E23: Part 2 Inside Microsoft’s Contact Center Revolution: AI Agents & CoPilot in Action with Jennifer Lowry

WEBVTT
00:00:00.017 --> 00:00:03.577
Hey everyone, welcome to Power Time. Let's have a listen to what's on today's show.
00:00:03.877 --> 00:00:08.337
And the other one, and this is also something that's inside of the contact center
00:00:08.337 --> 00:00:14.657
portfolio, is the agent for knowledge harvesting. And I will probably mess up
00:00:14.657 --> 00:00:17.777
the appropriate marketing term. You know, you guys can add it into it.
00:00:18.717 --> 00:00:23.997
Like there's somebody monitoring that something has gone wrong and they reach
00:00:23.997 --> 00:00:25.657
out to you or they reach out to your agent,
00:00:26.077 --> 00:00:30.237
you know, you know, this guy agent and like, they all sort it out for you and
00:00:30.237 --> 00:00:33.297
they let you know, say, Hey, we saw some, you know, like, you know,
00:00:33.497 --> 00:00:35.877
power cycling on your VM, we solved it.
00:00:36.297 --> 00:00:39.557
And like, you're, you know, you're, you're informed. But yes,
00:00:39.637 --> 00:00:43.877
I think we will get there. And I think we will get there faster than we think that we will.
00:00:44.097 --> 00:00:47.897
And again, given It's power time.
00:00:48.817 --> 00:00:50.397
Get ready. Let's go.
00:00:54.497 --> 00:00:55.457
Thank you.
00:01:00.426 --> 00:01:04.226
Hey everyone, welcome to Power Time. I'm really excited to introduce,
00:01:04.446 --> 00:01:09.746
actually, we will call Jen, as she's been back on the podcast now for the second
00:01:09.746 --> 00:01:13.346
time, our co-host for today. I think that's really appropriate.
00:01:13.946 --> 00:01:19.726
And with Matt as well, my other co-host. So Jen, welcome to a host on Power Time.
00:01:19.906 --> 00:01:23.346
That's what happens when you end up being on the show more than once. Of course.
00:01:23.646 --> 00:01:27.166
What do they say? No good deed goes unpunished, right? Yeah.
00:01:28.406 --> 00:01:32.486
For those who didn't listen to the first podcast, so you come to us from the
00:01:32.486 --> 00:01:37.646
Customer Zero team, which is one of the world's largest contact centers in Microsoft, of course.
00:01:37.846 --> 00:01:42.986
I will let you do the introduction, but the first podcast, everyone loved it. I will just say that.
00:01:43.166 --> 00:01:47.286
And we've got you back to talk about the contact center and the AI and things
00:01:47.286 --> 00:01:49.686
that's going through that at the moment and an update.
00:01:49.886 --> 00:01:55.586
But maybe just an intro, Jen, to the team, to people that haven't heard the
00:01:55.586 --> 00:01:59.906
first one. Yeah, well, first of all, just thank you to you and Matt for having me on again.
00:02:00.206 --> 00:02:03.586
I avoided all the swear words and stuff that it would embarrass me and keep
00:02:03.586 --> 00:02:05.086
me from getting invited back on.
00:02:05.246 --> 00:02:10.126
But I just have the pleasure of working in Microsoft support.
00:02:10.406 --> 00:02:13.746
So a little bit of the statistics on the size of Microsoft support.
00:02:13.746 --> 00:02:17.466
We're about 45,000 support agents strong.
00:02:17.726 --> 00:02:21.826
And, you know, we operate across 92 contact centers.
00:02:22.126 --> 00:02:28.766
And in those contact centers, we get a little over 1 billion consumers calling
00:02:28.766 --> 00:02:31.986
in and 10 million commercial customers.
00:02:31.986 --> 00:02:38.446
And so what that amounts to is about, I don't know, 300 million or more in contacts every year.
00:02:38.706 --> 00:02:42.526
And, you know, I think seeing Microsoft is growing substantially.
00:02:42.646 --> 00:02:45.886
And so the volume also grows.
00:02:46.006 --> 00:02:50.946
And so how do we use technology, this gift that is AI, to be able to serve our customers?
00:02:51.126 --> 00:02:55.986
And so it's just an amazing responsibility that we get to run here in Microsoft Support.
00:02:56.780 --> 00:03:02.860
So did you say 1 billion? Yes. That's just, yeah.
00:03:03.340 --> 00:03:07.300
In Australia, we have a population of 30 million, so to fathom a billion people
00:03:07.300 --> 00:03:10.040
calling into a contact center, that's just a big number.
00:03:10.740 --> 00:03:16.280
It is. It's a bit big. Yeah, a little less than 10% of the world's population
00:03:16.280 --> 00:03:19.020
is in some form or fashion probably calling into us.
00:03:19.660 --> 00:03:23.280
Yeah. Or you could see that as a bad thing, Jennifer, which is,
00:03:23.360 --> 00:03:27.780
you know, there's a large chunk of people calling in with problems about Microsoft Kits.
00:03:28.020 --> 00:03:33.260
So what's the majority of the calls and things that kind of come in?
00:03:33.660 --> 00:03:37.120
You know, it's a great question. And we get so much range.
00:03:37.120 --> 00:03:43.460
So we will get everything from a parent calling in because their 13-year-old
00:03:43.460 --> 00:03:49.160
son had gotten a hold of their credit card and decided to charge a whole bunch of Xbox credits.
00:03:49.680 --> 00:03:51.480
Not to say that that happened, personally.
00:03:54.580 --> 00:03:57.240
My son i will protect his innocence to an
00:03:57.240 --> 00:04:00.660
extent all the way up to we'll get questions on
00:04:00.660 --> 00:04:05.880
just high frequency trading applications that are running on azure and you know
00:04:05.880 --> 00:04:10.020
you don't you know we do some filtering so you're going to get you know consumer
00:04:10.020 --> 00:04:14.640
customers you know coming into consumer support engineers and then your azure
00:04:14.640 --> 00:04:18.240
support engineers but the technology isn't getting more simple,
00:04:18.600 --> 00:04:20.300
right? It's getting, you know, more complex.
00:04:20.820 --> 00:04:25.700
So, you know, and you know this because, you know, you talk to customers about
00:04:25.700 --> 00:04:29.040
the technology and it's not getting easier for you guys either.
