480 --> 00:03:42.307
Well,
I want to go back a little bit because

1b33f464-47c6-49b0-b784-d10f518a3a48-1
00:03:42.307 --> 00:03:46.518
organizationally KeyMark, you know, my,
myself, my, my career in,

1b33f464-47c6-49b0-b784-d10f518a3a48-2
00:03:46.518 --> 00:03:50.665
in this unstructured industry,
governance has been at the root of

1b33f464-47c6-49b0-b784-d10f518a3a48-3
00:03:50.665 --> 00:03:53.933
conversations forever,
But it's been in the root of

1b33f464-47c6-49b0-b784-d10f518a3a48-4
00:03:53.933 --> 00:03:58.772
conversations, you know, often related to,
you know, the core of what we do,

1b33f464-47c6-49b0-b784-d10f518a3a48-5
00:03:58.772 --> 00:04:03.360
which has been storing corporate
information, often corporate documents.

5351a83c-422a-4bae-80a0-bdf2c0441bcc-0
00:04:03.360 --> 00:04:06.613
They might originate digitally,
they might be scanned in or,

5351a83c-422a-4bae-80a0-bdf2c0441bcc-1
00:04:06.613 --> 00:04:07.520
or what have you.

282f84ce-b729-4d6a-872b-b039760596a2-0
00:04:08.000 --> 00:04:11.758
And, you know,
governance was things like what kind of

282f84ce-b729-4d6a-872b-b039760596a2-1
00:04:11.758 --> 00:04:12.920
document is this?

e70719be-2346-43f1-998b-db2c522f4a5c-0
00:04:13.240 --> 00:04:16.968
And based on the kind of document it is,
who's going to have rights to it and

e70719be-2346-43f1-998b-db2c522f4a5c-1
00:04:16.968 --> 00:04:19.120
who's going to be able to make it
available?

9f211695-0f27-4e78-a19f-f32319090893-0
00:04:19.120 --> 00:04:25.240
And with that comes this, you know,
governance model.

f1a6963c-de8d-4864-91d9-855a2cecda33-0
00:04:25.840 --> 00:04:28.954
But what happens when if you get it wrong,
right,

f1a6963c-de8d-4864-91d9-855a2cecda33-1
00:04:28.954 --> 00:04:33.065
if you get it wrong and you define a
document as the wrong thing,

f1a6963c-de8d-4864-91d9-855a2cecda33-2
00:04:33.065 --> 00:04:34.559
maybe you can't find it.

4eb50433-b5b4-4637-ab1c-344a32676458-0
00:04:34.920 --> 00:04:36.560
That's sort of the penalty, right?

0c2c3818-63de-4f01-847f-aedad57153f4-0
00:04:36.680 --> 00:04:37.520
You can't find it.

36027db0-a80d-4176-98bb-dc1232004bbf-0
00:04:38.320 --> 00:04:41.880
What if the data is wrong in an AMI model?

56922f55-8896-415f-854d-6089a6f3d5a5-0
00:04:42.160 --> 00:04:46.935
What if you're, you know, pull,
pulling data across, You know,

56922f55-8896-415f-854d-6089a6f3d5a5-1
00:04:46.935 --> 00:04:52.696
we've talked in this podcast about sort
of reaching into these unstructured

56922f55-8896-415f-854d-6089a6f3d5a5-2
00:04:52.696 --> 00:04:58.760
repositories, harvesting data from them,
structuring it, sharing it downstream.

02123dc9-e64b-4859-856b-1a93f8bcb6b7-0
00:04:59.600 --> 00:05:03.440
What if you didn't govern that and just
move that data across?

b264c71c-057a-40ed-b090-96d86fc80dc1-0
00:05:03.440 --> 00:05:05.520
Well, the penalty isn't, hey,
I can't just find it.

fe748a42-a0d5-4c3d-b72c-4c39b9eb6118-0
00:05:05.800 --> 00:05:09.440
The penalty could be an agent makes
decisions on flat out bad data.

95bdb117-a82f-424d-ab3a-f53dc78d10e1-0
00:05:10.400 --> 00:05:16.240
And now that's risk,
that's potential liability.

2535d844-3d3c-4057-be52-46cf867f13c4-0
00:05:16.240 --> 00:05:19.345
That's, you know,
if you're on the phone with a claimant

2535d844-3d3c-4057-be52-46cf867f13c4-1
00:05:19.345 --> 00:05:22.177
and you use an,
an agent to ask a question and it's

2535d844-3d3c-4057-be52-46cf867f13c4-2
00:05:22.177 --> 00:05:26.699
literally querying bad data that's been
brought across because nobody governed it,

2535d844-3d3c-4057-be52-46cf867f13c4-3
00:05:26.699 --> 00:05:27.680
nobody checked it.

bed304be-6858-4228-8d4a-e01e15cfb488-0
00:05:29.560 --> 00:05:33.810
There's a whole different level of risk
associated than what we've seen in the

bed304be-6858-4228-8d4a-e01e15cfb488-1
00:05:33.810 --> 00:05:34.080
past.

c7b314a1-d04c-493d-9f7e-8b3027affbd2-0
00:05:34.120 --> 00:05:34.480
Yeah.

0cd006dc-38aa-49d6-b333-0883a4a1e2c7-0
00:05:34.640 --> 00:05:37.640
And it, it has major ramifications.

b7695a20-6417-420a-b80c-c0cc2b8e92c7-0
00:05:38.120 --> 00:05:39.240
So we talked about that.

d7733aa2-d69a-4384-8db6-184457405571-0
00:05:39.240 --> 00:05:42.600
I think we're talking about, you know,
risk mitigation.

f70b5b66-ca93-4e03-9f14-4f2edfea0b07-0
00:05:42.800 --> 00:05:46.527
You talk about documents talking
traceability,

f70b5b66-ca93-4e03-9f14-4f2edfea0b07-1
00:05:46.527 --> 00:05:52.080
monitoring and auditing policy creation
is a whole plethora of items.

063995d9-8c15-4b64-9c1a-cf5bb9c2bad3-0
00:05:52.640 --> 00:05:59.082
When you think of all of those at an
enterprise level who should own this

063995d9-8c15-4b64-9c1a-cf5bb9c2bad3-1
00:05:59.082 --> 00:06:00.040
governance.

9ca233ea-a55c-42a4-a210-bc9fcd005076-0
00:06:00.080 --> 00:06:03.822
OK,
because I think that it in a lot of ways,

9ca233ea-a55c-42a4-a210-bc9fcd005076-1
00:06:03.822 --> 00:06:04.880
well it's IT?

a4bf2aeb-655b-400b-aaf0-338c9001a3f8-0
00:06:05.440 --> 00:06:07.520
But is it is there more to that?

cb2aa481-9b04-4267-bffb-44fd32f9a5b9-0
00:06:07.520 --> 00:06:12.082
Like what, who, who,
who's in charge of this was in in

cb2aa481-9b04-4267-bffb-44fd32f9a5b9-1
00:06:12.082 --> 00:06:14.240
thinking about this topic?

217fc79d-958c-46f2-8ffb-25b67cd35c09-0
00:06:14.240 --> 00:06:19.083
There's a couple of phrases that I ran
into that you know are already popping up

217fc79d-958c-46f2-8ffb-25b67cd35c09-1
00:06:19.083 --> 00:06:20.280
around some of this.

f024ef1f-e8fa-4a55-ab9e-4a084ffd688f-0
00:06:20.640 --> 00:06:25.228
One of this idea,
one of the ideas there is this idea of

f024ef1f-e8fa-4a55-ab9e-4a084ffd688f-1
00:06:25.228 --> 00:06:27.160
data and being AI ready.

a52fe00b-c8cb-4bca-97cf-4d43aa225d77-0
00:06:27.280 --> 00:06:29.480
And what does AI ready mean?

9a9e663d-3e21-4264-a937-457de0b332d4-0
00:06:29.480 --> 00:06:32.663
Well, you know,
one of the dangers associated with not

9a9e663d-3e21-4264-a937-457de0b332d4-1
00:06:32.663 --> 00:06:35.442
having governance is the things I talked
about,

9a9e663d-3e21-4264-a937-457de0b332d4-2
00:06:35.442 --> 00:06:39.668
pulling data off without knowing
necessarily what it is or validating it

9a9e663d-3e21-4264-a937-457de0b332d4-3
00:06:39.668 --> 00:06:41.000
or any of those things.

39665cf8-b6be-44e8-a411-4501dbf83166-0
00:06:41.280 --> 00:06:44.090
Not having the traceability,
not having an audit trail,

39665cf8-b6be-44e8-a411-4501dbf83166-1
00:06:44.090 --> 00:06:46.600
not knowing where it came from and
serving it up.

b60e469a-c347-4871-8e9e-1a7c63878385-0
00:06:47.040 --> 00:06:49.814
OK,
let's say we serve up bad data to

b60e469a-c347-4871-8e9e-1a7c63878385-1
00:06:49.814 --> 00:06:51.640
externally or internally.

761a1671-5314-4da4-8feb-fd4df6c1befb-0
00:06:52.760 --> 00:06:55.280
How do we have that back ungoverned?

919ce2a9-b3ea-4148-a538-e722e6ce69f6-0
00:06:55.520 --> 00:06:56.080
We don't.

6bf1c1ef-1296-4d01-b739-ff1d2f44d601-0
00:06:56.400 --> 00:06:59.560
And they're clearly there's a risk
associated with that.

860d6c70-f7c1-4af6-978a-2b45b8f8d986-0
00:06:59.560 --> 00:07:01.960
So to your point, who owns that?

7aac46b9-c1bb-48b4-baf8-d830541a23ec-0
00:07:02.120 --> 00:07:05.416
Well, there's typically, you know,
I think that lies in a couple of

7aac46b9-c1bb-48b4-baf8-d830541a23ec-1
00:07:05.416 --> 00:07:06.240
different places.

ad429a2f-0bf1-4a15-b007-a43741d3ed7a-0
00:07:06.240 --> 00:07:09.160
Certainly organizations have chief data
officers.

e76b0da6-a23a-4451-be0e-8ab0c17fc3eb-0
00:07:09.160 --> 00:07:14.263
You're seeing chief data information
officers and the ideas of data governance

e76b0da6-a23a-4451-be0e-8ab0c17fc3eb-1
00:07:14.263 --> 00:07:19.560
committees that involve various business
owners along with those CD OS and CDI OS

e76b0da6-a23a-4451-be0e-8ab0c17fc3eb-2
00:07:19.560 --> 00:07:22.080
that come together to put some of this.

a10c2ff2-175d-46d2-8c6a-d9c095e37f9a-0
00:07:22.360 --> 00:07:25.640
So I think that that ideas is is
incredibly important.

ac94e20f-d45e-4d01-a0fc-d66747c84ab7-0
00:07:26.280 --> 00:07:32.714
I think part of the importance there
though isn't just the model, the process,

ac94e20f-d45e-4d01-a0fc-d66747c84ab7-1
00:07:32.714 --> 00:07:35.240
but the the rules, so to speak.

280d2bcc-24c3-41a4-8482-b10ae0af6c96-0
00:07:35.240 --> 00:07:38.706
So one of the other terms that I ran
across,

280d2bcc-24c3-41a4-8482-b10ae0af6c96-1
00:07:38.706 --> 00:07:42.943
we've all heard about shadow IT and in
the age of SAS,

280d2bcc-24c3-41a4-8482-b10ae0af6c96-2
00:07:42.943 --> 00:07:45.640
shadow IT has just exploded, right?

5f1b2fa6-f702-48b1-82db-ac3f125a1030-0
00:07:45.640 --> 00:07:50.075
Because there's a, you know,
business owner or business unit owner

5f1b2fa6-f702-48b1-82db-ac3f125a1030-1
00:07:50.075 --> 00:07:53.650
with a budget,
I can go find a product that was built

5f1b2fa6-f702-48b1-82db-ac3f125a1030-2
00:07:53.650 --> 00:07:56.960
for me,
a SAS product that I don't have to depend

5f1b2fa6-f702-48b1-82db-ac3f125a1030-3
00:07:56.960 --> 00:07:58.880
on IT to utilize necessarily.

2ef4c1b5-696b-43da-8168-703d0d4b1d8e-0
00:07:58.880 --> 00:08:00.040
And you see the proliferation.

3d8637b7-6549-4d83-97d0-473e78ffd369-0
00:08:00.480 --> 00:08:04.280
What I hadn't heard of until recently was
this idea of shadow AI.

af1c68fe-d951-49c5-86bc-95bfe2035d65-0
00:08:05.880 --> 00:08:10.289
And so much of what's available are these
tool kits, these, you know,

af1c68fe-d951-49c5-86bc-95bfe2035d65-1
00:08:10.289 --> 00:08:15.140
this platform level tool kits from the
hyperscalers where I could, you know,

af1c68fe-d951-49c5-86bc-95bfe2035d65-2
00:08:15.140 --> 00:08:20.116
go in and use something to from AWS or
Azure or whatever and do my own version

af1c68fe-d951-49c5-86bc-95bfe2035d65-3
00:08:20.116 --> 00:08:24.400
of reaching and grab the data and push it
into an agentic solution.

764b644f-8df2-4dcb-aa08-660923dbc8c6-0
00:08:25.040 --> 00:08:26.880
Well, that's a problem, right?

