WEBVTT

00:00:00.000 --> 00:00:03.720
Be forceful,
but also allow degrees of freedom to your

00:00:03.720 --> 00:00:05.840
teams to re-engineer themselves.

00:00:07.240 --> 00:00:10.600
We're joined by Joe from Salesforce and
Jean Manuel from BCG.

00:00:10.600 --> 00:00:12.240
Welcome.

00:00:12.400 --> 00:00:16.400
You've described this moment as
transcendental as far as AI and the

00:00:16.400 --> 00:00:20.520
customer experiences is concerned,
which is a remarkable description.

00:00:20.520 --> 00:00:22.320
Just explain what's changed.

00:00:22.640 --> 00:00:24.320
Why is this such a big moment for it?

00:00:25.080 --> 00:00:28.360
Well, I think, you know,
I use transcendental because I think it's

00:00:28.360 --> 00:00:32.520
a great word that really talks about the
fact that the things that came before are

00:00:32.520 --> 00:00:34.560
just very different than where we are now.

00:00:35.440 --> 00:00:37.840
You know,
expecting an agent to be able to

00:00:37.840 --> 00:00:41.880
understand people in, you know,
all sorts of ways, nuances of language,

00:00:41.880 --> 00:00:46.160
indirect ways of asking a question,
spelling errors, all that sort of stuff.

00:00:46.600 --> 00:00:48.600
You know,
computers would have had a really hard

00:00:48.600 --> 00:00:49.920
time doing that three years ago.

00:00:50.120 --> 00:00:51.800
And now it just works.

00:00:51.880 --> 00:00:53.920
And they can, and they understand intent.

00:00:53.920 --> 00:00:57.160
They can, you know,
fake or express empathy, you know, like,

00:00:57.160 --> 00:00:59.680
however,
whatever side of the line you get on,

00:00:59.680 --> 00:01:02.520
but that then you stack those
capabilities

00:01:02.520 --> 00:01:03.960
and they drive real business outcomes.

00:01:04.320 --> 00:01:08.720
And so now you can really have a
meaningful relationship with your

00:01:08.720 --> 00:01:11.880
customer,
portraying your brand values in a way

00:01:11.880 --> 00:01:15.280
that just resonates,
similar to talking to a human.

00:01:15.720 --> 00:01:18.680
And that's just something that we just
never seen before in computer science.

00:01:18.720 --> 00:01:19.520
Always the promise.

00:01:20.000 --> 00:01:21.840
Now it's actually the delivery of that
promise.

00:01:22.200 --> 00:01:23.200
It's finally arrived.

00:01:23.200 --> 00:01:27.720
Jean Manuel,
we've seen AI transforming the customer experience.

00:01:27.720 --> 00:01:29.680
As we just said,
it's transcendental moment.

00:01:30.120 --> 00:01:31.880
How is that changing pricing strategy?

00:01:32.480 --> 00:01:36.800
I think it's changing the way companies
are going to get paid and extract value

00:01:36.800 --> 00:01:38.200
for the value they create.

00:01:39.000 --> 00:01:40.640
And you could see it already with AI.

00:01:40.960 --> 00:01:45.240
Right now the main model of AI is you pay
per person, right?

00:01:45.600 --> 00:01:50.360
So on Open AI and on, you know,
and so that means we are enhancing the

00:01:50.360 --> 00:01:53.480
role of a person,
the value a person can add.

00:01:53.800 --> 00:01:57.400
Soon with agents,
we're going to get the mole of getting

00:01:57.400 --> 00:02:00.360
paid per task with agents achieving
something.

00:02:01.440 --> 00:02:04.840
And so these two models are going to run
in parallels,

00:02:04.840 --> 00:02:08.160
corresponding to the two ways AI can add
value to us,

00:02:08.160 --> 00:02:11.120
either by doing things or by making us
better.

00:02:11.400 --> 00:02:12.920
Yeah, which is remarkable.

00:02:13.240 --> 00:02:17.640
Why are some companies benefiting from AI
and others not so much yet?

00:02:18.280 --> 00:02:22.920
Well, you know, I think AI is a tool,
but you have to have the raw materials

00:02:22.920 --> 00:02:25.440
and it really comes back down to the data.

00:02:25.440 --> 00:02:29.080
And I think what happens is if you look
at where people have been, you know,

00:02:29.080 --> 00:02:31.040
there's a mechanization phase of the data.

00:02:31.040 --> 00:02:33.760
Can you actually get it,
collect it and put it someplace?

00:02:33.840 --> 00:02:35.280
And then there's an automation phase.

00:02:35.720 --> 00:02:39.320
And then traditionally I would have said
machine learning phase where you can

00:02:39.320 --> 00:02:42.280
actually then start take actions based
upon real deep insights.