00:04:30.060 --> 00:04:36.120
Yeah, a lot of changes at the moment and particularly around AI and agents. And I think as well, the,
00:04:36.813 --> 00:04:39.853
the way that we kind of look at technology now is
00:04:39.853 --> 00:04:42.873
is in particularly in ai is becoming
00:04:42.873 --> 00:04:45.853
a little bit more mature i wouldn't say it's mature yet
00:04:45.853 --> 00:04:49.373
but um you know we're starting to look at multi-agents as
00:04:49.373 --> 00:04:52.553
trying to set up problems for us whether it's before it was
00:04:52.553 --> 00:04:55.293
kind of let's just work on prompting and come up with a
00:04:55.293 --> 00:04:58.913
really good way to get something back out of the large language model now
00:04:58.913 --> 00:05:01.793
we've moved into this this kind of world where
00:05:01.793 --> 00:05:04.793
we can actually get agents to do things on our behalf and what if
00:05:04.793 --> 00:05:07.573
we have multiple agents all running off and doing things and bringing
00:05:07.573 --> 00:05:10.513
data back to like a master agent and what would that look like
00:05:10.513 --> 00:05:13.673
and can we bring data back into different applications so
00:05:13.673 --> 00:05:18.353
that and we're you know really surprised at how accurate some of this stuff
00:05:18.353 --> 00:05:23.773
is as well and and how much it really saves time for our customers so yeah the
00:05:23.773 --> 00:05:27.093
next 12 months i think is going to be a real real testament to the work that
00:05:27.093 --> 00:05:31.613
everyone's created of course you guys were the first in the world, I think,
00:05:31.813 --> 00:05:35.733
to roll out CoPilot in a large way.
00:05:36.633 --> 00:05:39.013
Yeah, Customer Zero, which is why we call you Customer Zero.
00:05:39.313 --> 00:05:43.093
So we're talking two years on now. How's it all going?
00:05:43.433 --> 00:05:49.213
Yeah, well, I love that you mentioned the rise of the agentic capabilities because
00:05:49.213 --> 00:05:55.773
when we were talking last year, we talked in the context of how is CoPilot allowing
00:05:55.773 --> 00:05:57.753
our agents to become super agents.
00:05:58.073 --> 00:06:02.753
And that has evolved. So now when you look at the whole case flow,
00:06:02.953 --> 00:06:06.813
so in the world of support, we are now seeing process automation.
00:06:07.253 --> 00:06:12.293
So pieces of the case, one of the most important things is making sure that
00:06:12.293 --> 00:06:18.033
you have the customer intent connected appropriately with the right support engineer.
00:06:18.233 --> 00:06:23.653
So being able to leverage AI to do that, What that does is it translates into
00:06:23.653 --> 00:06:29.073
a much higher first call resolution. So you're not having to pass them off to other customers.
00:06:29.253 --> 00:06:33.193
And so these are all parts of the whole process that are fully automated.
00:06:34.149 --> 00:06:38.749
And then even some of the more toilsome things, I remember when I came on last
00:06:38.749 --> 00:06:44.729
time, I talked about, you know, we were very intentional about our design principles.
00:06:45.229 --> 00:06:50.729
There are so many different use cases that you could launch with Copilot that
00:06:50.729 --> 00:06:51.829
you can kind of get overwhelmed.
00:06:52.129 --> 00:06:57.769
And so for us, we said first, we want to make sure that we are removing the agent friction.
00:06:57.889 --> 00:07:02.169
We call it over here in the United States, agent toil. I know we joked about
00:07:02.169 --> 00:07:07.549
how toil means in Australia, so we're not at your vacation, don't you worry.
00:07:07.909 --> 00:07:13.589
But one of the other agentic capabilities is this whole case handling process.
00:07:13.829 --> 00:07:18.649
And for our support engineers, the reason why we hire them and recruit them
00:07:18.649 --> 00:07:23.749
is they are really good at solving tough problems under pressure.
00:07:23.749 --> 00:07:28.629
And so once that process is complete, once they've found the solution,
00:07:29.269 --> 00:07:33.169
like the paperwork, the follow-up, making sure that the customer has,
00:07:33.249 --> 00:07:36.349
in fact, remedied the issue and being able to close out the case,
00:07:36.489 --> 00:07:39.049
that paperwork, that's always so fun.
00:07:39.149 --> 00:07:44.609
And so we also now have this process automation around case handling where the
00:07:44.609 --> 00:07:49.329
agentic capabilities is doing that on behalf of our support engineers.
00:07:49.529 --> 00:07:55.769
So again, removing that agent friction. So it's just so cool to see the technology move so fast as well.
00:07:56.493 --> 00:07:59.393
I love that one. For anyone that doesn't like admin, I hate admin.
00:07:59.553 --> 00:08:02.793
So going to go in a meeting and then being able to populate my case,
00:08:02.953 --> 00:08:07.793
that case agent sounds something that would be of very interest to a lot of people.
00:08:07.953 --> 00:08:11.073
And you also touched on intent routing and that intent routing agent.
00:08:11.253 --> 00:08:14.393
I guess, what are the sort of other technologies that you're sort of excited about?
00:08:14.493 --> 00:08:16.873
You may not have enabled them yet within your contact center,
00:08:17.053 --> 00:08:19.153
but what are some of the other features that you've kind of seen that you're
00:08:19.153 --> 00:08:22.133
really interested in obviously exploring in the next 12 months? Yeah.
00:08:22.593 --> 00:08:25.453
You know, the other one, and this is also something that's inside
00:08:25.453 --> 00:08:28.953
of the contact center portfolio is the
00:08:28.953 --> 00:08:31.673
agent for knowledge harvesting and
00:08:31.673 --> 00:08:34.673
i will probably mess up the appropriate marketing term
00:08:34.673 --> 00:08:39.573
you know you guys can add it into the show notes but you know what happens so
00:08:39.573 --> 00:08:46.473
frequently is if you look at support over 90 of the issues have been seen before
00:08:46.473 --> 00:08:53.133
and you know we have ingested from our knowledge corpus all of the release notes,
00:08:53.393 --> 00:08:57.893
the training materials, the product documentation, to be able to help our support
00:08:57.893 --> 00:08:59.693
engineers resolve those problems quickly.
00:09:00.113 --> 00:09:07.613
But what do you do with the 10% where there is not yet knowledge articles available
00:09:07.613 --> 00:09:11.733
for those support engineers to use to solve the problem?