6c099087-988f-4029-a61c-dab1302dab5a-0
00:08:27.280 --> 00:08:33.774
So it's not only the governing the data,
but it's governing the use of these tools

6c099087-988f-4029-a61c-dab1302dab5a-1
00:08:33.774 --> 00:08:38.000
and who has access and what they can go
do with them.

ab87256a-d334-422c-b4a7-5f33b4cd6e81-0
00:08:38.000 --> 00:08:42.840
Because you could be doing everything
right on your policy side, etcetera.

828fd3bb-b7ae-43a1-9369-f0a8cd35e95f-0
00:08:42.840 --> 00:08:45.120
You can be reviewing data,
creating audit trails.

76ef547b-ef14-4094-8cf5-70ba46dfb30e-0
00:08:45.400 --> 00:08:50.333
You also have to take a look at this this
kind of shadow AI idea, which, you know,

76ef547b-ef14-4094-8cf5-70ba46dfb30e-1
00:08:50.333 --> 00:08:51.760
I think is is very real.

44ce75b0-26ca-4ed9-8caa-f0243c7f52ff-0
00:08:52.000 --> 00:08:52.200
Yeah.

38dc5e62-dfc4-46ed-9fd9-e73bb3f9f1fe-0
00:08:52.960 --> 00:08:56.978
So you're talking about the fact that
there's multiple people involved in the

38dc5e62-dfc4-46ed-9fd9-e73bb3f9f1fe-1
00:08:56.978 --> 00:08:59.760
organization that should be a part of
that committee.

925e1afb-7631-418d-8eb7-b9a316e95d24-0
00:08:59.760 --> 00:09:02.640
And I want to say this too about the
committee idea.

a36a17e2-ea51-4f63-9d77-96905c69abc5-0
00:09:03.080 --> 00:09:06.880
I think that's going to be table stakes
in the next two or three years.

747c53de-c7b8-4fa9-bc76-49324f86194f-0
00:09:07.240 --> 00:09:13.675
And by the way, I say two or three years,
I wonder if we shouldn't create an AI dog

747c53de-c7b8-4fa9-bc76-49324f86194f-1
00:09:13.675 --> 00:09:20.033
years kind of concept without a question,
because there's things that we've talked

747c53de-c7b8-4fa9-bc76-49324f86194f-2
00:09:20.033 --> 00:09:25.472
about 10 months ago that either or old
news have changed completely or

747c53de-c7b8-4fa9-bc76-49324f86194f-3
00:09:25.472 --> 00:09:28.919
irrelevant,
like it's just changing so fast.

d80753f5-13eb-4690-b3ff-e1af208fc4bf-0
00:09:28.920 --> 00:09:30.920
And like you say, well,
your dog's five years old.

a6a87561-8ddf-40bd-b5b1-7949bf199e30-0
00:09:30.920 --> 00:09:33.567
Well,
really he's 35 because one years we're

a6a87561-8ddf-40bd-b5b1-7949bf199e30-1
00:09:33.567 --> 00:09:33.920
seven.

fe963247-6b5b-4595-855d-0dbd3d14b567-0
00:09:34.120 --> 00:09:35.400
I think we should do that with AI.

9b761076-7d45-4629-a334-308aaa48ae11-0
00:09:35.760 --> 00:09:38.535
But I saw,
I saw a quote the other day that was

9b761076-7d45-4629-a334-308aaa48ae11-1
00:09:38.535 --> 00:09:42.640
attributed to Prime Minister Trudeau from
Canada back a few years ago.

1d12a9f6-60ae-4684-8d96-d08e01017100-0
00:09:42.640 --> 00:09:49.181
And he said something along the lines of
we've never moved this fast before and

1d12a9f6-60ae-4684-8d96-d08e01017100-1
00:09:49.181 --> 00:09:51.880
we'll never move this slow again.

8a4055d4-db92-42d2-a273-4ebce2b2f89a-0
00:09:52.320 --> 00:09:55.560
Wow,
which feels about right for where we are.

d5ad1ca6-55fe-4872-9a54-45365df3d666-0
00:09:55.840 --> 00:09:56.840
That's right, That's right.

011d68c5-93d1-406f-9097-3d28ad20e25d-0
00:09:57.200 --> 00:10:00.287
Well,
I say that because when we throw out

011d68c5-93d1-406f-9097-3d28ad20e25d-1
00:10:00.287 --> 00:10:05.456
stuff like, you know, two to three years,
you know, how long, how long,

011d68c5-93d1-406f-9097-3d28ad20e25d-2
00:10:05.456 --> 00:10:08.400
how long is 2 to three years in AI years?

f2b96906-8563-4e3c-b541-66248a1d6df6-0
00:10:08.400 --> 00:10:11.440
Like if I say 5,
that could be in the year and a half.

f9604820-a0da-4396-8872-ae0a64b2b6aa-0
00:10:11.440 --> 00:10:13.800
It's moving so, so quickly.

9ba785c2-4cf0-4012-af44-5806692c82a4-0
00:10:13.800 --> 00:10:18.726
But one of the things we, you know,
we're pointing out is that it is going to

9ba785c2-4cf0-4012-af44-5806692c82a4-1
00:10:18.726 --> 00:10:22.642
be important that they have a committee
in your organization,

9ba785c2-4cf0-4012-af44-5806692c82a4-2
00:10:22.642 --> 00:10:24.600
but that's not always possible.

a97c245d-c5be-4a91-8629-d6f0d126da00-0
00:10:24.880 --> 00:10:28.320
And I think a lot of people now don't
like how. Where do we even start?

a31a4f32-11f1-453d-ae39-0093d09a9d85-0
00:10:28.440 --> 00:10:31.160
And we don't want to go down a rabbit
hole of how to build a committee.

11b2369b-393e-43db-8b30-ced47d00468a-0
00:10:31.160 --> 00:10:35.174
But I didn't want to just throw this out
there that we're seeing a lot of people

11b2369b-393e-43db-8b30-ced47d00468a-1
00:10:35.174 --> 00:10:35.720
come to us.

9ae7f2bf-2be5-42dc-b291-78026ad4ada1-0
00:10:35.720 --> 00:10:40.360
We talked to other AI folks in our space
who are all kind of saying the same thing.

c1fa482c-1534-4e97-bdc1-3b2f0227f72a-0
00:10:40.360 --> 00:10:44.960
And that is most organizations are,
I know I have the ring.

b2a223dc-7efa-4e57-8e61-993b2a1804b0-0
00:10:46.200 --> 00:10:50.850
I need to get it into Mordor, you know,
I need AI, need a guide,

b2a223dc-7efa-4e57-8e61-993b2a1804b0-1
00:10:50.850 --> 00:10:56.002
I need somebody at least help me get
started to show me where to go and

b2a223dc-7efa-4e57-8e61-993b2a1804b0-2
00:10:56.002 --> 00:11:01.011
where's my Gandalf, so to speak,
right and so this idea of all right,

b2a223dc-7efa-4e57-8e61-993b2a1804b0-3
00:11:01.011 --> 00:11:04.159
you don't have to go set this up right
now.

74af6af2-77a3-4ac7-b669-b8d2727eb9fa-0
00:11:04.160 --> 00:11:09.286
The best thing you can do instead of
going and trying to start a committee,

74af6af2-77a3-4ac7-b669-b8d2727eb9fa-1
00:11:09.286 --> 00:11:13.198
lean into a partner,
lean into another AI expert that you

74af6af2-77a3-4ac7-b669-b8d2727eb9fa-2
00:11:13.198 --> 00:11:18.730
trust and work together to create that so
that you're getting your sea legs under

74af6af2-77a3-4ac7-b669-b8d2727eb9fa-3
00:11:18.730 --> 00:11:19.000
you.

690802ba-b2c8-48ee-a123-ee87ece2709b-0
00:11:19.000 --> 00:11:22.335
At some point, it'll be,
I think you would probably agree it's

690802ba-b2c8-48ee-a123-ee87ece2709b-1
00:11:22.335 --> 00:11:24.400
going to be part of every organization.

b9ccf602-cf29-431f-a81e-c18ab229ee57-0
00:11:24.520 --> 00:11:27.860
Like, you know,
that round table of AI people that are in

b9ccf602-cf29-431f-a81e-c18ab229ee57-1
00:11:27.860 --> 00:11:28.840
the organization.

28a5122e-c557-4093-8bf6-fe8715dc7768-0
00:11:28.840 --> 00:11:30.360
Yeah, I, I, I completely agree.

77588355-7cfe-4490-ba10-5592f3967453-0
00:11:30.360 --> 00:11:33.089
And I, you know,
I want to go back to your time comment

77588355-7cfe-4490-ba10-5592f3967453-1
00:11:33.089 --> 00:11:35.916
just just for a moment that, you know,
idea of, you know,

77588355-7cfe-4490-ba10-5592f3967453-2
00:11:35.916 --> 00:11:37.720
what three years from now looks like.

0a3e7df3-2faf-4310-ac06-f60ee16c9c9c-0
00:11:37.720 --> 00:11:41.083
You know,
I keep coming back to the idea that three

0a3e7df3-2faf-4310-ac06-f60ee16c9c9c-1
00:11:41.083 --> 00:11:45.547
years and three months ago,
ChatGPT had not yet been unveiled to the

0a3e7df3-2faf-4310-ac06-f60ee16c9c9c-2
00:11:45.547 --> 00:11:46.000
public.

95dd57e2-95c9-4b51-bdd9-8e36c5b2c274-0
00:11:46.640 --> 00:11:50.960
And you think about the ubiquitous use of
these tools today.

35f9f06c-fe11-4a62-9029-54d22798666e-0
00:11:50.960 --> 00:11:55.602
It's every conversation in technology,
it's every conversation in the boardroom,

35f9f06c-fe11-4a62-9029-54d22798666e-1
00:11:55.602 --> 00:11:59.556
The influence in the stock market lately
on AI and what's happening,

35f9f06c-fe11-4a62-9029-54d22798666e-2
00:11:59.556 --> 00:12:00.760
what's not happening.

0eca1824-9bdd-4513-aadf-e7581e802c6a-0
00:12:00.760 --> 00:12:05.986
And so this idea of trying to project out
more than three years at this point,

0eca1824-9bdd-4513-aadf-e7581e802c6a-1
00:12:05.986 --> 00:12:10.550
I think is an impossibility because
you've you've now got, you know,

0eca1824-9bdd-4513-aadf-e7581e802c6a-2
00:12:10.550 --> 00:12:12.800
AI driving AI based conversations.

1b483c51-f517-4131-a389-752b1652abd9-0
00:12:12.800 --> 00:12:15.680
You're you're going into AI saying what's
going to happen with AI?

d3cdba11-22d1-4e74-ae54-6393b10e0aa1-0
00:12:15.680 --> 00:12:17.760
I mean that that's literally where we are.

e00ba706-fd4a-47ea-a84f-22a92acf741b-0
00:12:18.160 --> 00:12:20.570
So you back to your point about the,
you know,

e00ba706-fd4a-47ea-a84f-22a92acf741b-1
00:12:20.570 --> 00:12:22.520
those committees and things like that.

82f06697-3492-4089-9eba-486ebd45a63d-0
00:12:22.880 --> 00:12:28.120
One of the things that we've talked about
before is failure rates, right?

6dc7681a-5f8c-4125-ba90-4d25a7a241f3-0
00:12:28.120 --> 00:12:33.319
This this now famous or infamous MIT
study of the 95% failure rate and those

6dc7681a-5f8c-4125-ba90-4d25a7a241f3-1
00:12:33.319 --> 00:12:34.400
kinds of things.

b96cdb8f-9387-45d4-8ba5-c46e8b10a62c-0
00:12:34.920 --> 00:12:38.523
I think what's, you know,
what's what's occurring is you're,

b96cdb8f-9387-45d4-8ba5-c46e8b10a62c-1
00:12:38.523 --> 00:12:42.363
you're seeing this sort of first bastions
of, OK, we tried this,

b96cdb8f-9387-45d4-8ba5-c46e8b10a62c-2
00:12:42.363 --> 00:12:43.840
it didn't go as expected.

63bb83e4-2150-43d2-b3ba-222400492cb6-0
00:12:43.840 --> 00:12:44.720
Now what?

0aadafa6-11be-4ff0-bd58-d2e15d7cf85a-0
00:12:45.160 --> 00:12:50.252
And I think a lot of those decisions got
made with people's big platform vendors,

0aadafa6-11be-4ff0-bd58-d2e15d7cf85a-1
00:12:50.252 --> 00:12:52.240
big platform consultants, right?

2c8c339b-c7ff-4172-b928-7f8e26ba0971-0
00:12:52.240 --> 00:12:54.488
They're,
they're going to the people they spend

2c8c339b-c7ff-4172-b928-7f8e26ba0971-1
00:12:54.488 --> 00:12:57.720
the most money with and saying, OK, help,
you know, help me do this.

74d73105-87b5-477c-a609-6cd1faa23589-0
00:12:58.080 --> 00:13:03.881
That doesn't necessarily lend itself to
deep understanding of the customer's

74d73105-87b5-477c-a609-6cd1faa23589-1
00:13:03.881 --> 00:13:04.560
business.