00:02:42.720 --> 00:02:44.360
And now we have sort of the agentic side.

00:02:44.720 --> 00:02:47.120
I'm not even sure you'd need three,
as much anymore.

00:02:47.360 --> 00:02:50.720
But if you haven't started with one,
then you're going to have a real hard

00:02:50.720 --> 00:02:51.720
time getting to agents.

00:02:51.760 --> 00:02:54.320
Whereas if companies have curated,
they have a data lake,

00:02:54.320 --> 00:02:57.720
they're already taking some actions,
maybe some machine learning in there,

00:02:57.720 --> 00:02:58.760
they're already prepped.

00:02:58.760 --> 00:03:01.160
So at that point,
all of those data sources just become

00:03:01.160 --> 00:03:02.600
sauce for the goose of the agent.

00:03:03.120 --> 00:03:05.320
But if you haven't started that journey,
then you have to catch up.

00:03:05.720 --> 00:03:08.240
Are you seeing this two-tier system now
effectively

00:03:08.240 --> 00:03:09.920
with some companies speeding ahead?

00:03:10.240 --> 00:03:10.760
That's right.

00:03:10.920 --> 00:03:14.320
And I think in addition to the data that
you talked about,

00:03:14.320 --> 00:03:19.040
there is also a readiness that is needed
about changing your entire processes and

00:03:19.040 --> 00:03:21.320
to re-imagine what you are going to do.

00:03:22.520 --> 00:03:25.080
Usually AI can't replace completely what
humans do.

00:03:25.080 --> 00:03:29.360
They do some part of it much better,
but they allow humans to focus on other

00:03:29.360 --> 00:03:33.760
things that are most of the time much
more value-added, much more interesting.

00:03:34.040 --> 00:03:37.640
But you need to allow the degrees of
freedom for people to change their roles.

00:03:37.640 --> 00:03:42.320
So you need to look at the incentives,
the organization structure and to rethink

00:03:42.320 --> 00:03:43.480
that simultaneously.

00:03:43.880 --> 00:03:48.600
So companies that are natives and already
start from scratch tend to invent the new

00:03:48.600 --> 00:03:49.760
processes right away.

00:03:50.160 --> 00:03:54.960
Companies that need to transform from an
existing to the new need to really be

00:03:54.960 --> 00:03:57.400
deciding very forcefully to go this way.

00:03:57.760 --> 00:04:00.760
And if you're not very forceful,
you might be stuck

00:04:00.760 --> 00:04:02.520
in the old ways for quite a while.

00:04:02.760 --> 00:04:07.160
What's one piece of advice that you'd
give to any company now on how they can

00:04:07.160 --> 00:04:08.240
get ahead using AI?

00:04:08.560 --> 00:04:15.040
Well, the first of all is to be forceful,
but also allow degrees of freedom to your

00:04:15.040 --> 00:04:17.520
teams to re-engineer themselves.

00:04:17.720 --> 00:04:21.120
So if you leave everybody in very narrow
roles and you ask yourself,

00:04:21.120 --> 00:04:23.720
can we replace that role or all the roles,

00:04:23.720 --> 00:04:25.800
maybe none of the roles could be replaced completely.

00:04:25.920 --> 00:04:29.920
But you rethink through the entire value
chain and the process that people were

00:04:29.920 --> 00:04:33.920
were doing and there's new roles that are
going to emerge and you need to think

00:04:33.920 --> 00:04:34.960
already through that.

00:04:35.560 --> 00:04:37.080
Yeah, I totally agree.

00:04:37.360 --> 00:04:38.840
I think that's the way I look at it as
well.

00:04:38.840 --> 00:04:42.400
But I also think it's a question of be
really, really,

00:04:42.400 --> 00:04:46.200
really crisp on the what you want to get
done and be very,

00:04:46.200 --> 00:04:48.920
very flexible on the how you get it done.

00:04:49.520 --> 00:04:52.920
And I think that's where some people
confuse those two things.

00:04:52.920 --> 00:04:56.800
They confuse the what and the how because
of people, because of jobs,

00:04:56.800 --> 00:04:58.840
and I think that does make it harder.

00:04:58.840 --> 00:05:02.120
But if you're really clear about the what
and you're flexible about the how,

00:05:02.120 --> 00:05:04.640
you can get some amazing things
accomplished very quickly.

00:05:05.000 --> 00:05:05.920
Excellent advice.

00:05:06.280 --> 00:05:08.440
Well,
thank you both for joining us here in Cannes.

00:05:08.640 --> 00:05:09.320
Thanks for having us.

00:05:09.680 --> 00:05:10.320
Thank you very much, Manveen.