00:09:12.418 --> 00:09:17.958
And we run into this conflict between competing goals.
00:09:18.178 --> 00:09:23.498
Our support engineers, the two things that they care the most about is making
00:09:23.498 --> 00:09:25.458
sure that they have happy customers.
00:09:25.658 --> 00:09:30.258
So CSAT, you know, they want to make sure that their customers' issues are resolved.
00:09:30.518 --> 00:09:33.798
And also making sure that you're clearing the queue, you know, your backlog.
00:09:33.798 --> 00:09:39.238
And so those kind of behaviors, sometimes when you prioritize those,
00:09:39.418 --> 00:09:45.838
it means that when you close out a case that didn't have a knowledge article already available,
00:09:46.238 --> 00:09:51.158
even though you know for future support engineers, you should go write a knowledge
00:09:51.158 --> 00:09:58.118
article that documents all of that stuff that you've done, but that doesn't get handled very often.
00:09:58.118 --> 00:10:05.698
And so what's so cool now is our knowledge management agent can look at the
00:10:05.698 --> 00:10:12.778
flow of the support case, and it could suggest and write the knowledge management article.
00:10:12.958 --> 00:10:17.658
Now, it's so important. We talked a lot about the important ecosystem that you
00:10:17.658 --> 00:10:21.358
need to build where you've got knowledge management custodians that are making
00:10:21.358 --> 00:10:24.378
sure that that asset is appropriate to be incorporated.
00:10:24.378 --> 00:10:29.758
But again, just a game changer of how do you increase the knowledge that's out
00:10:29.758 --> 00:10:31.958
there to solve issues. So isn't that fun?
00:10:33.438 --> 00:10:37.698
Do you think we'll have an agent to kind of do, to be that custodian rather
00:10:37.698 --> 00:10:39.978
than have a person in the middle for that process?
00:10:40.418 --> 00:10:44.158
You know, potentially, I, you know, I don't, don't know just yet.
00:10:44.358 --> 00:10:49.938
You know, there, I, I'm, things are moving so quickly that I imagine the next time we talk.
00:10:50.641 --> 00:10:53.941
You know, potentially, you know, I don't think we're there yet,
00:10:53.941 --> 00:10:57.681
but it's just because we want to make sure that the knowledge articles that
00:10:57.681 --> 00:10:59.461
we have is to a really high level.
00:10:59.601 --> 00:11:03.461
I don't know if you guys remember me talking about this, but when we first ingested
00:11:03.461 --> 00:11:10.701
all of our knowledge articles back a couple of years ago, we ingested 1.1 million knowledge articles.
00:11:11.261 --> 00:11:14.541
And that whole adage that we say of garbage in, garbage out,
00:11:15.001 --> 00:11:17.041
a whole lot of, you know, garbage.
00:11:17.301 --> 00:11:19.421
So we had to go through and cull everything.
00:11:19.721 --> 00:11:24.081
So we want to maintain that high level. So I think we're just being more careful
00:11:24.081 --> 00:11:27.041
about, you know, what gets incorporated into the knowledge base.
00:11:27.161 --> 00:11:28.281
Yeah, stay tuned for more.
00:11:28.921 --> 00:11:33.501
Yeah, looking forward to seeing how agents can support that moving forward.
00:11:33.721 --> 00:11:38.701
I was talking to a learned colleague actually in Seattle around the contact
00:11:38.701 --> 00:11:42.041
center and some comments that we were kind of talking about was,
00:11:42.101 --> 00:11:45.221
you know, the future of the contact center, we have an application
00:11:45.221 --> 00:11:48.381
that humans use and with the advent of agents
00:11:48.381 --> 00:11:51.321
agents are going to go do a lot of the actions so yes
00:11:51.321 --> 00:11:54.581
what we're thinking is you know why do you need contact center
00:11:54.581 --> 00:12:00.121
kind of solution for the 90 that you already have an answer for of calls coming
00:12:00.121 --> 00:12:04.001
into the contact center so what about your digital deflection because i can
00:12:04.001 --> 00:12:07.361
see a world where you know someone gets onto a chatbot and the agent just goes
00:12:07.361 --> 00:12:11.861
and resolves it doesn't even need to go to the contact center i Just go out,
00:12:12.041 --> 00:12:14.841
come back, and it's sold. Yes.
00:12:15.441 --> 00:12:18.021
I'm excited about that as well. In fact, if you think about this,
00:12:18.241 --> 00:12:22.981
when a customer calls in to support, that is the last resort.
00:12:23.261 --> 00:12:27.261
I don't know about you guys, but I always do the mode, you know?
00:12:28.781 --> 00:12:31.281
The best contact center is one that no one calls.
00:12:31.620 --> 00:12:36.020
So that's the theory. I'm on Teams meetings chatting to the live chat trying
00:12:36.020 --> 00:12:39.500
to solve my problems and I'm on calls that I'm just attending and stuff like that.
00:12:39.660 --> 00:12:43.260
So if you see me doing this, it's because I'm trying to pay my phone bill or
00:12:43.260 --> 00:12:44.520
trying to get something like that.
00:12:44.800 --> 00:12:50.800
Of course. Do you ever do this where you'll actually race to see which mode
00:12:50.800 --> 00:12:55.860
can fix your problem faster where you'll try chatbot and then you actually call up?
00:12:55.860 --> 00:13:00.560
I did that recently where I was just kind of curious who was going to solve the issue faster.
00:13:01.340 --> 00:13:05.760
That's awesome. You've told me a loophole about contacting sales and that gets
00:13:05.760 --> 00:13:07.160
you through faster than anything.
00:13:07.400 --> 00:13:11.340
I've used that before. So thank you on that one, Jennifer. We'll close that loop now.
00:13:13.680 --> 00:13:16.600
Yes, we'll be talking about some other vendor. Yes.
00:13:19.340 --> 00:13:24.820
So a reminder of what Matt is talking about is this was a funny sort of surprise
00:13:24.820 --> 00:13:28.880
when we were first rolling out our agent determination. It was on voice.
00:13:28.880 --> 00:13:33.260
So when customers were calling in for issues...
00:13:34.177 --> 00:13:38.297
We suddenly saw a dramatic change in the volume demand.
00:13:38.697 --> 00:13:44.897
So particularly our inside sales volume dropped precipitously.