68a9e130-8ba8-438d-96e1-d6e61bef6e41-0
00:13:04.760 --> 00:13:08.740
You know, our,
our business being more of a boutique,

68a9e130-8ba8-438d-96e1-d6e61bef6e41-1
00:13:08.740 --> 00:13:13.162
we're in very deeply with the customers
at solution levels,

68a9e130-8ba8-438d-96e1-d6e61bef6e41-2
00:13:13.162 --> 00:13:16.847
at business understanding,
process understanding,

68a9e130-8ba8-438d-96e1-d6e61bef6e41-3
00:13:16.847 --> 00:13:21.932
often people understanding and those
things become critical in these

68a9e130-8ba8-438d-96e1-d6e61bef6e41-4
00:13:21.932 --> 00:13:26.797
conversations because one,
you're talking about the governance of

68a9e130-8ba8-438d-96e1-d6e61bef6e41-5
00:13:26.797 --> 00:13:28.640
their sacred cows, right?

ed056d94-5b74-4a58-acee-4c8db7199f5e-0
00:13:28.640 --> 00:13:30.720
This is the keys to their Kingdom.

9facfde1-bbd2-49e7-89f7-1dae879e6752-0
00:13:30.720 --> 00:13:35.443
We don't we don't generally get involved
with kind of less critical information

9facfde1-bbd2-49e7-89f7-1dae879e6752-1
00:13:35.443 --> 00:13:40.343
for organizations we're typically in this
is the most critical information we have

9facfde1-bbd2-49e7-89f7-1dae879e6752-2
00:13:40.343 --> 00:13:44.240
might be medical charts,
might be claims data, might be whatever.

bfb169bc-8750-4c36-8287-ce410083fd9d-0
00:13:44.480 --> 00:13:49.511
And as part of that,
these discussions lend themselves to a

bfb169bc-8750-4c36-8287-ce410083fd9d-1
00:13:49.511 --> 00:13:55.800
need for advice and the customers are in
an incredibly difficult position.

14b75857-0ffb-4efa-a2f8-90412c742d97-0
00:13:56.120 --> 00:13:59.840
The speed at which they're being required
to learn, assess,

14b75857-0ffb-4efa-a2f8-90412c742d97-1
00:13:59.840 --> 00:14:04.366
decide with the pressures, you know,
from above to make all those things

14b75857-0ffb-4efa-a2f8-90412c742d97-2
00:14:04.366 --> 00:14:04.800
happen.

5bb094bf-2913-4b91-b3d1-038738ec1367-0
00:14:04.800 --> 00:14:07.813
And from the outside and from the outside
and this partner,

5bb094bf-2913-4b91-b3d1-038738ec1367-1
00:14:07.813 --> 00:14:09.120
you need this AI solution.

730e3027-250b-4f9d-aa65-83954356a03f-0
00:14:09.120 --> 00:14:12.000
I mean, it's, it's and we all feel,
we feel right.

91d70025-ea36-45c7-b0fc-b3a4633c843e-0
00:14:12.160 --> 00:14:14.800
And, you know,
are we are we moving fast enough?

90514c1d-ce20-4f5b-b5b0-476104bf3e5e-0
00:14:14.800 --> 00:14:17.880
Because there's this sort of like
pressure around that.

54fcff87-8d5e-43aa-8111-a5e6fcf852d3-0
00:14:18.120 --> 00:14:23.754
So I think that where you know someone,
someone like KeyMark can be of incredible

54fcff87-8d5e-43aa-8111-a5e6fcf852d3-1
00:14:23.754 --> 00:14:28.613
help to customers is to say, one,
this is what we're seeing in other

54fcff87-8d5e-43aa-8111-a5e6fcf852d3-2
00:14:28.613 --> 00:14:29.600
organizations.

604aed0e-d067-4837-9633-975ba551cb8a-0
00:14:29.600 --> 00:14:35.425
This is how they're approaching and that
sort of best practice two it's helping them

604aed0e-d067-4837-9633-975ba551cb8a-1
00:14:35.425 --> 00:14:37.040
pull the problem apart.

47b10770-a69a-47a4-9923-f62030b910cc-0
00:14:37.320 --> 00:14:39.160
You know,
what are you trying to accomplish?

bd7b2f1b-1f0f-474d-bab6-dc4328abed72-0
00:14:39.160 --> 00:14:41.120
What data are we talking about?

a7418a06-09ea-46fd-9b32-5bd941eced34-0
00:14:41.400 --> 00:14:44.502
What are the capabilities and
possibilities, you know,

a7418a06-09ea-46fd-9b32-5bd941eced34-1
00:14:44.502 --> 00:14:48.960
as it as it relates to not only the
solution, but the governance aspect of it.

d28a0034-e0ca-49c2-b9df-e319f32721e0-0
00:14:50.200 --> 00:14:53.360
It's generally speaking,
not a customer's job to figure that out.

825a8177-40c6-46e3-8719-b4ea60499e7a-0
00:14:53.720 --> 00:14:56.720
And they need trusted sources to to be
able to help.

27a20379-61ae-415a-9105-4ea96e342050-0
00:14:57.240 --> 00:14:57.560
Yeah.

c2efd540-42b7-4dd6-aece-5e2d9f6d03aa-0
00:14:57.720 --> 00:14:58.680
What we’re your Gandalf.

952e2513-4298-4714-ba7c-1cd4699017a4-0
00:14:58.880 --> 00:14:59.200
Yeah.

224eeb21-7d66-4d2d-9fe7-6a33881eb789-0
00:14:59.360 --> 00:15:00.200
Or your Yoda.

144bd1cd-ff0b-4287-a4ec-ca35927d14ac-0
00:15:00.200 --> 00:15:00.480
Yeah.

882da11b-98d4-4f87-93d3-83807e6e5054-0
00:15:00.640 --> 00:15:02.000
Or you've got the beard.

bcbff7cc-0e56-4c8a-b765-f32829b3ed84-0
00:15:02.920 --> 00:15:03.120
Whoa.

62e56f0f-b356-4566-b44d-eeb42a0063f4-0
00:15:03.120 --> 00:15:05.280
Well, well, that the ears too.

b8f0d085-2af4-4d9e-9bf4-19b99fd256cf-0
00:15:05.320 --> 00:15:06.160
The green ears for the.

06de49e4-f4ee-459b-b708-5e68e270e6be-0
00:15:06.160 --> 00:15:08.324
I think I'm more like,
I'm looking more like that Gollum guy at

06de49e4-f4ee-459b-b708-5e68e270e6be-1
00:15:08.324 --> 00:15:09.440
this point, but that's all right.

7ee0865b-646c-4b93-a2ae-cf76e22dc9a9-0
00:15:10.320 --> 00:15:11.800
Well, it's better than the Eye of Sauron,
I guess.

a55b9ce5-3e4d-4082-aaa7-2702af839822-0
00:15:12.760 --> 00:15:13.840
It's fiery face.

3824a73e-0c8a-4599-8b56-be30f4dc3395-0
00:15:13.840 --> 00:15:14.120
Yeah.

723dbca5-b4a2-4564-b27b-d1b84f4e5038-0
00:15:14.960 --> 00:15:18.759
So the,
the World Economic Forum has described

723dbca5-b4a2-4564-b27b-d1b84f4e5038-1
00:15:18.759 --> 00:15:22.640
the essential pillars of, of,
of AI governance.

2a09c28d-7cd8-4219-b299-daded1e4afa7-0
00:15:22.640 --> 00:15:25.240
And I think it's,
it's really helpful lens to look at.

77abcab7-f4a3-4d19-8df0-cb742e15b65b-0
00:15:25.240 --> 00:15:26.240
It's really three of them.

154e306a-0db1-47e5-b792-5e332626f02e-0
00:15:26.240 --> 00:15:31.520
There's a responsible AI,
ethical AI and trustworthy AI.

d95550e4-3427-4211-8361-321df7b4139c-0
00:15:31.520 --> 00:15:39.301
What in your personal assessment,
what do you think is the most

d95550e4-3427-4211-8361-321df7b4139c-1
00:15:39.301 --> 00:15:43.800
underestimated aspect of those three?

2502c0c4-353e-4491-9a05-b53ff734e8b9-0
00:15:44.920 --> 00:15:48.492
I lean in,
I lean towards trustworthy only because I

2502c0c4-353e-4491-9a05-b53ff734e8b9-1
00:15:48.492 --> 00:15:53.750
think the people just assume to a certain
extent that the information they're

2502c0c4-353e-4491-9a05-b53ff734e8b9-2
00:15:53.750 --> 00:15:58.200
getting back from whatever AI tools
they're using is trustworthy.

30d7c318-5aec-42e7-ac2d-58dcad1a9308-0
00:15:59.200 --> 00:16:02.706
And we talked a little bit on a previous
podcast about if you,

30d7c318-5aec-42e7-ac2d-58dcad1a9308-1
00:16:02.706 --> 00:16:06.768
if you really want to experiment,
go in and ask it something you already

30d7c318-5aec-42e7-ac2d-58dcad1a9308-2
00:16:06.768 --> 00:16:10.720
know the answer to and see how you feel
about what it comes back with.

ca46fb4a-b91d-4540-91de-be63d49dcba3-0
00:16:10.720 --> 00:16:14.599
And you know,
hopefully it's comes back pretty strong,

ca46fb4a-b91d-4540-91de-be63d49dcba3-1
00:16:14.599 --> 00:16:17.280
but you're likely to find differences.

b36cfcb4-3433-4c75-b4ad-bc0cd1ed6204-0
00:16:17.280 --> 00:16:23.805
And that's kind of that leading indicator
that we're we're just assuming that what

b36cfcb4-3433-4c75-b4ad-bc0cd1ed6204-1
00:16:23.805 --> 00:16:26.400
it's spitting out is, is correct.

fa39f5b5-8ecf-44c3-8aca-4c30fbf2aedf-0
00:16:26.520 --> 00:16:31.827
And I think that this is sort of the
whole point of this governance discussion

fa39f5b5-8ecf-44c3-8aca-4c30fbf2aedf-1
00:16:31.827 --> 00:16:33.440
is you, you can't right.

d147d334-7012-4816-8a5a-150df16fb0ff-0
00:16:33.440 --> 00:16:38.042
I don't think, you know,
as you listed off things like ethical,

d147d334-7012-4816-8a5a-150df16fb0ff-1
00:16:38.042 --> 00:16:41.709
I think everybody,
the AI vendors themselves, the,

d147d334-7012-4816-8a5a-150df16fb0ff-2
00:16:41.709 --> 00:16:46.600
the companies using it has a desire to be
ethical at at some level.

0986e399-539d-4a7d-a247-18cb76ea2396-0
00:16:46.600 --> 00:16:52.025
I think I, I like to think, however,
things like bias, you know, you, you, you,

0986e399-539d-4a7d-a247-18cb76ea2396-1
00:16:52.025 --> 00:16:55.415
you talk,
you think about a word like ethical and

0986e399-539d-4a7d-a247-18cb76ea2396-2
00:16:55.415 --> 00:16:56.840
you think about bias.

2c058108-740d-4989-a94f-c9b2ef420338-0
00:16:56.840 --> 00:16:59.880
There is bias built in not only to AI.

e423af7c-ec83-47c4-937e-1a8ff08e3d3d-0
00:16:59.880 --> 00:17:04.040
There's bias built into each
organization's own data.

264060ef-4ab7-4520-aaad-deb698a98dc7-0
00:17:04.520 --> 00:17:07.533
How they make decisions,
how they write contracts,

264060ef-4ab7-4520-aaad-deb698a98dc7-1
00:17:07.533 --> 00:17:09.720
how they adjudicate claims, how they.

be45650e-ae78-4eb8-b4e2-c7e85f026fdc-0
00:17:09.760 --> 00:17:12.920
All those kinds of things are inherent
bias.

98d9ce15-729f-44ac-9a38-9536bed4af63-0
00:17:12.960 --> 00:17:18.143
If you simply take that data and feed it
into AI,

98d9ce15-729f-44ac-9a38-9536bed4af63-1
00:17:18.143 --> 00:17:25.400
you've now introduced your bias into AI
to be carried forward to now.

02007df5-c426-4024-94c7-c423f42c1ecd-0
00:17:25.640 --> 00:17:30.720
Maybe that's your intent,
but that should be a decision, right?

ce3e4b28-0d2e-43ce-b0bb-f3c7394f6fc6-0
00:17:30.840 --> 00:17:34.741
Our data committee, our,
our governance committee looks and says,

ce3e4b28-0d2e-43ce-b0bb-f3c7394f6fc6-1
00:17:34.741 --> 00:17:37.520
do do we want that or is there a,
a scrubbing?

e1a9fc48-7644-4c53-8909-21520c47b3c9-0
00:17:37.520 --> 00:17:42.510
Is there some sort of a semantic layer
that gets built that says we don't

e1a9fc48-7644-4c53-8909-21520c47b3c9-1
00:17:42.510 --> 00:17:45.814
necessarily want to carry that forward
that way,

e1a9fc48-7644-4c53-8909-21520c47b3c9-2
00:17:45.814 --> 00:17:48.040
we want to make it more agnostic.

6c604fd4-0946-4a30-9f78-0422c3b9c69a-0
00:17:48.200 --> 00:17:54.497
So we talked about just what it is,
but let's continue talking about why it

6c604fd4-0946-4a30-9f78-0422c3b9c69a-1
00:17:54.497 --> 00:17:55.160
matters.

c450e821-3308-4ca3-9c1b-19799aa53b0f-0
00:17:55.960 --> 00:18:00.840
I think there's this regulatory landscape
that we need to discuss.

e732a05d-6826-4fad-8a87-5cb7b0610293-0
00:18:00.840 --> 00:18:07.206
We are moving toward a world where AI
governance isn't just optional,

e732a05d-6826-4fad-8a87-5cb7b0610293-1
00:18:07.206 --> 00:18:10.480
it's it's regulated, it's mandatory.

