WEBVTT

00:00:00.125 --> 00:00:02.752
We are here at AWS re:Invent.

00:00:02.752 --> 00:00:05.046
We are connecting
with numerous executives.

00:00:05.046 --> 00:00:07.465
Delighted to do so with David Rokeach

00:00:07.465 --> 00:00:11.302
He is the VP Enterprise AI at Scale AI.

00:00:11.386 --> 00:00:13.471
David, great to see you. Good to see you.

00:00:13.471 --> 00:00:17.851
What problems are Scale
AI helping to solve?

00:00:17.934 --> 00:00:22.147
So Scale’s mission is to accelerate
the deployment of AI applications

00:00:22.230 --> 00:00:25.900
and we were founded on day
one with the observation

00:00:25.984 --> 00:00:29.195
that data is the bottleneck
to creating great AI.

00:00:29.279 --> 00:00:33.033
And so over the course of the last seven
years, we've been working very closely

00:00:33.241 --> 00:00:37.579
with the most sophisticated model builders
in the world to power the data

00:00:37.662 --> 00:00:40.665
that helps to train and improve
the performance of their models.

00:00:40.749 --> 00:00:44.461
So from all of the self-driving
vehicle technology you see out there

00:00:44.544 --> 00:00:48.339
to the commercial generative
AI and large language models

00:00:48.339 --> 00:00:51.968
that are now taking the world
by storm, Scale has been quietly powering

00:00:51.968 --> 00:00:55.597
the development of those capabilities
over the past seven years.

00:00:55.680 --> 00:01:00.602
And where we're now seeing that translate
is into the enterprise.

00:01:00.602 --> 00:01:04.439
And so some of those
same needs in high-quality data

00:01:04.522 --> 00:01:08.401
and being able to customize and improve
performance of models is the same thing

00:01:08.401 --> 00:01:11.321
that enterprises need to start
really getting value from this technology.

00:01:11.321 --> 00:01:15.825
So David, what kinds of challenges
do the enterprise companies have?

00:01:15.909 --> 00:01:19.162
Well, I think the biggest thing is
that there's this real need

00:01:19.162 --> 00:01:22.457
to get to a clear view
of return on investment

00:01:22.540 --> 00:01:24.793
and kind of the dirty little secret in

00:01:24.793 --> 00:01:26.753
the space
is that there are all these flashy demos

00:01:26.753 --> 00:01:30.090
and all these exciting opportunities
out there, but there are very few

00:01:30.131 --> 00:01:33.927
applications really in production
in the real world delivering value

00:01:34.010 --> 00:01:34.761
and I think

00:01:34.761 --> 00:01:38.598
what we're seeing in the enterprise
is that there's this gap in being able

00:01:38.598 --> 00:01:42.644
to deploy some of the technology
quickly in a demo fashion,

00:01:42.727 --> 00:01:44.896
but it's really, really difficult
to get to production.

00:01:44.896 --> 00:01:48.691
And so enterprise
executives are looking for solutions that

00:01:48.775 --> 00:01:51.486
help them to move faster
on really hard challenges,

00:01:51.486 --> 00:01:55.448
but also create the infrastructure
they need to build on top of,

00:01:55.448 --> 00:01:58.451
you know, the first couple of use cases
and actually scale up the capability.

00:01:58.535 --> 00:02:02.497
So we know that Scale
AI and BCG are collaborating.

00:02:02.497 --> 00:02:03.873
How does that work?

00:02:03.873 --> 00:02:09.045
Well, BCG and Scale have really
complementary skill sets and solutions.

00:02:09.129 --> 00:02:11.422
on the BCG side,

00:02:11.506 --> 00:02:12.715
really, truly global

00:02:12.715 --> 00:02:16.553
expertise and industry expertise
and being able to deliver, you know,

00:02:16.594 --> 00:02:20.515
insight and solutions
for large scale clients

00:02:20.598 --> 00:02:23.434
driving business transformation
and change management.

00:02:23.434 --> 00:02:26.896
And then on the Scale side,
we really focus on deploying production

00:02:26.896 --> 00:02:28.148
grade infrastructure

00:02:28.148 --> 00:02:29.190
that an enterprise team

00:02:29.190 --> 00:02:32.652
can rely on for some of their hardest
generative AI challenges.

00:02:32.735 --> 00:02:36.406
And when you bring those two things
together, you get to faster

00:02:36.406 --> 00:02:40.368
return on investment,
faster impact, production-grade quality

00:02:40.451 --> 00:02:44.622
capabilities that can scale,
and we do it responsibly and safely

00:02:44.622 --> 00:02:48.376
with the infrastructure and support
needed to really embed responsible

00:02:48.543 --> 00:02:50.545
AI practices
into the solutions that they're building.

00:02:50.545 --> 00:02:55.258
So if 2023 is a stand out year,
what are you excited about for 2024?

00:02:55.341 --> 00:02:59.512
Well, you know, I think 2023
really it's the first inning,

00:02:59.512 --> 00:03:02.515
maybe the bottom
half of the first inning in generative AI,

00:03:02.515 --> 00:03:05.518
what I'm excited about in 2024 is
I think this is the time

00:03:05.518 --> 00:03:08.646
when we really start to see enterprises
deploying these solutions

00:03:08.646 --> 00:03:09.898
in the real world.

00:03:09.898 --> 00:03:11.900
And, you know,
I think at the end of the day,

00:03:11.900 --> 00:03:14.277
AI is going to be more impactful
than the Internet.

00:03:14.277 --> 00:03:16.988
But what it means to get
there is that enterprises

00:03:16.988 --> 00:03:19.324
and consumers need to start
using it in real life settings.

00:03:19.324 --> 00:03:21.659
And I think that's
what we'll start seeing in 2024.

00:03:21.659 --> 00:03:22.827
David, thank you very much.

00:03:22.827 --> 00:03:24.662
David Rokeach. Thank you.