00:13:44.897 --> 00:13:49.697
And what we had found out was human behavior was taking over,
00:13:49.857 --> 00:13:55.817
where people believed correctly so that they would get access to a person sooner
00:13:55.817 --> 00:13:59.177
if that person was implying that they wanted to make a sale.
00:13:59.377 --> 00:14:02.837
Now, what ended up happening is the moment you talked to that warm body,
00:14:02.837 --> 00:14:05.737
they realize that oh you need an issue with
00:14:05.737 --> 00:14:08.817
support and they will pass you over to support so the issue
00:14:08.817 --> 00:14:11.937
doesn't get resolved faster so in that process
00:14:11.937 --> 00:14:14.637
you know we worked through making sure that
00:14:14.637 --> 00:14:17.497
we had the right team staff for the for
00:14:17.497 --> 00:14:23.557
the call intent yeah interesting did that were you was your team a part of the
00:14:23.557 --> 00:14:26.117
i'm trying to think of the name of it mate you might be able to remind me there
00:14:26.117 --> 00:14:32.257
was a a clive agent that was created that was fixing the you know the cables
00:14:32.257 --> 00:14:35.197
and and whatnot around the world, the fiber cables that run.
00:14:35.457 --> 00:14:41.017
So those fiber cables have a whole bunch of faults and things that come into a contact center.
00:14:41.237 --> 00:14:46.317
And they were able to create an agent that automated that process and made it
00:14:46.317 --> 00:14:49.337
extremely more streamlined pretty much overnight.
00:14:49.657 --> 00:14:51.217
Was your team a part of that, Jeff?
00:14:51.597 --> 00:14:55.517
So part of CSS was I was not specifically involved.
00:14:55.557 --> 00:14:59.837
A few things that my team had gotten involved in is other issues,
00:14:59.837 --> 00:15:04.477
Other sort of quirky issues like the Xbox.
00:15:04.797 --> 00:15:09.237
I'm not sure if you guys knew this, but not Xbox. There was a hardware device
00:15:09.237 --> 00:15:14.577
where the resolution to fix it was actually to put it in the refrigerator or the freezer.
00:15:15.389 --> 00:15:24.009
And so these, we had, I know, something about, you know, putting it into a cold
00:15:24.009 --> 00:15:25.989
environment would help with some of the
00:15:25.989 --> 00:15:29.269
temperature issues that we were experiencing. This was a long time ago.
00:15:29.629 --> 00:15:33.309
And, you know, those are kind of their volume based inquiries.
00:15:33.629 --> 00:15:37.569
And so being able to help resolve things through agents, through voice agents.
00:15:37.709 --> 00:15:41.509
So we've done it through that. We've done a few other ones with power cycling.
00:15:42.249 --> 00:15:44.889
But but my team wasn't particularly involved on the
00:15:44.889 --> 00:15:47.549
on the cables but a lot of the other kind of you know
00:15:47.549 --> 00:15:50.269
similar low complexity issues where you know agents have
00:15:50.269 --> 00:15:53.069
been able to take out the volume so yeah because i
00:15:53.069 --> 00:15:55.989
guess you could you could look at all of
00:15:55.989 --> 00:15:58.929
the calls that come into the contact center now whether it be a digital channel
00:15:58.929 --> 00:16:01.649
or voice channel and kind of bulk those together and
00:16:01.649 --> 00:16:04.469
go is there a possibility that we can create an agent that would
00:16:04.469 --> 00:16:07.369
would actually go and resolve this before we
00:16:07.369 --> 00:16:10.109
you know in the years of old if you go back years of
00:16:10.109 --> 00:16:13.209
old two years ago we would have looked at you
00:16:13.209 --> 00:16:17.329
know what are your top three or four inquiries coming to the contact center
00:16:17.329 --> 00:16:21.029
and you know for a lot of organizations reset my password kind of stuff and
00:16:21.029 --> 00:16:25.869
then we'd go why don't you put a chatbot up and you know we can resolve that
00:16:25.869 --> 00:16:29.089
really quickly just through you know a power automate flow or something in the
00:16:29.089 --> 00:16:32.919
background with power virtual agent back in the day and that was out what we did.
00:16:33.399 --> 00:16:36.779
Now, though, we can look at that at really complex.
00:16:36.959 --> 00:16:40.999
So we could probably go down to, you know, what are your top 50 calls that come
00:16:40.999 --> 00:16:46.119
into the contact center, and then look at agents, how they can really address that entire span.
00:16:46.419 --> 00:16:48.459
And I think that's really the big win.
00:16:49.299 --> 00:16:54.319
My team did a demo recently for a financial institution in Australia,
00:16:54.479 --> 00:16:58.339
where we did multi-agent onboarding, and it was really sophisticated.
00:16:58.539 --> 00:17:02.279
I was really surprised at how we could get that business process and really
00:17:02.279 --> 00:17:05.039
make it work. So I guess my question here is,
00:17:05.559 --> 00:17:10.019
Do you regularly do that? Do you check at these processes and then assess them
00:17:10.019 --> 00:17:14.299
of whether an agent can kind of resolve that end-to-end?
00:17:14.499 --> 00:17:18.179
Or is it something that you review all the time? Yes, I mean, absolutely.
00:17:18.379 --> 00:17:23.519
And I'll tell you, a lot of the work now, if you think about this next phase
00:17:23.519 --> 00:17:28.459
that we're in before, the phase was IT is going to give you these features built
00:17:28.459 --> 00:17:33.199
into dynamics, which we use things like case summary, you know, writing an email.
00:17:33.199 --> 00:17:39.839
Now, and this is perfectly aligning with Satya's vision that he expressed last
00:17:39.839 --> 00:17:46.779
week at MECAP Start, that, you know, we are, you know, the software factory for the planet.
00:17:46.779 --> 00:17:53.279
But it's taking it one further where we are creating the factory for people
00:17:53.279 --> 00:17:57.039
to create software built for their specific needs.
00:17:57.039 --> 00:18:02.939
And so we're starting to see where our support engineers are creating agents that.
00:18:03.461 --> 00:18:08.821
For their specific tasks, right? So like one, you know, one specific example
00:18:08.821 --> 00:18:14.261
is, you know, we have a support manager that wants to make sure that he is being
00:18:14.261 --> 00:18:17.241
supportive to his support engineers.