35548016-febb-4279-8d79-17e9837973b6-0
00:18:10.480 --> 00:18:16.000
And we already have GDPR in place that's
been around for years now.

d2e39984-d79b-4602-bb95-ccc0a417553d-0
00:18:16.560 --> 00:18:23.327
But I don't know that everybody's aware
of the new act and EU, EUAI act that's a

d2e39984-d79b-4602-bb95-ccc0a417553d-1
00:18:23.327 --> 00:18:30.095
lot of lot of letters in that one that
that has some real specific concerns if

d2e39984-d79b-4602-bb95-ccc0a417553d-2
00:18:30.095 --> 00:18:36.948
you're not paying attention to it where
you you have to have more disclosure of

d2e39984-d79b-4602-bb95-ccc0a417553d-3
00:18:36.948 --> 00:18:43.202
use of AI or how are you using it
especially for public facing solutions

d2e39984-d79b-4602-bb95-ccc0a417553d-4
00:18:43.202 --> 00:18:46.800
and then around the specifics of industry.

87c013c5-9aee-4837-94af-f089be90f4b5-0
00:18:46.800 --> 00:18:51.404
You mentioned that earlier this finance,
healthcare, so again you had this,

87c013c5-9aee-4837-94af-f089be90f4b5-1
00:18:51.404 --> 00:18:55.887
it will we kind of talk about this,
this internal piece and this external

87c013c5-9aee-4837-94af-f089be90f4b5-2
00:18:55.887 --> 00:18:59.280
piece and we'll we'll reference that in
different ways.

a7938937-0f02-4c30-b01d-d24a6e809d11-0
00:18:59.280 --> 00:19:04.254
Sometimes you don't want an internal
person getting access to another piece of

a7938937-0f02-4c30-b01d-d24a6e809d11-1
00:19:04.254 --> 00:19:08.284
information internally,
but then there's the external access to

a7938937-0f02-4c30-b01d-d24a6e809d11-2
00:19:08.284 --> 00:19:12.000
your organization that you also need to
to buffer as well.

506a1b87-48b6-41e9-9d7a-57f88227c9d4-0
00:19:12.360 --> 00:19:19.760
I am concerned that not enough US based
enterprises are paying attention to this.

26e5e414-f43b-4fa6-a190-51135fcd1774-0
00:19:19.760 --> 00:19:24.870
I don't know that they're aware of what's
changing and the actually the

26e5e414-f43b-4fa6-a190-51135fcd1774-1
00:19:24.870 --> 00:19:27.000
responsibility that they have.

c25ccea8-48e0-4b44-837c-058337a21891-0
00:19:27.960 --> 00:19:30.877
No, I,
I agree with you and I think that

c25ccea8-48e0-4b44-837c-058337a21891-1
00:19:30.877 --> 00:19:34.079
traditionally,
without getting on a soapbox,

c25ccea8-48e0-4b44-837c-058337a21891-2
00:19:34.079 --> 00:19:38.562
Europe has been well ahead of the US as
it relates to privacy,

c25ccea8-48e0-4b44-837c-058337a21891-3
00:19:38.562 --> 00:19:41.480
personal information, that sort of thing.

0e842c37-71d6-49eb-89cf-782af4e7faa9-0
00:19:41.640 --> 00:19:45.280
For for better or worse, GDPR is,
is been well out in front.

90cc32c5-4485-4b8e-99d3-90b84f5b7f1f-0
00:19:46.000 --> 00:19:50.290
What is also true though,
is that there are states in the US that

90cc32c5-4485-4b8e-99d3-90b84f5b7f1f-1
00:19:50.290 --> 00:19:54.320
have been traditional fast followers on
some of those things.

511e8552-8a43-4b06-a77a-4ba203af7378-0
00:19:54.320 --> 00:19:58.847
So you've seen privacy acts that have
popped up in various States and the

511e8552-8a43-4b06-a77a-4ba203af7378-1
00:19:58.847 --> 00:20:03.129
United States and, and with that,
any company in KeyMark being a good

511e8552-8a43-4b06-a77a-4ba203af7378-2
00:20:03.129 --> 00:20:07.839
example, we operate all across the,
the US, right and, and sometimes out of.

6f1b39d5-cc57-4a14-87c4-66cd79fa2bbd-0
00:20:08.640 --> 00:20:10.360
That's going to apply for us too.

914e556c-ce1a-436c-b321-b6b42a5d7c36-0
00:20:10.360 --> 00:20:12.240
We do business in some of those states.

54ee2b9e-96f4-4919-aa2c-923d0d6adf5c-0
00:20:12.280 --> 00:20:17.469
And and so there's the optionality is
going to go away fairly quickly into your

54ee2b9e-96f4-4919-aa2c-923d0d6adf5c-1
00:20:17.469 --> 00:20:20.000
point that AI act has some teeth to it.

247f36cb-7cce-43c3-a2ad-802db3f54155-0
00:20:20.080 --> 00:20:26.619
And so organizations are now going to
have to not only think in terms of things

247f36cb-7cce-43c3-a2ad-802db3f54155-1
00:20:26.619 --> 00:20:31.033
like, to your point,
trustworthiness and ethical and,

247f36cb-7cce-43c3-a2ad-802db3f54155-2
00:20:31.033 --> 00:20:36.592
and that sort of thing, but also risk,
right, like government risk,

247f36cb-7cce-43c3-a2ad-802db3f54155-3
00:20:36.592 --> 00:20:42.804
legislative risk, litigious risk,
not just the risk of I gave bad data to a

247f36cb-7cce-43c3-a2ad-802db3f54155-4
00:20:42.804 --> 00:20:49.344
customer or which is not good either,
but doesn't necessarily or hasn't had the

247f36cb-7cce-43c3-a2ad-802db3f54155-5
00:20:49.344 --> 00:20:51.960
teeth that you're talking about.

2dd127a1-4647-4979-9fca-eaebdae3f6b3-0
00:20:52.000 --> 00:20:54.774
And so, you know,
one of the things I think maybe to, to,

2dd127a1-4647-4979-9fca-eaebdae3f6b3-1
00:20:54.774 --> 00:20:56.880
to turn it around on you just for a
second.

aea6faa1-5c8b-42d0-949a-0d6c14654771-0
00:20:56.880 --> 00:21:01.521
We, we spoke even earlier today, you know,
from a marketing standpoint,

aea6faa1-5c8b-42d0-949a-0d6c14654771-1
00:21:01.521 --> 00:21:05.776
you think about our website,
you've got decisions to make around,

aea6faa1-5c8b-42d0-949a-0d6c14654771-2
00:21:05.776 --> 00:21:09.128
you know,
sort of ethical and trustworthy and those

aea6faa1-5c8b-42d0-949a-0d6c14654771-3
00:21:09.128 --> 00:21:10.160
sorts of things.

dcf8018a-a3a0-4ea7-875a-43097505e491-0
00:21:10.640 --> 00:21:16.389
And we talked earlier about how do you,
how do you want to represent your

dcf8018a-a3a0-4ea7-875a-43097505e491-1
00:21:16.389 --> 00:21:21.440
organization as you think about labeling
things as AI generated?

4f2da371-bc49-48dc-984b-dbeff2a027be-0
00:21:21.480 --> 00:21:27.304
Do we create little bobbing heads on
there that are AI don't know, you know,

4f2da371-bc49-48dc-984b-dbeff2a027be-1
00:21:27.304 --> 00:21:32.524
AI created like how you're thinking about
it, that it comes down to,

4f2da371-bc49-48dc-984b-dbeff2a027be-2
00:21:32.524 --> 00:21:36.080
I think knowing who you are as a brand,
right?

696fa6a1-d7e2-456a-becf-cbd70d6ee9ae-0
00:21:36.480 --> 00:21:40.980
And without getting on too much marketing
speak,

696fa6a1-d7e2-456a-becf-cbd70d6ee9ae-1
00:21:40.980 --> 00:21:47.960
the responsibility that comes from that
EUAI act is, I think, a good thing.

db170e9f-d5e4-4a04-a408-05afd982c0fc-0
00:21:47.960 --> 00:21:54.037
And the fact that so many people look at
whether they see it on social media or

db170e9f-d5e4-4a04-a408-05afd982c0fc-1
00:21:54.037 --> 00:22:00.040
wherever they interact with content,
even my kids like that, that's AI, right?

f7b83f4d-e643-44d7-91b4-9b4043255938-0
00:22:00.280 --> 00:22:02.320
And they're like, no, it's not,
it's not AI.

36669658-9353-48b3-82ed-96b582f7113a-0
00:22:02.400 --> 00:22:03.360
No, no, that's AI.

fd3b46b0-eb1a-446f-9a34-be7e80fb69a5-0
00:22:03.360 --> 00:22:07.360
And then this is argument which is is
really breaking trust down.

9369a9c9-f19a-493c-a773-23296c3985a9-0
00:22:07.720 --> 00:22:13.174
So apply that not just globally to
whether you need to watermark your images

9369a9c9-f19a-493c-a773-23296c3985a9-1
00:22:13.174 --> 00:22:16.503
or correct,
add to your privacy policy that we

9369a9c9-f19a-493c-a773-23296c3985a9-2
00:22:16.503 --> 00:22:21.320
occasionally use AI generated content,
whatever that may look like.

a4eb2ad1-aed7-4045-8bb6-7e308dc34217-0
00:22:22.120 --> 00:22:25.076
Think,
think backwards into an enterprise

a4eb2ad1-aed7-4045-8bb6-7e308dc34217-1
00:22:25.076 --> 00:22:29.581
mindset, because that's the, you know, we,
we bring our biases,

a4eb2ad1-aed7-4045-8bb6-7e308dc34217-2
00:22:29.581 --> 00:22:32.960
we bring our behaviors into the office
with us.

f1768b57-899f-4c93-8786-dc0de97b6a56-0
00:22:33.200 --> 00:22:37.088
And so all right,
if I'm going to query in an agentic

f1768b57-899f-4c93-8786-dc0de97b6a56-1
00:22:37.088 --> 00:22:41.768
solution right on top of my, my data,
am I going to go, come on,

f1768b57-899f-4c93-8786-dc0de97b6a56-2
00:22:41.768 --> 00:22:45.728
It said this right, which is trust, right,
Which it's,

f1768b57-899f-4c93-8786-dc0de97b6a56-3
00:22:45.728 --> 00:22:50.120
it's interesting how there's been this
complete trust in it.

c94cfeca-34ae-40e3-b625-46ee97ceed01-0
00:22:50.120 --> 00:22:53.664
And there are times when, well,
I think I even find myself doing that,

c94cfeca-34ae-40e3-b625-46ee97ceed01-1
00:22:53.664 --> 00:22:56.160
like, all right, well,
what it says is that's it.

28b39fe5-4e81-4d72-b01f-f1dc8399475c-0
00:22:56.160 --> 00:22:56.680
That's good.

5eed7685-9967-43dc-bc6c-5172bda06c87-0
00:22:56.680 --> 00:22:57.280
That's great.

0149347e-9e7a-4698-9969-43347598211c-0
00:22:57.800 --> 00:23:03.671
And then later on, I'm like, I don't know,
it's an odd thing that we're all kind of

0149347e-9e7a-4698-9969-43347598211c-1
00:23:03.671 --> 00:23:04.720
wrestling with.

2d47fdee-63a0-4a4b-a549-14f523d92cbe-0
00:23:04.720 --> 00:23:10.523
And I think that that's so important to
consider as we move into, you know,

2d47fdee-63a0-4a4b-a549-14f523d92cbe-1
00:23:10.523 --> 00:23:14.800
the drivers of what is what matters in
your governance.

644e6edd-796b-4919-b20a-e3196bcc7093-0
00:23:14.800 --> 00:23:20.720
And so I think the companies that are
taking this seriously and that operate,

644e6edd-796b-4919-b20a-e3196bcc7093-1
00:23:20.720 --> 00:23:26.641
operationalize governance now are going
to move a lot faster and are going to

644e6edd-796b-4919-b20a-e3196bcc7093-2
00:23:26.641 --> 00:23:28.160
scale a lot quicker.

33c60500-1ff9-4e99-a94c-bcfbcbaf2ac7-0
00:23:29.600 --> 00:23:33.595
And, and I think, you know,
that we can look at it as friction,

33c60500-1ff9-4e99-a94c-bcfbcbaf2ac7-1
00:23:33.595 --> 00:23:37.840
which it can be in a way,
or we can look at it as a growth enabler.

ab0c750b-885a-45d2-a1b4-962ff0c8dcb3-0
00:23:38.360 --> 00:23:39.840
No, I, I, I totally agree.

24da72ad-41ee-46e1-bae5-87384316efa0-0
00:23:39.840 --> 00:23:44.421
And, and I think it,
it comes back to this idea of you have to

24da72ad-41ee-46e1-bae5-87384316efa0-1
00:23:44.421 --> 00:23:50.021
have a centralized decisioning function
on how you're going to address these

24da72ad-41ee-46e1-bae5-87384316efa0-2
00:23:50.021 --> 00:23:51.040
issues, right?

22ec6088-a8eb-49f2-8951-7ead2ea8fc9e-0
00:23:51.400 --> 00:23:54.639
If you in your role said, hey,
we're going to,

22ec6088-a8eb-49f2-8951-7ead2ea8fc9e-1
00:23:54.639 --> 00:23:59.807
we're going to create a bunch of AI
characters that are going to be videos

22ec6088-a8eb-49f2-8951-7ead2ea8fc9e-2
00:23:59.807 --> 00:24:02.840
and they're going to tell our story for
us.

ec790def-f7b9-4b1c-8ee3-c81759ef3095-0
00:24:02.840 --> 00:24:06.880
And we're going to use AI to generate
everything they say, etc.

cc44c30a-6652-4f24-9590-961e3bb7a623-0
00:24:07.760 --> 00:24:11.000
That's not necessarily on the surface
right or wrong, right?

