WEBVTT

00:00:00.125 --> 00:00:01.835
So we are seeing a really big shift

00:00:01.835 --> 00:00:05.797
in terms of rethinking, of
moving away from assistance view

00:00:05.797 --> 00:00:09.175
to really take an end-to-end
process workflow view

00:00:09.175 --> 00:00:10.135
to solve this.

00:00:10.135 --> 00:00:11.469
Dirk, Marcus, welcome.

00:00:11.469 --> 00:00:13.138
Marcus, to you first.

00:00:13.138 --> 00:00:15.682
Generally, how are you
seeing telcos manage

00:00:15.682 --> 00:00:18.310
the implementation of AI and agentic AI?

00:00:18.310 --> 00:00:21.354
I think there's a lot of
maturing in the industry now,

00:00:21.354 --> 00:00:23.523
and it's awesome to see companies moving

00:00:23.523 --> 00:00:28.486
from isolated use cases,
experimentation, to really at scale,

00:00:28.737 --> 00:00:33.116
end-to-end workflow usage
of AI and also agentic AI.

00:00:33.116 --> 00:00:35.994
And for me, that's the
biggest change now in 2026,

00:00:35.994 --> 00:00:40.499
moving from small implementations
to add scale end-to-end.

00:00:40.499 --> 00:00:42.584
What about customer service, Dirk?

00:00:42.584 --> 00:00:46.379
How are telcos approaching
that when it comes to AI?

00:00:46.379 --> 00:00:49.174
What we've seen historically
that telcos have managed

00:00:49.174 --> 00:00:50.884
to unify the entry gate

00:00:50.884 --> 00:00:53.720
for customers contacting and reaching out.

00:00:53.720 --> 00:00:56.264
But what still happened in the backend,

00:00:56.264 --> 00:00:57.933
the service agents still needed

00:00:57.933 --> 00:01:00.352
to look into five, six different systems

00:01:00.352 --> 00:01:02.145
to actually solve their case.

00:01:02.145 --> 00:01:05.732
So now AI is really a catalyst behind it

00:01:05.732 --> 00:01:07.776
because that no longer works.

00:01:07.776 --> 00:01:10.070
An agent has trouble to branch

00:01:10.070 --> 00:01:11.738
into so many different systems.

00:01:11.738 --> 00:01:15.534
So we are seeing a really big
shift in terms of rethinking,

00:01:15.534 --> 00:01:17.577
of moving away from assistance view

00:01:17.577 --> 00:01:20.956
to really take an end-to-end
process workflow view

00:01:20.956 --> 00:01:21.790
to solve this.

00:01:21.790 --> 00:01:23.125
And this is what the transformation

00:01:23.125 --> 00:01:24.626
that we are currently seeing.

00:01:24.626 --> 00:01:26.003
What about your own customers?

00:01:26.003 --> 00:01:28.171
What are you hearing from them?

00:01:28.171 --> 00:01:31.174
Yeah, so let me just pick one
of our existing customers.

00:01:31.258 --> 00:01:34.553
They have huge contact centers.

00:01:34.553 --> 00:01:37.222
They have 12,000 people
in their field service.

00:01:37.222 --> 00:01:40.642
And what they were seeing,
that they were doing okay

00:01:40.642 --> 00:01:42.644
in terms of deploying the field agents

00:01:42.644 --> 00:01:44.188
and the efficiency around there,

00:01:44.188 --> 00:01:47.316
but the amount of work
that was necessary manually

00:01:47.316 --> 00:01:50.068
to actually get the work
orders into the system

00:01:50.068 --> 00:01:52.321
with agents needing to pull data

00:01:52.321 --> 00:01:54.782
from so many different
systems to validate,

00:01:54.782 --> 00:01:58.369
to qualify, to ensure that
all the background information

00:01:58.369 --> 00:02:01.413
was in the case file was really a problem.

00:02:01.413 --> 00:02:02.498
And they flipped it around.

00:02:02.498 --> 00:02:06.085
They had an agent before a
human ever touches that case

00:02:06.085 --> 00:02:08.420
do all the legwork for them to make sure

00:02:08.420 --> 00:02:10.047
that everything is in order

00:02:10.047 --> 00:02:11.882
and really giving thousands of hours back

00:02:11.882 --> 00:02:13.342
to the organization.

00:02:13.342 --> 00:02:17.012
And the second bit that is,
for me, even more striking,

00:02:17.012 --> 00:02:19.598
they also made sure that
the quality is right.

00:02:19.598 --> 00:02:22.851
So the feedback we've
got ex-post is like 90%

00:02:22.851 --> 00:02:25.312
of the work orders are
really of great quality,

00:02:25.312 --> 00:02:27.022
so no need to rework.

00:02:27.022 --> 00:02:31.235
So we're seeing a change also
in the mental, cultural piece

00:02:31.235 --> 00:02:33.112
of trusting the system suddenly

00:02:33.112 --> 00:02:35.906
because it no longer
was a POC or anything,

00:02:35.906 --> 00:02:36.907
but really making sure

00:02:36.907 --> 00:02:39.118
that the system is delivering quality.

00:02:39.118 --> 00:02:41.203
90% is a huge rate,

00:02:41.203 --> 00:02:43.205
but that only works if
you take a different view

00:02:43.205 --> 00:02:44.957
on your system landscape.

00:02:44.957 --> 00:02:48.252
Finally, to you both, can you
share a real world moment,

00:02:48.252 --> 00:02:51.922
real world example, where
AI really delivered value?

00:02:51.922 --> 00:02:55.009
What was it that made
it work do you think?

00:02:55.009 --> 00:02:57.803
For me, one moment that brought value

00:02:57.803 --> 00:02:59.638
to life was when people moved away

00:02:59.638 --> 00:03:01.640
from the technical capabilities-

00:03:01.640 --> 00:03:03.392
so note summarization, whatever-

00:03:03.392 --> 00:03:06.437
to actually what generates
value, to Dirk's point,

00:03:06.437 --> 00:03:07.772
to make the work order right

00:03:07.772 --> 00:03:10.149
and to understand if I
get the work order right,

00:03:10.149 --> 00:03:12.443
I get the business value realized.

00:03:12.443 --> 00:03:13.277
Dirk?

00:03:13.277 --> 00:03:14.236
Yeah, I want to build on that.

00:03:14.236 --> 00:03:17.365
I mean, what I'm seeing
is really the shift,

00:03:17.365 --> 00:03:19.367
the separation of having data somewhere

00:03:19.367 --> 00:03:21.869
and having workflows
and systems somewhere.

00:03:21.869 --> 00:03:25.081
Agentic AI only works if you
bring those two together.

00:03:25.081 --> 00:03:28.543
That you can contextualize
the agent's work

00:03:28.543 --> 00:03:30.294
with all the relevant information,

00:03:30.294 --> 00:03:32.797
and that's a hard piece to crack.

00:03:32.797 --> 00:03:34.465
And everybody's talking about it there.

00:03:34.465 --> 00:03:36.717
So that's for me also
the demarcation point

00:03:36.717 --> 00:03:39.595
that we are seeing right now
that everybody's working on.

00:03:39.595 --> 00:03:41.764
Dirk, Marcus, thank you so much.

00:03:41.764 --> 00:03:42.598
Thank you.

00:03:42.598 --> 00:03:43.808
Thank you for having us.