00:18:17.661 --> 00:18:24.221
And so he's able to create agents to help him prepare for in-depth one-on-ones
00:18:24.221 --> 00:18:29.361
where, you know, he's got the backlog of case cues to be able to look at.
00:18:29.361 --> 00:18:35.001
And so that shortens his prep time dramatically, and it also allows him to be
00:18:35.001 --> 00:18:38.661
present and engaged with his support engineers.
00:18:38.901 --> 00:18:43.621
And so, you know, it's no longer, you know, being fed these specific features.
00:18:44.061 --> 00:18:48.101
But, you know, the agentic capabilities that you have across the portfolio,
00:18:48.101 --> 00:18:52.601
it's really all about your creativity. In fact, I think one of your guests that
00:18:52.601 --> 00:18:58.261
you had on a few weeks ago talked about that, both the creativity as well as the taste, right?
00:18:58.361 --> 00:19:01.461
You need to have both creativity and good. Dara, the taste, yeah.
00:19:02.861 --> 00:19:07.041
It's interesting you touched on something there, which Matt and I have been
00:19:07.041 --> 00:19:12.741
talking about of late, which is, you know, AI can improve the human connection a lot.
00:19:12.861 --> 00:19:16.601
And the example we used was, I think it was my wife, I've got to remember,
00:19:16.601 --> 00:19:21.421
But anyway, she went to a new doctor and the doctor used AI to do case summaries and notes.
00:19:21.641 --> 00:19:28.021
And it meant that the doctor could 100% focus on her rather than be sitting
00:19:28.021 --> 00:19:29.541
there taking notes all day.
00:19:29.761 --> 00:19:34.641
And I think this is the thing, right? Like if, and I'm seeing it in management as well.
00:19:34.641 --> 00:19:39.381
If we can help us through what we have, our internal kind of performance connects
00:19:39.381 --> 00:19:44.441
and things, If we can use AI to kind of help us bring all that manual data input
00:19:44.441 --> 00:19:48.161
together and then allow us to spend more time with the people,
00:19:48.461 --> 00:19:52.081
I think human connection will be really improved because if you think about
00:19:52.081 --> 00:19:54.801
it and everyone that's listening in thinks about it,
00:19:55.121 --> 00:19:58.841
you spend a lot of time on devices and you're distracted.
00:19:59.429 --> 00:20:02.909
You're not in the moment with people. And AI kind of allows that.
00:20:03.029 --> 00:20:06.949
And I'm glad that, you know, this is the second story now where I've heard that,
00:20:07.089 --> 00:20:10.249
you know, people are thinking about that. It's so true.
00:20:10.529 --> 00:20:15.649
And I love that, that, you know, with your wife and her talking to her medical
00:20:15.649 --> 00:20:20.469
professional, I've said this in the context of our support community.
00:20:21.189 --> 00:20:25.769
AI, a piece of technology, is allowing us to be more empathetic.
00:20:26.409 --> 00:20:30.649
So I'll give you an example. Yeah. You know, particularly in our consumer space
00:20:30.649 --> 00:20:33.589
where they do so much more in volume.
00:20:34.449 --> 00:20:38.109
And a lot of times when customers are calling in, they're calling in because
00:20:38.109 --> 00:20:41.389
they're having a bad day. You know, like they've been locked out of their account.
00:20:41.649 --> 00:20:43.889
They don't have access to their family photos.
00:20:44.869 --> 00:20:51.129
And our support engineers, they absorb that. And also their empathy bucket,
00:20:51.249 --> 00:20:52.569
if you will, it's a technical term.
00:20:52.829 --> 00:20:57.269
It's not going to be the same at 8 o'clock in the morning as it is at 4 p.m. at night.
00:20:57.749 --> 00:21:03.769
And so what co-pilot and AI is doing is allowing them to be more human, right?
00:21:03.849 --> 00:21:06.369
It's carrying the burden for them. It's not inauthentic.
00:21:06.901 --> 00:21:12.061
So it's such a surprise, right, because you always think about AI as being inhuman.
00:21:12.061 --> 00:21:17.381
And robotic, but to your point, Scott, it is so, so, so much more in terms of
00:21:17.381 --> 00:21:21.841
benefiting both patients, customers, as well as, you know, professionals.
00:21:21.841 --> 00:21:24.921
So yeah on our last podcast we
00:21:24.921 --> 00:21:28.481
had the ceo of kids help line on on
00:21:28.481 --> 00:21:31.701
the podcast and we spoke a lot about their contact center
00:21:31.701 --> 00:21:34.501
which runs on microsoft technology and you
00:21:34.501 --> 00:21:37.281
know the difference for them is that they can see more they can see
00:21:37.281 --> 00:21:41.341
more children the counselors can have a better conversation and
00:21:41.341 --> 00:21:44.381
you know we raised do we think we'll ever be in a position where
00:21:44.381 --> 00:21:47.761
ai will do more more of that frontline counseling
00:21:47.761 --> 00:21:50.521
just so we can you know talk to more more help more
00:21:50.521 --> 00:21:53.481
kids and they're kind of working through that like
00:21:53.481 --> 00:21:56.381
it wasn't it wasn't tracy said it wasn't a definite
00:21:56.381 --> 00:21:59.781
like no we'll never do that we'll always have uh
00:21:59.781 --> 00:22:02.521
it's kind of a person there it's just like what's best for the kids
00:22:02.521 --> 00:22:05.501
and how can we how can we help them so i
00:22:05.501 --> 00:22:08.481
think this kind of mixed ai human in the contact
00:22:08.481 --> 00:22:11.381
center kind of relationship that we have now i
00:22:11.381 --> 00:22:14.701
can actually see a lot of benefits for people at
00:22:14.701 --> 00:22:17.621
the end of that which is surprising because when
00:22:17.621 --> 00:22:20.301
we first kind of launched it was like oh we're going to automate everything and just
00:22:20.301 --> 00:22:23.081
put ai in front of everyone or no it's actually we're actually
00:22:23.081 --> 00:22:28.421
seeing some really good results now yeah i think jennifer i think jennifer touched
00:22:28.421 --> 00:22:32.161
on that in her first podcast as well part of the benefits of ai is being her
00:22:32.161 --> 00:22:34.901
new staff members are more supported as well they're more comfortable in taking
00:22:34.901 --> 00:22:38.541
these calls so there's a really like human empathetic element to actual internal
00:22:38.541 --> 00:22:41.521
staff not just the external customers that are coming through So.