9f251ce4-aecb-48f2-a2d6-97f3a0dcaa27-0
00:24:11.880 --> 00:24:14.320
It's one to your point, what's your brand?

8c60c0c1-6848-49f4-8d9d-a14becea91ef-0
00:24:14.320 --> 00:24:15.880
What's your stance?

1b62cd33-c573-4f32-af15-9ade53ee2deb-0
00:24:16.280 --> 00:24:20.600
2 How important is it that that date is
right?

62e74f5c-fdcc-40a0-8cee-052ad15d3109-0
00:24:20.920 --> 00:24:22.200
It's of the utmost importance.

900cad3f-3cce-4722-86b4-8d9fb4204fdf-0
00:24:22.200 --> 00:24:25.480
This this is representing you and who you
are.

b2a368f2-a24e-4c79-9a94-624850985418-0
00:24:25.960 --> 00:24:30.716
And if you don't have a process in place
by which that's going to get reviewed and

b2a368f2-a24e-4c79-9a94-624850985418-1
00:24:30.716 --> 00:24:33.810
what comes out,
then you should have to live with the

b2a368f2-a24e-4c79-9a94-624850985418-2
00:24:33.810 --> 00:24:36.331
result that comes from that,
the liability,

b2a368f2-a24e-4c79-9a94-624850985418-3
00:24:36.331 --> 00:24:38.280
the the risk that comes from that.

da1fd54e-0bb1-4d7b-b3f9-b029245792e8-0
00:24:38.960 --> 00:24:43.624
And this goes back, you know,
earlier I touched on this,

da1fd54e-0bb1-4d7b-b3f9-b029245792e8-1
00:24:43.624 --> 00:24:47.880
this idea of really taking, you know,
in our world.

87b698c6-9c99-412b-b888-b2ebc0a0b0f4-0
00:24:47.880 --> 00:24:50.000
So coming back to kind of what we do.

4dc1ac0a-7cb2-4f03-b012-84ce28751198-0
00:24:50.240 --> 00:24:57.287
So I've got this goldmine of unstructured
information and let's use, I don't know,

4dc1ac0a-7cb2-4f03-b012-84ce28751198-1
00:24:57.287 --> 00:24:59.240
contracts, for example.

715fcdb3-52ac-41b3-b426-e99d747372c7-0
00:24:59.640 --> 00:25:04.184
And I want to use an IDP tool or I want
to use a, a tool to, you know,

715fcdb3-52ac-41b3-b426-e99d747372c7-1
00:25:04.184 --> 00:25:08.920
go in vectorize that that data and start
to make it available downstream.

1928e1a2-8879-403a-b7df-fc6c3b98f5ee-0
00:25:09.160 --> 00:25:11.200
So what type of document is it?

c00928e2-d330-4bce-83e0-a76dd741b6e0-0
00:25:11.920 --> 00:25:14.600
OK, what if, what if I get that wrong?

828fb47e-40e4-4061-be79-fd94e9760a7f-0
00:25:15.960 --> 00:25:21.490
If I get that wrong and the the rules
that follow and I pass that downstream to

828fb47e-40e4-4061-be79-fd94e9760a7f-1
00:25:21.490 --> 00:25:23.080
an agent, what happens?

9b0ab803-2cb4-4d0a-a029-5651faab4035-0
00:25:23.080 --> 00:25:26.233
Well,
maybe something shows up that really

9b0ab803-2cb4-4d0a-a029-5651faab4035-1
00:25:26.233 --> 00:25:28.360
shouldn't show up by example.

30ba3573-3c4b-4ff7-a890-bce707bc00d8-0
00:25:30.560 --> 00:25:33.320
What tools do we have in place?

bab631be-a15b-45ea-a1b0-bf2ecdd5d9b3-0
00:25:33.320 --> 00:25:36.120
So we talked about where does that
boutique element come in?

9cdd49d3-a3d7-43a5-a215-b0080fdd8f03-0
00:25:36.120 --> 00:25:39.840
One of the things we know from our
history is you better validate.

18c3ba58-b1e6-44e6-ae43-388bfb5fe696-0
00:25:40.080 --> 00:25:41.040
It's pretty important.

ad561ece-22eb-4a29-b41b-ef4ab795ef49-0
00:25:41.440 --> 00:25:44.160
If I go to read a bunch of
information, I should probably validate it.

7c6c8026-d824-4e6e-87e3-255fc638cbf8-0
00:25:44.360 --> 00:25:45.880
I should check the confidence.

ab3d85ec-365a-4afe-ace3-59e1834ecafd-0
00:25:45.880 --> 00:25:50.237
How confident are you about this document,
this word, this paragraph, this,

ab3d85ec-365a-4afe-ace3-59e1834ecafd-1
00:25:50.237 --> 00:25:51.040
what have you?

db5d698a-3a4a-4442-a7ef-bd4982909363-0
00:25:51.600 --> 00:25:55.200
And there's a lot of tools out there that
are being built right now.

30d8a4bf-9dfa-484e-b5f7-29c99b00c5d3-0
00:25:55.200 --> 00:25:58.800
They're calling themselves IDP or they're
calling themselves different things.

b986af62-506a-46a3-9439-6083f52562b4-0
00:25:59.400 --> 00:26:01.520
Well, they don't necessarily do that.

d5d8b189-b674-4d5a-bc76-5d629368eb88-0
00:26:01.520 --> 00:26:03.920
They just read it and say, here it is.

72bdf80a-8be0-492e-8527-680ae8863338-0
00:26:04.360 --> 00:26:08.898
And obviously customers are enamored with
the possibilities that, well,

72bdf80a-8be0-492e-8527-680ae8863338-1
00:26:08.898 --> 00:26:10.600
I can just build it myself.

41c4d7f4-0ef7-4a80-923c-90d4e463aa2c-0
00:26:11.160 --> 00:26:11.960
You can.

24e254bd-77bf-4b8f-ab8c-5b4d0a53f331-0
00:26:12.080 --> 00:26:13.040
You're absolutely right.

1294ebcc-8ce9-4252-a028-992c6f60e514-0
00:26:13.360 --> 00:26:14.520
You can build it yourself.

c20e84f6-2c39-40ee-96d2-fbcf5f1d3022-0
00:26:15.680 --> 00:26:19.630
It won't have the controls,
it won't have the validation steps,

c20e84f6-2c39-40ee-96d2-fbcf5f1d3022-1
00:26:19.630 --> 00:26:23.520
it won't have the audit trail,
it won't have the traceability.

6e253817-de54-44b8-bb8f-ebb0628c7ea3-0
00:26:23.520 --> 00:26:24.720
It won't have these things.

4099edba-d0d7-4b70-8e46-9e05cc3e9672-0
00:26:24.720 --> 00:26:25.600
It'll read the data.

f2996a70-0297-46dd-b02a-f23a598aff51-0
00:26:26.200 --> 00:26:30.048
Maybe it's right, maybe it's wrong,
and it's just going and you know,

f2996a70-0297-46dd-b02a-f23a598aff51-1
00:26:30.048 --> 00:26:34.006
so I think if we talk about governance,
it's like it's the policies and

f2996a70-0297-46dd-b02a-f23a598aff51-2
00:26:34.006 --> 00:26:36.480
procedures,
it's also the technology itself.

059e258b-0f60-43f4-b593-edc029702457-0
00:26:36.840 --> 00:26:40.586
It's the decision making involved in,
you know what,

059e258b-0f60-43f4-b593-edc029702457-1
00:26:40.586 --> 00:26:42.920
what should we be pulling across?

3b104485-3531-41dd-83a5-38dc1022e8a9-0
00:26:42.920 --> 00:26:44.080
What are we making available?

fc54c13f-a992-4784-965f-8024366a7bd1-0
00:26:44.280 --> 00:26:49.137
Because if you unleash the agent and this
is the difference from the human in the

fc54c13f-a992-4784-965f-8024366a7bd1-1
00:26:49.137 --> 00:26:52.632
loop versus the agent,
agents going to do exactly what you

fc54c13f-a992-4784-965f-8024366a7bd1-2
00:26:52.632 --> 00:26:53.640
program it to do.

e820c4a1-6e2d-4acf-9f3b-193ac63aaf4a-0
00:26:54.440 --> 00:26:58.040
Go find this, bring it back as an example,
go make this decision.

9875ee81-cfc4-4b0c-a9ec-3d9a15aecce2-0
00:26:59.360 --> 00:27:02.486
If you'd give an agent decision making
authority,

9875ee81-cfc4-4b0c-a9ec-3d9a15aecce2-1
00:27:02.486 --> 00:27:07.612
which is where people are saying, well,
I don't need the knowledge worker because

9875ee81-cfc4-4b0c-a9ec-3d9a15aecce2-2
00:27:07.612 --> 00:27:12.364
it can decide on behalf of a person,
How much more important is it that the

9875ee81-cfc4-4b0c-a9ec-3d9a15aecce2-3
00:27:12.364 --> 00:27:13.239
data is right?

d4de676f-4ede-4b72-b3ae-e43522a31f11-0
00:27:14.560 --> 00:27:18.409
So these are the things that, you know,
I think you've got to be grappled with,

d4de676f-4ede-4b72-b3ae-e43522a31f11-1
00:27:18.409 --> 00:27:21.200
got to be wrestled with and,
and have to have a plan for.

15428f9d-5610-4f8c-81eb-61805b97f91c-0
00:27:21.200 --> 00:27:24.223
And I think it's the reason, you know,
going back to kind of your, your,

15428f9d-5610-4f8c-81eb-61805b97f91c-1
00:27:24.223 --> 00:27:26.377
your first comments about why should you,
you know,

15428f9d-5610-4f8c-81eb-61805b97f91c-2
00:27:26.377 --> 00:27:28.200
why are we excited about this
conversation?

ccb779e5-2e26-4ed3-bbec-77034b0213a0-0
00:27:28.880 --> 00:27:34.072
Because I think it's really sort of the
next phase of these AI discussions where

ccb779e5-2e26-4ed3-bbec-77034b0213a0-1
00:27:34.072 --> 00:27:39.200
the all the good that's out there with it
is also met with the realities of it.

2a9ac65d-9398-452a-99cc-bceffa83ab37-0
00:27:39.480 --> 00:27:40.240
How do we address that?

07d74694-25b2-420a-abc2-8de51b1a6c40-0
00:27:40.560 --> 00:27:40.880
Yeah.

322a787e-b7d5-45fa-9b7c-ebe9d615b752-0
00:27:40.960 --> 00:27:45.604
And we talked last time I think about the
yeah, the the failures of AI,

322a787e-b7d5-45fa-9b7c-ebe9d615b752-1
00:27:45.604 --> 00:27:49.668
specifically with the,
it's an older use case now with the Air

322a787e-b7d5-45fa-9b7c-ebe9d615b752-2
00:27:49.668 --> 00:27:50.120
Canada.

995431ba-58b4-4c4f-ad9a-cccb36892d92-0
00:27:50.160 --> 00:27:50.880
Air Canada, right?

350e93b6-09ea-48d4-9e06-2775ab324710-0
00:27:50.960 --> 00:27:51.200
Yeah.

fad91997-caf1-45e3-b0e2-2b6b9686923b-0
00:27:51.560 --> 00:27:54.677
I think most AI failures,
they're not model failures,

fad91997-caf1-45e3-b0e2-2b6b9686923b-1
00:27:54.677 --> 00:27:58.372
They're actually governance failures,
100% because they've not,

fad91997-caf1-45e3-b0e2-2b6b9686923b-2
00:27:58.372 --> 00:28:02.240
you're not setting those up up front,
you're not gating it, right.

4356c5a3-0610-47d3-80bc-1e603fb4891b-0
00:28:02.480 --> 00:28:08.664
And I think that's something we want to
drive home, that this really is super,

4356c5a3-0610-47d3-80bc-1e603fb4891b-1
00:28:08.664 --> 00:28:11.640
super important that you get it right.

93c7539e-5590-4460-a261-95152f40038b-0
00:28:11.720 --> 00:28:14.736
Well, you use that example,
and I don't know this, you know,

93c7539e-5590-4460-a261-95152f40038b-1
00:28:14.736 --> 00:28:17.011
I don't know exactly what led to the
problem,

93c7539e-5590-4460-a261-95152f40038b-2
00:28:17.011 --> 00:28:19.880
but let's say they have a policies and
procedures manual.

ed949519-3b82-4eb4-a25d-a582747ac655-0
00:28:20.040 --> 00:28:20.320
Yeah.

0f2c756b-9fc5-4f77-8fca-2638b33f8e45-0
00:28:20.560 --> 00:28:22.000
How often do you think that gets updated?

ea3aa3b8-1a0d-4d0f-bc44-169a3ed5ea63-0
00:28:23.000 --> 00:28:24.080
Multiple times per year.

7d358f98-305c-49d9-8eb2-2ef7544b79cf-0
00:28:24.360 --> 00:28:25.440
How many versions are there?

3c4ded6c-3937-434c-bdb0-a5f5a8247277-0
00:28:25.640 --> 00:28:26.480
100 of them.

5dc20362-f147-41b7-a826-f7bcada3e287-0
00:28:27.040 --> 00:28:28.120
200 of them.

5a99705f-cf92-4e49-a4c4-d13745009214-0
00:28:29.040 --> 00:28:32.615
If you just unleash, you know,
a tool to go crawl and bring the

5a99705f-cf92-4e49-a4c4-d13745009214-1
00:28:32.615 --> 00:28:35.520
information back,
which one's the right one, Right.

0617279e-ef72-4f11-88fc-ffde41424119-0
00:28:36.680 --> 00:28:37.640
How would the agent know?

90efe28b-0472-4baa-a198-41e633afd5f0-0
00:28:39.160 --> 00:28:41.720
So what what governs the process to get
the right version?

aec7d5db-c8e5-4b13-97d6-6c858a0ac60d-0
00:28:43.360 --> 00:28:46.600
It's and and and it's not just, you know,
you're bringing up.

e3d641bf-5089-4be4-8579-bf0c5743ae8b-0
00:28:46.640 --> 00:28:51.120
I love what you're bringing up around
different areas of an organization.

43c3fa43-7e09-4746-b43c-f2f3ce73d4e3-0
00:28:51.400 --> 00:28:53.965
Yeah,
that that may be an HR thing that you

43c3fa43-7e09-4746-b43c-f2f3ce73d4e3-1
00:28:53.965 --> 00:28:54.840
just mentioned.

2780d0be-a94f-467a-a462-d6c9a0a69f6f-0
00:28:54.840 --> 00:29:00.675
It could be that department,
but I think people I'm afraid their

2780d0be-a94f-467a-a462-d6c9a0a69f6f-1
00:29:00.675 --> 00:29:06.601
mindset goes to the one,
this is an IT problem and it's not an IT

2780d0be-a94f-467a-a462-d6c9a0a69f6f-2
00:29:06.601 --> 00:29:07.320
problem.

346cdec0-785e-4382-9d93-02e913095858-0
00:29:07.320 --> 00:29:13.177
There is there's involvement there,
but before you ever train a model,

346cdec0-785e-4382-9d93-02e913095858-1
00:29:13.177 --> 00:29:17.880
there is this governance question of is
your data ready?

626bc219-4b6d-48ba-b26e-1f9c0459e1c3-0
00:29:19.280 --> 00:29:24.293
And you know,
if you go put milk in your car tank,

626bc219-4b6d-48ba-b26e-1f9c0459e1c3-1
00:29:24.293 --> 00:29:27.440
gas tank, it's not going to run.

e8112b8b-a2b5-454c-ae92-d671d9a99608-0
00:29:27.440 --> 00:29:28.480
We we all know that.

26fdfe0a-8bf3-4a8b-a7d1-35287af58d27-0
00:29:29.560 --> 00:29:36.760
But if you put dirty gas in your car
unintentionally stop off.

1fcb4b3a-31a1-4e9b-a1f5-70dde01380e4-0
00:29:36.760 --> 00:29:38.440
And we won't name a town.

ceebd1eb-9934-4b99-85c7-baba84d09ece-0
00:29:38.440 --> 00:29:42.928
We don't want to insult anybody,
But if you just stop at that little one

ceebd1eb-9934-4b99-85c7-baba84d09ece-1
00:29:42.928 --> 00:29:46.680
pump station in the middle of nowhere,
you're taking a risk.

85a51c83-5d60-42fa-97ef-e47f8e9ffcf8-0
00:29:47.400 --> 00:29:49.120
Your engine’s going to run accordingly.

fdcd6847-5478-41bc-84d2-d06805fba9c2-0
00:29:49.560 --> 00:29:51.880
You don't put diesel in a gas running car.

b7320dca-7c27-44c8-ab10-17d77bf1e53f-0
00:29:52.200 --> 00:29:56.220
You don't put milk in there like,
and so it's this analogy I'm trying to

b7320dca-7c27-44c8-ab10-17d77bf1e53f-1
00:29:56.220 --> 00:29:59.800
paint of what you put into your engine is
how it's going to run.

d54795d7-7e92-44ad-b651-ae4519e1a998-0
00:29:59.960 --> 00:30:04.202
And we come back to that over and over
and over again about what are you putting

d54795d7-7e92-44ad-b651-ae4519e1a998-1
00:30:04.202 --> 00:30:04.360
in?

ae11a7b0-d9dd-4cae-ae47-df10ca9c561a-0
00:30:04.360 --> 00:30:06.720
What are you governing is going into your
tank?

c78131ac-e1bb-4c79-8d31-5534b802aca5-0
00:30:07.080 --> 00:30:08.720
What are you deciding upfront?

0e047f16-e897-4e94-9e3a-5e5a178f90f6-0
00:30:08.720 --> 00:30:14.311
Because what you put in, what you get out,
then the work that you put in upfront on

0e047f16-e897-4e94-9e3a-5e5a178f90f6-1
00:30:14.311 --> 00:30:17.440
governance,
it pays dividends in the long run.

396022dc-2194-4112-ac54-bea35903dadf-0
00:30:17.680 --> 00:30:21.120
Yeah, I think your,
your analogy is exactly right.

3028afc4-d08a-43e3-9d77-543ba46db381-0
00:30:21.120 --> 00:30:24.137
I,
I think that the there is almost this

3028afc4-d08a-43e3-9d77-543ba46db381-1
00:30:24.137 --> 00:30:28.773
assumption or this thought process like,
well, AI is so smart,

3028afc4-d08a-43e3-9d77-543ba46db381-2
00:30:28.773 --> 00:30:30.759
it just figures things out.

69b0b331-8fb2-4227-9a77-a8b23adeea63-0
00:30:30.760 --> 00:30:35.783
Yeah, my AI, yeah, there's, there's,
there's never been a more garbage in

69b0b331-8fb2-4227-9a77-a8b23adeea63-1
00:30:35.783 --> 00:30:40.467
garbage out technology more than just the
old BI, you know, not old,

69b0b331-8fb2-4227-9a77-a8b23adeea63-2
00:30:40.467 --> 00:30:42.640
but the that BI thought process.

0df57532-2ab0-4fe7-99ac-196fec4820f0-0
00:30:42.640 --> 00:30:48.118
I mean, this is now,
unfortunately, what you're doing is you're

0df57532-2ab0-4fe7-99ac-196fec4820f0-1
00:30:48.118 --> 00:30:53.858
saying if garbage goes in,
you've ratcheted up the power level to

0df57532-2ab0-4fe7-99ac-196fec4820f0-2
00:30:53.858 --> 00:31:01.076
extrapolate and use that garbage in much
more meaningful and risk driven ways than

0df57532-2ab0-4fe7-99ac-196fec4820f0-3
00:31:01.076 --> 00:31:02.120
ever before.

d4f0da4f-eae2-4376-bc66-fad8e7a95879-0
00:31:02.120 --> 00:31:05.560
And you know, that's,
that's kind of the piece that's missing.

8eee5aed-87fc-4383-9ebe-cf2fb0a97e72-0
00:31:05.560 --> 00:31:08.920
You know, this,
this idea of what does AI ready mean?

71c69e32-6eae-471e-a560-c48f438cc9fa-0
00:31:08.960 --> 00:31:13.691
I would encourage every organization to
ask themselves that question as it relates

71c69e32-6eae-471e-a560-c48f438cc9fa-1
00:31:13.691 --> 00:31:14.520
to their data.

536eb16d-9ef1-43ad-b2c0-774c2532a903-0
00:31:14.640 --> 00:31:16.560
What does AI ready mean?

d0504087-91ca-4dad-a688-e5e91dbe2496-0
00:31:17.800 --> 00:31:24.146
Go back to because I think IBM has this
agent OPS concept and I really like what

d0504087-91ca-4dad-a688-e5e91dbe2496-1
00:31:24.146 --> 00:31:27.280
you were saying a minute ago about that.

320e2bb9-0980-47ad-bdfc-2409cb1d6667-0
00:31:27.280 --> 00:31:31.721
The agents,
because they are going to be making more

320e2bb9-0980-47ad-bdfc-2409cb1d6667-1
00:31:31.721 --> 00:31:38.006
autonomous decisions and governance
becomes insanely important when you're

320e2bb9-0980-47ad-bdfc-2409cb1d6667-2
00:31:38.006 --> 00:31:40.520
letting AI make that decision.

1398fff0-80ee-4174-bdbc-597e37103213-0
00:31:40.520 --> 00:31:44.948
Because like if an agent makes 42
decisions,

1398fff0-80ee-4174-bdbc-597e37103213-1
00:31:44.948 --> 00:31:51.640
traceability wants to know that you can
track all 42 of those back.

235b8c28-20f2-49e3-a462-0d911363ea9d-0
00:31:51.760 --> 00:31:53.080
Yes, right.

dc5c8b59-b545-4e71-a9b8-e8cabc1a39d4-0
00:31:53.160 --> 00:31:53.560
Yes.

9defb13f-077d-4b19-b4f7-878c52e199f7-0
00:31:53.640 --> 00:31:57.836
And even outside of an agentic element,
if you do that,

9defb13f-077d-4b19-b4f7-878c52e199f7-1
00:31:57.836 --> 00:32:03.980
like even if you're querying something,
doing a current query on a large language

9defb13f-077d-4b19-b4f7-878c52e199f7-2
00:32:03.980 --> 00:32:07.501
model,
if you're not checking your references,

9defb13f-077d-4b19-b4f7-878c52e199f7-3
00:32:07.501 --> 00:32:09.600
you're over trusting the AI.

7aff1efb-19f7-4e0c-9fa1-14c1c93c78e4-0
00:32:10.440 --> 00:32:14.313
And you can go to just about whether it's
a master class or an online AI training

7aff1efb-19f7-4e0c-9fa1-14c1c93c78e4-1
00:32:14.313 --> 00:32:14.880
or whatever.

d33db1ed-b7b3-4a5c-b472-89af67eca4b1-0
00:32:14.880 --> 00:32:18.492
And almost all of them are going to
advise you to do that,

d33db1ed-b7b3-4a5c-b472-89af67eca4b1-1
00:32:18.492 --> 00:32:19.840
check your references.

29f004be-1edd-461d-bf70-ae1ab95f8ade-0
00:32:20.040 --> 00:32:21.680
Same thing as table stakes.

d927ac59-fe35-4786-a9ac-03e6ebdcec7a-0
00:32:21.680 --> 00:32:25.241
I think with especially on that
traceability in a governance side in an

d927ac59-fe35-4786-a9ac-03e6ebdcec7a-1
00:32:25.241 --> 00:32:27.814
organization that says, all right,
if this was the,

d927ac59-fe35-4786-a9ac-03e6ebdcec7a-2
00:32:27.814 --> 00:32:31.919
if you're saying that this customer owed
this much on invoice, for example, right?

ae3dae41-46c0-49bc-8cc0-51c410285eb7-0
00:32:32.200 --> 00:32:34.480
I want to be able to see that invoice if
I need to.

398f3724-ba47-43c3-b660-70edf4786f51-0
00:32:34.760 --> 00:32:35.120
That's right.

f8478942-5597-488b-a4a6-163428ce2957-0
00:32:35.720 --> 00:32:36.040
That's right.

e91a522e-4274-40b5-b084-81eca4b479f2-0
00:32:36.400 --> 00:32:36.640
Yeah.

faa843e9-405e-4b46-ae2a-6e02a7ffe1e0-0
00:32:36.680 --> 00:32:41.992
I, I think that the, you know,
this whole idea of the power that that

faa843e9-405e-4b46-ae2a-6e02a7ffe1e0-1
00:32:41.992 --> 00:32:44.800
agent has is so critically important.

c4e4f335-fd10-42cc-8d19-11a5d8315b98-0
00:32:44.920 --> 00:32:48.150
I mean,
we have never in the history that I know

c4e4f335-fd10-42cc-8d19-11a5d8315b98-1
00:32:48.150 --> 00:32:52.501
of putting in, you know,
workflow or process automation solutions

c4e4f335-fd10-42cc-8d19-11a5d8315b98-2
00:32:52.501 --> 00:32:55.600
not been asked the question in the demo
stage.

8452c235-a190-4540-a8be-ee81936d0550-0
00:32:55.760 --> 00:32:56.720
Is there an audit trail?

d9332f44-a700-46f7-bd92-5c8a181e1105-0
00:32:57.600 --> 00:32:59.240
Can I track who approved what?

d1b14a0f-0d66-4e69-96b0-5c51da3d7c4b-0
00:32:59.360 --> 00:33:02.489
Can I go all the way back and say,
you know, all right,

d1b14a0f-0d66-4e69-96b0-5c51da3d7c4b-1
00:33:02.489 --> 00:33:06.960
this went to Clay and and he approved it
and he shouldn't have whatever, right?

d7eb53a7-914e-4dbf-b353-62c395340276-0
00:33:06.960 --> 00:33:07.880
I have to have that.

63535387-c128-41cd-99b4-72e5ae92a595-0
00:33:08.320 --> 00:33:13.288
So the idea of unleashing an agent and
saying have it make these decisions

63535387-c128-41cd-99b4-72e5ae92a595-1
00:33:13.288 --> 00:33:14.680
without traceability.

27a9d4bb-b8d6-4a18-90da-c0e38fbd87de-0
00:33:14.720 --> 00:33:17.380
I mean,
that's just flying without a net and

27a9d4bb-b8d6-4a18-90da-c0e38fbd87de-1
00:33:17.380 --> 00:33:20.040
that's, that's putting it, you know,
nicely.

ba39ff34-6802-4282-9e07-0091398623d6-0
00:33:21.280 --> 00:33:23.040
What I think is, is really interesting.