00:22:41.621 --> 00:22:42.841
And again, Tracy from your town.
00:22:42.981 --> 00:22:45.901
Providing that information to the counsellors, again, the kids are getting better support.
00:22:46.295 --> 00:22:49.495
The counselors are more comfortable as well when providing that service.
00:22:49.695 --> 00:22:52.975
So, again, it really resonated with me more turning to Tracy and it took me
00:22:52.975 --> 00:22:55.555
a trip down memory lane to our initial kind of conversation,
00:22:55.715 --> 00:22:57.535
Jennifer, as well. So, yeah. Yeah, yeah.
00:22:57.995 --> 00:23:00.055
Like I'm not sure if I can meet Trixie.
00:23:01.495 --> 00:23:04.835
Her conversation sounds amazing in terms of what they're doing for the community.
00:23:05.355 --> 00:23:06.895
Yeah, definitely, 100%.
00:23:07.555 --> 00:23:12.755
So we're kind of two years on. How's the CSAC going for your internal employees?
00:23:12.895 --> 00:23:17.195
And, you know, are they still super excited about the role? Yes.
00:23:17.495 --> 00:23:22.435
Yeah, absolutely. But both, you know, internal agent satisfaction and also we've
00:23:22.435 --> 00:23:25.175
had a high watermark for our CSAT as well.
00:23:25.595 --> 00:23:28.695
You know, when I talked about, you know, some of the, you know,
00:23:28.895 --> 00:23:33.175
any of your support organizations are really measured across three areas.
00:23:33.315 --> 00:23:35.655
It's your cost, right?
00:23:35.815 --> 00:23:41.615
You know, most of your contact centers run as a percentage of GNL, a percentage of GNA.
00:23:41.915 --> 00:23:45.395
So, you know, you never get more money the following year. obviously
00:23:45.395 --> 00:23:48.595
also your customers and also your agent experience and
00:23:48.595 --> 00:23:51.555
so you know with ai and we're seeing it with
00:23:51.555 --> 00:23:54.695
the results it is impacting all
00:23:54.695 --> 00:23:57.715
of those three metrics in a positive way you
00:23:57.715 --> 00:24:00.995
know but if you think about some of the other transformations that you know
00:24:00.995 --> 00:24:06.275
we've had over the last 25 years many of those were you know cost-cutting measures
00:24:06.275 --> 00:24:10.175
you know i would say that with the exception of say covid where everybody suddenly
00:24:10.175 --> 00:24:14.035
got to work from home in their pajamas which also it was a cost-effective measure
00:24:14.035 --> 00:24:15.735
because we weren't paying for getting an AC.
00:24:16.775 --> 00:24:18.695
I don't know, my super budget went down.
00:24:20.035 --> 00:24:21.815
Matt's just in the comments right now, I'm sure.
00:24:23.395 --> 00:24:24.195
No comment.
00:24:27.015 --> 00:24:31.975
But yeah, look, I'll tell you that where we're going with some of the next capabilities,
00:24:32.095 --> 00:24:35.915
I mentioned the Agenta capabilities, but AI,
00:24:36.475 --> 00:24:42.775
conversational AI, it's very good in language, particularly around translation and,
00:24:43.234 --> 00:24:47.554
And what we've started to work, and this is now in GA instead of Microsoft Teams,
00:24:47.974 --> 00:24:51.074
is this ability to do live translation.
00:24:51.674 --> 00:24:57.514
So you can now be having a conversation, a technical conversation in Mandarin.
00:24:57.934 --> 00:25:01.894
And, you know, on the other end, you know, you're speaking in French or Italian.
00:25:02.314 --> 00:25:05.934
By the way, I was also curious about if it would translate between Australian
00:25:05.934 --> 00:25:09.174
English and American English. It doesn't do that yet.
00:25:10.054 --> 00:25:13.774
Okay. I really need that because a funny story, when we're in Seattle,
00:25:14.174 --> 00:25:17.814
we're walking down one of the main streets and we just wanted some Australian
00:25:17.814 --> 00:25:20.334
food because it is a lot different to American food.
00:25:20.534 --> 00:25:24.514
And we see this shop that says the best pies ever. And I've gone.
00:25:24.894 --> 00:25:28.514
Meat pies. What are we doing? Meat pie. Meat pies. It's going to be awesome.
00:25:28.654 --> 00:25:32.014
We walk in there, we open up the menu and it's full of pizzas.
00:25:32.274 --> 00:25:34.294
And I've gone, since where is a pizza?
00:25:34.594 --> 00:25:37.734
A pie? Australia isn't known for pizza. Yeah, yeah, yeah.
00:25:39.314 --> 00:25:42.434
We're very disappointed. we can test
00:25:42.434 --> 00:25:45.094
that ai by the word toil yeah we'll put toil in and
00:25:45.094 --> 00:25:48.094
we'll see if it translate the australian version of toil to the american soil
00:25:48.094 --> 00:25:51.494
we've got the word to test it with to see if it works so we
00:25:51.494 --> 00:25:53.934
can do it that way so it's interesting i know we're not allowed to talk about
00:25:53.934 --> 00:25:57.074
most stuff that went on at mcap start but there was one demo i want to bring
00:25:57.074 --> 00:26:02.694
up in particular which was seth's demo i don't remember it he demoed uh talking
00:26:02.694 --> 00:26:09.014
to advanced ai that was interacting directly with a shop, I think,
00:26:09.134 --> 00:26:13.094
at the time, and it was flipping between, you know, Japanese and English and
00:26:13.094 --> 00:26:16.794
things like that, and you're getting a pause in the audience because it was so accurate. Yeah.
00:26:17.874 --> 00:26:21.594
Phenomenal. We saw an iteration of that, I think, at Ignite or maybe the last
00:26:21.594 --> 00:26:25.454
MCAP star, but now, Matt, I know you haven't seen the demo, but it was really
00:26:25.454 --> 00:26:26.634
good. I'm building it now.
00:26:26.934 --> 00:26:29.394
How good the language translation is and things.
00:26:29.956 --> 00:26:33.496
Yeah, we're literally building it now. We've got a big CCAS event.
00:26:33.576 --> 00:26:36.436
I think you're actually dialing in for Jen as well, I think.
00:26:37.336 --> 00:26:40.556
So we're doing that at the moment. We're kind of the scenario of like you're
00:26:40.556 --> 00:26:44.356
a university student and you obviously speak some English as well.