76a137f2-4145-4a0e-b78d-1a95a29053af-0
00:33:23.040 --> 00:33:27.086
I think that the question that's being
asked of the industry right now and,

76a137f2-4145-4a0e-b78d-1a95a29053af-1
00:33:27.086 --> 00:33:31.240
and I don't think everybody understands
the ramifications of the answers yet.

7458e37c-a3f9-4824-9595-5c22450bb91c-0
00:33:32.720 --> 00:33:37.541
Couple of really headlining announcements
lately of companies that have done these

7458e37c-a3f9-4824-9595-5c22450bb91c-1
00:33:37.541 --> 00:33:41.840
significant layoffs related to, you know,
what we're going to use agents.

a452a032-4d81-40ec-8bfe-b41b79f24462-0
00:33:41.840 --> 00:33:46.985
I think H&R Block, you know,
just did a, a, a big announcement where,

a452a032-4d81-40ec-8bfe-b41b79f24462-1
00:33:46.985 --> 00:33:50.114
you know,
they're cutting a significant number

a452a032-4d81-40ec-8bfe-b41b79f24462-2
00:33:50.114 --> 00:33:52.200
because the AI can just do it.

00e5c5b9-30a2-4cf2-8466-57d77ebad3c8-0
00:33:52.440 --> 00:33:56.479
I just want to underscore yet again,
because I think sometimes we, we, we,

00e5c5b9-30a2-4cf2-8466-57d77ebad3c8-1
00:33:56.479 --> 00:33:59.320
we can sound like we're not on the AI
train.

93acc062-3494-4fd1-a140-6bc333fe2d1d-0
00:33:59.320 --> 00:34:00.920
We're 100% on the AI train.

2bb3b950-7b28-47f5-ad96-6d7e08ec00bb-0
00:34:00.920 --> 00:34:02.440
I couldn't be more on the AI train.

ca8805e1-4c47-4e3a-ad56-9667a0664780-0
00:34:02.440 --> 00:34:04.200
We can't move fast enough with it.

e3cc1fdb-13ba-46e1-b2e5-d665022c2065-0
00:34:04.200 --> 00:34:07.200
However, it's got to be governed.

dfe6bf13-63a9-4965-becf-b5fd4fc7d6d1-0
00:34:07.440 --> 00:34:10.665
And I think what's going to be
interesting over the next year to two

dfe6bf13-63a9-4965-becf-b5fd4fc7d6d1-1
00:34:10.665 --> 00:34:14.123
years is going to be those companies that
have said, you know, all right,

dfe6bf13-63a9-4965-becf-b5fd4fc7d6d1-2
00:34:14.123 --> 00:34:16.040
throwing out the baby with the bathwater.

c28387c2-1407-40b2-9200-d9935f1f8f31-0
00:34:16.040 --> 00:34:16.600
Here we go.

ef8315b8-3a51-4fb3-9761-0f44b01b6b57-0
00:34:16.640 --> 00:34:18.520
It's all agents all the time.

196edaf4-c85e-4059-8f42-6de474a1ca50-0
00:34:19.440 --> 00:34:24.951
I, it,
it strikes me as difficult to believe

196edaf4-c85e-4059-8f42-6de474a1ca50-1
00:34:24.951 --> 00:34:31.320
they have done the work on the data
readiness side.

8b439627-f555-48bf-9002-2c3226162b74-0
00:34:31.640 --> 00:34:33.640
And now I'm not taking anything away from
their people.

50f056bb-569a-4224-9a9a-68069cce7902-0
00:34:33.640 --> 00:34:35.080
It just feels irresponsible.

8cab9446-b6c8-4d09-88ad-0956e8aaed44-0
00:34:35.080 --> 00:34:40.732
It's just a how would you have had the
time at a company of that size to know

8cab9446-b6c8-4d09-88ad-0956e8aaed44-1
00:34:40.732 --> 00:34:45.515
your data's good And you know,
there's that, you know, that what,

8cab9446-b6c8-4d09-88ad-0956e8aaed44-2
00:34:45.515 --> 00:34:49.718
what I think we're we're trying to drive
to is this is a,

8cab9446-b6c8-4d09-88ad-0956e8aaed44-3
00:34:49.718 --> 00:34:52.400
a slow down to go fast moment, right?

c6a5baee-662e-4de2-b5f0-f9754a3b77b5-0
00:34:52.960 --> 00:34:55.849
Take the time to make sure you're you're
ready,

c6a5baee-662e-4de2-b5f0-f9754a3b77b5-1
00:34:55.849 --> 00:34:59.160
get some help and what are the right
questions to ask?

7c53c896-6a85-412b-ae3f-2076f6cd43da-0
00:34:59.720 --> 00:35:01.960
And that's going to allow you to unleash
the full power.

d2a363a5-8353-44f3-ab05-6c50152610a1-0
00:35:02.920 --> 00:35:07.240
So let's wrap a bow on this with just a
few more closing thoughts.

0f71cb5d-0de9-4f8d-be0b-995220e0f4a5-0
00:35:08.200 --> 00:35:14.401
I was thinking the other day, you know,
as we were preparing for this podcast,

0f71cb5d-0de9-4f8d-be0b-995220e0f4a5-1
00:35:14.401 --> 00:35:16.600
what's the future look like?

eeb91608-c2a5-49d9-96db-026c793f5baf-0
00:35:18.120 --> 00:35:27.773
And I think are we heading into a point
where hackers were going to start using

eeb91608-c2a5-49d9-96db-026c793f5baf-1
00:35:27.773 --> 00:35:35.979
agents to start uncovering
vulnerabilities within organizations who

eeb91608-c2a5-49d9-96db-026c793f5baf-2
00:35:35.979 --> 00:35:39.720
have public facing information?

d3f920e2-413b-48ff-b190-ab452929975f-0
00:35:40.240 --> 00:35:44.792
Not to be a fear mongerer, no,
but that's kind of the point I want to

d3f920e2-413b-48ff-b190-ab452929975f-1
00:35:44.792 --> 00:35:49.280
get across is I think in five years or
less, back to those AI years,

d3f920e2-413b-48ff-b190-ab452929975f-2
00:35:49.280 --> 00:35:54.418
whatever that looks like, very soon,
I think AI governance is going to take on

d3f920e2-413b-48ff-b190-ab452929975f-3
00:35:54.418 --> 00:35:57.280
essentially the same role as
cybersecurity.

bf74da5d-9835-4fa5-aaf7-ba8de3855739-0
00:35:57.280 --> 00:36:03.173
It will be so incredibly important to
protect yourself from outside forces or

bf74da5d-9835-4fa5-aaf7-ba8de3855739-1
00:36:03.173 --> 00:36:07.480
inside your own 4 walls,
if you want to put it that way.

4838f351-a81a-4d6b-98e6-e77234c01605-0
00:36:09.000 --> 00:36:14.960
And really to emphasize this aspect,
that of this thought that AI strategy

4838f351-a81a-4d6b-98e6-e77234c01605-1
00:36:14.960 --> 00:36:20.285
without governance is just
experimentation and it's just downright

4838f351-a81a-4d6b-98e6-e77234c01605-2
00:36:20.285 --> 00:36:21.079
dangerous.

c2ee4272-c9a9-4882-a4d4-451cea2cb38d-0
00:36:22.160 --> 00:36:26.827
Adding in governance to your AI strategy
is what's going to bring about

c2ee4272-c9a9-4882-a4d4-451cea2cb38d-1
00:36:26.827 --> 00:36:27.800
transformation.

c9075b5a-6ee1-4357-bbbc-77aec8a996b1-0
00:36:28.520 --> 00:36:28.720
Yeah.

e1623dd2-fd6c-4b5a-be19-12b70aab2b1b-0
00:36:28.720 --> 00:36:32.067
And I think, you know,
we've spent a lot of time today talking

e1623dd2-fd6c-4b5a-be19-12b70aab2b1b-1
00:36:32.067 --> 00:36:36.159
about governance as it relates to the
data and, and the quality of the data,

e1623dd2-fd6c-4b5a-be19-12b70aab2b1b-2
00:36:36.159 --> 00:36:39.720
the quality of the work the agents do,
and all those kinds of things.

cd6e4171-c71d-41da-bd68-aa2d48b448d5-0
00:36:39.720 --> 00:36:41.640
But I think the point you're making is
exactly right.

7114dd41-2921-41e2-b82e-0bb8bd586c67-0
00:36:41.640 --> 00:36:45.253
And I, and I think that, you know,
this idea that, you know,

7114dd41-2921-41e2-b82e-0bb8bd586c67-1
00:36:45.253 --> 00:36:49.637
hackers using AI mean, you know,
I'm quite sure that's been going on for,

7114dd41-2921-41e2-b82e-0bb8bd586c67-2
00:36:49.637 --> 00:36:51.000
for, for quite a while.

96ad0756-e49c-4ae2-a073-ef02290eccd4-0
00:36:51.000 --> 00:36:54.115
I, I don't relish the job of,
of a cybersecurity, you know,

96ad0756-e49c-4ae2-a073-ef02290eccd4-1
00:36:54.115 --> 00:36:56.504
person at all,
because I think to your point,

96ad0756-e49c-4ae2-a073-ef02290eccd4-2
00:36:56.504 --> 00:37:00.607
you're literally going to have agents
fighting agents and who can evolve them,

96ad0756-e49c-4ae2-a073-ef02290eccd4-3
00:37:00.607 --> 00:37:03.360
you know,
fast enough and all those kinds of things.

47eb3be4-23df-4777-bb41-c68bc8540d58-0
00:37:03.880 --> 00:37:06.413
And it's, you know,
it's the sad reality of the, the,

47eb3be4-23df-4777-bb41-c68bc8540d58-1
00:37:06.413 --> 00:37:08.760
the world that we're in, But it's,
it's a real 1.

cbb59a04-0c9d-4cbd-a728-c83f79fa9a7e-0
00:37:08.760 --> 00:37:12.639
And I think the,
one of the things that's going to be,

cbb59a04-0c9d-4cbd-a728-c83f79fa9a7e-1
00:37:12.639 --> 00:37:17.083
you know, critical is that again,
what data we make available,

cbb59a04-0c9d-4cbd-a728-c83f79fa9a7e-2
00:37:17.083 --> 00:37:21.950
how we make it available,
how we govern it, The, you know, the, the,

cbb59a04-0c9d-4cbd-a728-c83f79fa9a7e-3
00:37:21.950 --> 00:37:24.560
the guardrails that we put around it.

d78f16a4-ca72-4e8a-88f5-7dd843354601-0
00:37:24.560 --> 00:37:28.760
You think about our customers,
we've got PII of all different sorts.

31e4e13d-e03a-45bc-95de-c3775d965f95-0
00:37:28.760 --> 00:37:32.197
We've got banking information,
we've got mortgage information,

31e4e13d-e03a-45bc-95de-c3775d965f95-1
00:37:32.197 --> 00:37:35.688
we've got claim information,
we've got Social Security numbers,

31e4e13d-e03a-45bc-95de-c3775d965f95-2
00:37:35.688 --> 00:37:39.507
credit card numbers, you know,
in our customers, We've got, you know,

31e4e13d-e03a-45bc-95de-c3775d965f95-3
00:37:39.507 --> 00:37:43.927
all these kinds of things that, you know,
hackers are trying to identify all the

31e4e13d-e03a-45bc-95de-c3775d965f95-4
00:37:43.927 --> 00:37:44.199
time.

82894840-09b2-4dc7-b1ff-e6d8b90da6a7-0
00:37:44.200 --> 00:37:46.920
Well, you know,
set hackers aside for the moment.

28785a8f-737a-482d-bebd-7b0a5326efef-0
00:37:47.200 --> 00:37:50.831
What's the risk of letting an agent that
you know, at the end of the day,

28785a8f-737a-482d-bebd-7b0a5326efef-1
00:37:50.831 --> 00:37:53.040
just follows the instructions that you
give.

41967b57-04c6-499f-a409-2225048b49ef-0
00:37:53.040 --> 00:37:56.896
It brings information back and you know,
oops, here's a, you know,

41967b57-04c6-499f-a409-2225048b49ef-1
00:37:56.896 --> 00:38:01.387
Social Security numbers or here's a real world
how you know, here's a medical chart,

41967b57-04c6-499f-a409-2225048b49ef-2
00:38:01.387 --> 00:38:03.920
a real world, you know, a real,
real world.

7d6f3416-f221-416e-b4e2-e6cf9b2236d7-0
00:38:03.920 --> 00:38:08.440
And, and the, the AI, you know,
will follow its instructions.

db6116b9-78e3-486c-94ab-2be43f2fca0b-0
00:38:08.440 --> 00:38:09.920
It's going to bring that information back.

8556649c-e02f-4d34-bfc1-54388352ae2a-0
00:38:10.280 --> 00:38:15.588
This is why that that whole kind of
intermediary step of safeguarding what it

8556649c-e02f-4d34-bfc1-54388352ae2a-1
00:38:15.588 --> 00:38:20.760
is that you're making available becomes,
you know, that much more critical.

d1fdf89c-ae6c-48da-854b-99339e36e11d-0
00:38:20.840 --> 00:38:26.969
And it's why I think it'll be very
interesting over the next couple years to

d1fdf89c-ae6c-48da-854b-99339e36e11d-1
00:38:26.969 --> 00:38:28.960
see how companies handle.

bb7a496e-ee87-4e72-8f6c-01fb46ae762c-0
00:38:28.960 --> 00:38:31.080
We've talked about MCP, for example.

fb0e0bff-297a-4711-947d-db5f7ef448cf-0
00:38:31.480 --> 00:38:36.174
So there is this idea that I could have
my agent leverage or agents leveraging

fb0e0bff-297a-4711-947d-db5f7ef448cf-1
00:38:36.174 --> 00:38:40.453
MCP to go to N number of endpoints to get
data aggregate, decide on it,

fb0e0bff-297a-4711-947d-db5f7ef448cf-2
00:38:40.453 --> 00:38:41.879
do those kind of things.

439b1d4c-ca78-490e-adeb-abd03e1476ad-0
00:38:42.240 --> 00:38:45.233
Or,
and or I could centralized my data in a

439b1d4c-ca78-490e-adeb-abd03e1476ad-1
00:38:45.233 --> 00:38:49.520
data lake and I have my agent communicate
with that data lake.

46bc44be-622c-4564-83f8-faf7a2033d75-0
00:38:49.520 --> 00:38:53.903
Whichever you're choosing, you know,
you have this security element, you know,

46bc44be-622c-4564-83f8-faf7a2033d75-1
00:38:53.903 --> 00:38:57.233
associated with it, of course,
to make sure that, you know,

46bc44be-622c-4564-83f8-faf7a2033d75-2
00:38:57.233 --> 00:39:00.840
I'm safeguarding those endpoints and you
know, all of that data.

a19f6744-a6b8-4c85-8f17-5c9ab4b463da-0
00:39:02.160 --> 00:39:09.280
But without a plan as to what are we
pulling back and how are we presenting it?

03d9dbe2-173d-412b-9183-a8b4f38ea79e-0
00:39:09.880 --> 00:39:15.480
What are we giving an age in the
authority to decide on?

018d9a25-8982-4345-be13-05d80df35445-0
00:39:17.800 --> 00:39:22.038
You know,
not letting efficiency get in the way of

018d9a25-8982-4345-be13-05d80df35445-1
00:39:22.038 --> 00:39:25.280
good safe practices is super important.

cd38f36e-8306-435a-a132-7bc04083b875-0
00:39:26.800 --> 00:39:30.334
So, you know,
I want to just quote by saying, if,

cd38f36e-8306-435a-a132-7bc04083b875-1
00:39:30.334 --> 00:39:34.646
if someone's listening to this,
who is an enterprise leader,

cd38f36e-8306-435a-a132-7bc04083b875-2
00:39:34.646 --> 00:39:38.958
a business leader, um,
I imagine they're past the question,

cd38f36e-8306-435a-a132-7bc04083b875-3
00:39:38.958 --> 00:39:40.160
should we use AI?

fda4df62-c490-4494-b844-bbdaa5eccd3c-0
00:39:40.160 --> 00:39:45.760
Most people are have moved on to how or
why did we use it the wrong way?

880434fc-b8cd-45f7-b32b-66d1871b3f00-0
00:39:45.760 --> 00:39:47.440
Whatever, we've talked about that before.

7f4f4f8f-c41c-497c-8d13-e82772a4883a-0
00:39:47.440 --> 00:39:53.191
But I think the question really becomes,
do we have the governance maturity to

7f4f4f8f-c41c-497c-8d13-e82772a4883a-1
00:39:53.191 --> 00:39:54.720
scale it responsibly?

40053c98-28c2-461b-8bff-b6ad78951b40-0
00:39:55.480 --> 00:39:59.120
I think that's where the question needs
to start in the boardroom.

0191a429-8334-4202-ad4c-825d4605d402-0
00:39:59.640 --> 00:40:04.537
Do we have what we need to do this in a
responsible way to protect our business,

0191a429-8334-4202-ad4c-825d4605d402-1
00:40:04.537 --> 00:40:08.467
to protect our people,
protect our patients, whatever, you know,

0191a429-8334-4202-ad4c-825d4605d402-2
00:40:08.467 --> 00:40:10.160
to protect our constituents.

7f1b4225-eb7b-4c4a-aeca-8d8171a1dddf-0
00:40:10.160 --> 00:40:13.760
If you're in government, like there's,
there's a lot of that's the kind of

7f1b4225-eb7b-4c4a-aeca-8d8171a1dddf-1
00:40:13.760 --> 00:40:15.680
question that needs to be asked upfront.

51301c38-b62b-4621-b80c-6d24ffdd02f5-0
00:40:16.280 --> 00:40:18.957
Yeah,
particularly as you you think about

51301c38-b62b-4621-b80c-6d24ffdd02f5-1
00:40:18.957 --> 00:40:20.360
venturing down a path.

b5c71585-4012-4bbe-af2f-61ad9c27bda6-0
00:40:20.360 --> 00:40:22.240
I mean, this is true of AI generally, but.

2d0734db-a689-441d-ae91-7b4f0990ec69-0
00:40:23.240 --> 00:40:27.987
As we've talked about here quite a bit,
there is this, you know,

2d0734db-a689-441d-ae91-7b4f0990ec69-1
00:40:27.987 --> 00:40:33.465
kind of final frontier of all of this
unstructured data that's been as yet

2d0734db-a689-441d-ae91-7b4f0990ec69-2
00:40:33.465 --> 00:40:38.650
untouched and as yet ungoverned in the
sense that in a typical content

2d0734db-a689-441d-ae91-7b4f0990ec69-3
00:40:38.650 --> 00:40:41.279
repository you have metadata, right?

eab25e72-f3a3-4b79-b1b0-fbb31ba1e997-0
00:40:41.400 --> 00:40:45.030
This is how I go find this document and
you'll find the document,

eab25e72-f3a3-4b79-b1b0-fbb31ba1e997-1
00:40:45.030 --> 00:40:47.120
which is effectively a picture, right?

9d1b00e0-5eaf-4156-b99c-fcdf69007cc3-0
00:40:47.160 --> 00:40:49.320
And I can read through it and I can find
information.

d5553c96-2dbd-4d9c-b05c-165b2a716c1f-0
00:40:49.320 --> 00:40:54.808
Now what we're saying is no,
we're going to use tools to lift data out

d5553c96-2dbd-4d9c-b05c-165b2a716c1f-1
00:40:54.808 --> 00:40:57.360
of that for downstream use by AI.

0c544dff-d151-4cf0-a779-b2cc33f2bc1c-0
00:40:57.840 --> 00:41:00.111
Well,
all the models that went into the

0c544dff-d151-4cf0-a779-b2cc33f2bc1c-1
00:41:00.111 --> 00:41:04.710
discrete data in your ERP or your CRM or
your HR system that have been carefully

0c544dff-d151-4cf0-a779-b2cc33f2bc1c-2
00:41:04.710 --> 00:41:07.720
curated with security and all those kinds
of things.

b8059b11-a677-4644-97e1-5511ba3b7b47-0
00:41:08.080 --> 00:41:13.302
Now all of a sudden you're saying, oh,
go out and go get this new pile of data

b8059b11-a677-4644-97e1-5511ba3b7b47-1
00:41:13.302 --> 00:41:15.880
and what, you're just going to dump it.

596f70e2-5a14-4d88-b612-1863d9afa4c9-0
00:41:16.360 --> 00:41:19.988
You know, and,
and I know I'm being extreme,

596f70e2-5a14-4d88-b612-1863d9afa4c9-1
00:41:19.988 --> 00:41:21.520
but it's this idea.

54506ab7-d7e9-42bf-a23b-cd9079992912-0
00:41:21.520 --> 00:41:25.458
And I think the root point we're trying
to get at is you put all this time,

54506ab7-d7e9-42bf-a23b-cd9079992912-1
00:41:25.458 --> 00:41:28.878
energy and thought into safeguarding the
data in those, you know,

54506ab7-d7e9-42bf-a23b-cd9079992912-2
00:41:28.878 --> 00:41:32.040
kind of endpoint discrete data
application, CRM, HR, etc.

0928c304-7e03-4c0a-a29a-672b3551684d-0
00:41:32.440 --> 00:41:35.200
Now all of a sudden you're saying,
all right, I've got this.

3f7eceb7-3eb7-4822-86f5-4b02ae0b9899-0
00:41:36.160 --> 00:41:38.164
Well, you,
you have to apply the same thought

3f7eceb7-3eb7-4822-86f5-4b02ae0b9899-1
00:41:38.164 --> 00:41:38.600
processes.

b72988d2-f610-432f-bde3-6156cd07a498-0
00:41:38.600 --> 00:41:43.134
You have to thought, you know,
apply policies, procedures, you've got to,

b72988d2-f610-432f-bde3-6156cd07a498-1
00:41:43.134 --> 00:41:45.953
you know,
get the data sort of into different

b72988d2-f610-432f-bde3-6156cd07a498-2
00:41:45.953 --> 00:41:51.040
levels of what I want to secure versus,
you know, what's I can be more open about.

cdc3eb09-3183-4727-aab5-2d58710e3ecd-0
00:41:51.400 --> 00:41:55.046
And you want to have, you know,
all the things we talked about earlier in

cdc3eb09-3183-4727-aab5-2d58710e3ecd-1
00:41:55.046 --> 00:41:59.137
terms of the ability to traceability and
all those kinds of things to know exactly

cdc3eb09-3183-4727-aab5-2d58710e3ecd-2
00:41:59.137 --> 00:42:03.079
where did it come from, who originated,
who touched it, how did it end up here?

edf46005-17e6-4c88-963e-3cee79e19817-0
00:42:03.480 --> 00:42:07.234
And you know,
when and if we run into something where

edf46005-17e6-4c88-963e-3cee79e19817-1
00:42:07.234 --> 00:42:10.920
boy, this agent got this wrong,
how did this happen?

7e601514-e283-46b0-afe0-79ce3536d4f7-0
00:42:11.400 --> 00:42:13.160
Well, that can't map back to.

fbb367ca-683e-4b20-8d64-66114b9a3fcc-0
00:42:13.240 --> 00:42:14.440
I don't know, it was in this pile.

0fcebe4a-6260-4120-a581-410a76832331-0
00:42:14.840 --> 00:42:15.680
Yeah, right.

fa79be2e-cb00-42cf-b46c-4a209dc495c4-0
00:42:16.200 --> 00:42:16.480
Yeah.

201a7cca-1af0-4f2f-9f8c-580e2886222c-0
00:42:16.760 --> 00:42:19.440
There are a lot we could go into here.

8cd813e8-9934-4c0a-94d2-b6513075feda-0
00:42:19.640 --> 00:42:21.800
This has been a good a good conversation.

f43ac386-2120-4a28-98a7-3b4fc645859e-0
00:42:21.800 --> 00:42:22.560
As always.

b9cf4c1a-8f50-4924-a116-b716eacf89db-0
00:42:23.440 --> 00:42:28.052
It's always fun to talk about this stuff
and then find out that in six months from

b9cf4c1a-8f50-4924-a116-b716eacf89db-1
00:42:28.052 --> 00:42:31.720
now we need to re record this because
something else has changed.

df8f0cc3-d297-4d40-a4da-ee600647f949-0
00:42:31.720 --> 00:42:32.240
No doubt.

70e0a0b8-9b4d-437e-8f86-33cbc95b85ad-0
00:42:32.840 --> 00:42:33.560
No doubt.

897cede1-d7d1-41d9-9ebc-480c3cc30bf4-0
00:42:33.720 --> 00:42:36.734
On that note,
thanks for joining us on the Mostly

897cede1-d7d1-41d9-9ebc-480c3cc30bf4-1
00:42:36.734 --> 00:42:38.000
Unstructured podcast.

3d1f3a0d-bb03-407a-9858-a1241de69d3d-0
00:42:38.680 --> 00:42:39.880
Ed and Clay signing off.

19f43a55-2aa7-486e-b5ca-f94ed7eedc27-0
00:42:40.160 --> 00:42:41.960
I do want to invite you to engage with us.

e7e25cc5-397e-48a3-80c2-ce4912e3fa22-0
00:42:41.960 --> 00:42:46.012
If you are watching our podcast on
YouTube, make sure that you subscribe,

e7e25cc5-397e-48a3-80c2-ce4912e3fa22-1
00:42:46.012 --> 00:42:47.600
like and then please comment.

295274ca-0a98-4bf3-955e-44df0fd4b845-0
00:42:47.600 --> 00:42:48.720
We would love to engage with you.

7a415007-6b44-461c-ac88-b4c757e9b672-0
00:42:48.720 --> 00:42:52.928
If you have take issue with something
we've said or you have more questions,

7a415007-6b44-461c-ac88-b4c757e9b672-1
00:42:52.928 --> 00:42:54.240
we would love to engage.

fcefabc5-d666-4b84-a050-5880de6c81b4-0
00:42:54.480 --> 00:42:57.240
You can also go to kemarkinc.com,
drop us a line there.

532dce38-f887-4618-a61e-f06d3e6fa1d2-0
00:42:57.240 --> 00:42:58.360
We'd love to chat with you.

2df251a7-bdb4-4cb0-8df2-aac57eaff9c7-0
00:42:58.760 --> 00:43:00.824
And if you heard something brilliant,
well,

2df251a7-bdb4-4cb0-8df2-aac57eaff9c7-1
00:43:00.824 --> 00:43:02.560
just assume that Ed and I planned it.

765563ce-7179-4302-96a8-a46793d41d52-0
00:43:03.200 --> 00:43:03.800
See you next time.

b23e7589-f65a-484f-a4a9-c521adc4ff85-0
00:43:03.840 --> 00:43:04.440
Have a good one.