00:26:44.456 --> 00:26:45.856
But I go, hey, my natural language is this.
00:26:45.976 --> 00:26:49.536
Do you mind if I have this conversation here? And it flips across to that natural language.
00:26:49.796 --> 00:26:53.096
So again, we're in the process of building that out now.
00:26:53.256 --> 00:26:55.676
And again, I think it's really cool. But that kind of spawned from the idea
00:26:55.676 --> 00:26:58.276
that we were chatting on Teams a while back. and that's what you were talking
00:26:58.276 --> 00:27:00.616
about, this multilingual language translation. I was like, oh,
00:27:00.636 --> 00:27:02.436
I need to play with that. That sounds super cool.
00:27:02.636 --> 00:27:05.036
So yeah, you spawned my demo. So thank you.
00:27:05.356 --> 00:27:09.136
So that must help a lot because I can imagine if someone rings in,
00:27:09.596 --> 00:27:11.496
you know, who only speaks Mandarin,
00:27:12.036 --> 00:27:17.316
you've then got a route to someone that only speaks Mandarin and then they've
00:27:17.316 --> 00:27:20.116
got to have, you know, an open queue for that to happen. Otherwise,
00:27:20.216 --> 00:27:21.316
you might have long wait times.
00:27:22.236 --> 00:27:23.736
Have you seen like a,
00:27:24.492 --> 00:27:30.252
you know, I guess a bump in CSAT for people who don't speak English in their
00:27:30.252 --> 00:27:32.932
contact center because you can do that kind of thing?
00:27:33.132 --> 00:27:37.652
Yeah, it's a great question. So we're still early days in this and the preliminary
00:27:37.652 --> 00:27:40.992
feedback is very positive in terms of CSAT.
00:27:41.192 --> 00:27:44.752
So high CSAT, no DSAT. We also measure DSAT.
00:27:45.292 --> 00:27:49.392
For the next six months, we will probably have lots more exciting stuff to talk about.
00:27:49.612 --> 00:27:55.452
But I mean, to your point of like, you know, having all of those attributes matched up in time.
00:27:55.652 --> 00:28:01.252
So not only, you know, both being able to speak the same language in the same
00:28:01.252 --> 00:28:05.412
time zone with capacity, but then also the technology.
00:28:05.612 --> 00:28:08.752
And, you know, I talked about this earlier, like the technology is not getting easier.
00:28:09.132 --> 00:28:12.832
So you've got across Azure, you know, people will argue about like upwards of
00:28:12.832 --> 00:28:14.852
like 70 different potential technologies.
00:28:15.252 --> 00:28:16.852
And that's just Azure, Right.
00:28:17.292 --> 00:28:21.372
You have all of our other applications as well. Like that's just incredibly complex.
00:28:22.132 --> 00:28:28.572
And, you know, sometimes, you know, the best possible engineer for the job doesn't
00:28:28.572 --> 00:28:32.092
speak the language. And so this is now an ability to be able to both reduce
00:28:32.092 --> 00:28:35.752
cost, but also increase customer set because what do they want?
00:28:35.832 --> 00:28:38.512
They want their issue solved quickly, you know.
00:28:39.229 --> 00:28:44.389
I didn't think about that. That's a really good point. Access to people that,
00:28:44.389 --> 00:28:51.189
you know, have the right skill set but can't communicate. That's a fantastic use case for it.
00:28:52.409 --> 00:28:56.409
So looking forward, what are you most excited about?
00:28:56.669 --> 00:29:00.449
Is there some use cases that you're going to tackle that might help our customers
00:29:00.449 --> 00:29:02.989
that are listening in that are thinking the same thing?
00:29:03.349 --> 00:29:06.569
Yeah. And this is, you know, if I was, you know, boss for a day.
00:29:06.569 --> 00:29:11.969
The thing that I'm really excited about as a potential and is,
00:29:12.289 --> 00:29:17.889
again, I mentioned that inside of support, we are frequently a cost center.
00:29:18.529 --> 00:29:24.869
But because we're in this position of being able to help out customers and we
00:29:24.869 --> 00:29:28.269
understand potentially there's resiliency issues,
00:29:28.269 --> 00:29:34.589
things that they could potentially benefit from in terms of new products that
00:29:34.589 --> 00:29:37.069
would make sense for what their estate is.
00:29:37.860 --> 00:29:45.620
With now the agents presenting potentially, I'll call it upsell opportunities to that customer.
00:29:45.740 --> 00:29:48.100
Now, of course, you have to do it appropriately.
00:29:48.360 --> 00:29:51.840
You don't, when the customer calls in and they say, my VM isn't working,
00:29:52.020 --> 00:29:55.940
you don't say, hey, can I sell you something, you know, like UltraDisk?
00:29:56.100 --> 00:29:57.580
That's not the time to do that.
00:29:58.200 --> 00:30:04.440
But being able to present suggestions of, hey, here's an opportunity that you may consider.
00:30:04.820 --> 00:30:07.520
Like, I'll give you one example. it's a pretty
00:30:07.520 --> 00:30:11.620
straightforward example if somebody's got a m365
00:30:11.620 --> 00:30:14.600
family plan but they notice
00:30:14.600 --> 00:30:20.300
that there's only one you know account or one user that's an example of where
00:30:20.300 --> 00:30:25.560
the customer is not getting you know full value for what they're purchasing
00:30:25.560 --> 00:30:29.560
and that is classic customer success it's not selling them more but making sure
00:30:29.560 --> 00:30:32.460
that they are getting the value for what they bought from Microsoft.
00:30:32.900 --> 00:30:37.480
And so, you know, being able to present that to a support engineer that is probably
00:30:37.480 --> 00:30:41.520
more wired for, I need to fix this issue and move on to the other.
00:30:41.640 --> 00:30:47.500
So that's kind of that overarching theme of how do you change your support organization
00:30:47.500 --> 00:30:53.600
into an adoption organization that drives not only sustaining and maintaining
00:30:53.600 --> 00:30:55.880
the revenue, but potentially growing it.
00:30:55.980 --> 00:30:59.480
And I'm hearing some of my colleagues I talked to outside of Microsoft,
00:31:00.100 --> 00:31:03.440
in support leadership capacities, also thinking about that as a potential.
00:31:03.720 --> 00:31:06.660
So how about you? Are you seeing that from the people that you're talking to in Oz?
00:31:07.370 --> 00:31:12.110
Yeah, we are. I mean, one of the things that we noticed about a year ago was,
00:31:12.230 --> 00:31:17.390
if you look at the engineering org, both sales and service now report up into
00:31:17.390 --> 00:31:19.470
Jeff Comstock and then through to Charles.
00:31:19.670 --> 00:31:24.410
And those products, we've seen technology kind of coexist between the two dynamics
00:31:24.410 --> 00:31:25.830
apps, if that makes sense.
00:31:25.990 --> 00:31:29.330
So they're leveraging sales and customer service and vice versa.
00:31:29.330 --> 00:31:33.650
Because we're thinking that, you know, AI is a really great use case for sales.
00:31:33.850 --> 00:31:36.750
And at the moment, everyone kind of looks at it for customer service.
00:31:36.930 --> 00:31:41.830
But I can definitely see in the future that, you know, a customer service agent
00:31:41.830 --> 00:31:44.830
will be able to create a marketing journey.
00:31:45.210 --> 00:31:50.570
Perhaps in the moment as you're chatting to, you know, IVR or a chatbot,
00:31:50.810 --> 00:31:55.330
it can make suggestions real time about perhaps products and things like that.
00:31:55.470 --> 00:31:58.430
So it's interesting that you're going on that journey as well because I think.
00:31:58.430 --> 00:31:59.550
That's what we're seeing in the market.
00:31:59.979 --> 00:32:02.619
Kind of lean towards and i think there's a
00:32:02.619 --> 00:32:05.599
really good use case everyone hates cold calling i think
00:32:05.599 --> 00:32:10.379
there's a really good use case for automated kind
00:32:10.379 --> 00:32:13.819
of cold calling i just don't know initial screen initial
00:32:13.819 --> 00:32:17.379
screen or something like that just vetting if they are interested and then having
00:32:17.379 --> 00:32:21.479
a human call out where it's a slightly more than just a cold it's slightly warm
00:32:21.479 --> 00:32:25.659
yeah i think that's going to be quite interesting yeah we had this huge scandal
00:32:25.659 --> 00:32:30.379
in australia where it It was called the robocall problem where basically our,
00:32:30.599 --> 00:32:34.379
I don't know what's our version at Centrelink in the US, it would be social
00:32:34.379 --> 00:32:35.739
security, I think. Okay, yeah.
00:32:36.159 --> 00:32:39.979
Kind of just auto-dialed everyone and sent them a message saying they owed a
00:32:39.979 --> 00:32:43.879
debt and they didn't actually owe a debt. So it was really bad.
00:32:44.019 --> 00:32:47.019
A lot of people went into a lot of mental health problems and all sorts of things.
00:32:47.479 --> 00:32:52.559
So I think there's a sensitivity around outbound dialing with AI that it gets it wrong.
00:32:52.919 --> 00:32:56.519
But I definitely think we'll get there because it's a job that,
00:32:56.799 --> 00:32:59.959
you know just humans don't like doing yeah i know i
00:32:59.959 --> 00:33:03.159
know it's it nobody likes rejection you know
00:33:03.159 --> 00:33:07.299
yes no no one
00:33:07.299 --> 00:33:10.679
likes that yeah you can tell me i was on the dating apps before so i'm after
00:33:10.679 --> 00:33:17.739
chatting to advanced voice i could definitely see a day where ai could convince
00:33:17.739 --> 00:33:20.359
me to buy something i didn't want to buy i'm just gonna have to go i don't want
00:33:20.359 --> 00:33:23.859
to chat to the ai because it's going to sell me something and I want it to.
00:33:25.839 --> 00:33:30.459
Very good. Another scenario that we're exploring is
00:33:30.884 --> 00:33:34.864
sort of like proactive outbound when it comes to like student retention at universities
00:33:34.864 --> 00:33:36.884
as well. So that would be college over there, Jen.
00:33:37.044 --> 00:33:40.704
So if your grades are slipping or you haven't attended some classes and something
00:33:40.704 --> 00:33:43.524
like that, how do you be proactive, get in contact with the student,
00:33:43.724 --> 00:33:46.064
make sure they're okay, all of those types of things.
00:33:46.184 --> 00:33:50.044
So again, I think those proactive use cases, I think AI is getting to a point
00:33:50.044 --> 00:33:55.504
where it can look at data and make really informed decisions to be more proactive
00:33:55.504 --> 00:33:57.504
when it comes to support and outbound and reach.
00:33:57.724 --> 00:34:00.584
So yeah, interested to see if you're seeing that stuff in the US as well.
00:34:00.884 --> 00:34:07.184
Yeah, I mean, we're seeing, in that piece around picking up your clientele,
00:34:07.244 --> 00:34:08.684
your community that's at risk,
00:34:08.964 --> 00:34:13.004
not so much a use case that we're seeing inside of support here,
00:34:13.204 --> 00:34:17.804
but over the last year, I've spent a significant amount of time in Europe,
00:34:18.064 --> 00:34:26.484
and there are statutory laws around making sure that you, your citizenship needs.
00:34:27.490 --> 00:34:32.650
Identifying citizenships at risk, and I will probably miss on some of the details, I'll summarize.
00:34:33.110 --> 00:34:42.210
But when an agent at the municipality gets an incoming call from a citizen,
00:34:42.210 --> 00:34:49.530
they are trained to listen for certain things that the citizen says that potentially
00:34:49.530 --> 00:34:51.130
puts them at risk, right? Right.
00:34:51.550 --> 00:34:54.850
And we know that agents are human.
00:34:55.150 --> 00:35:01.150
Sometimes they miss on that. And so being able to have that extra pair of ears,
00:35:01.150 --> 00:35:05.230
they're listening for it, especially when it comes to, you know,
00:35:05.850 --> 00:35:08.030
protecting your most vulnerable.
00:35:08.030 --> 00:35:10.630
So, these are people calling into, you know, say the equivalents,
00:35:10.770 --> 00:35:13.110
you know, Social Security or your utilities.
00:35:13.410 --> 00:35:17.510
And so, you are starting to see that, you know, be done to protect folks that
00:35:17.510 --> 00:35:18.250
need protecting, right?
00:35:18.410 --> 00:35:20.750
You know, not taking advantage of it. But we also know that,
00:35:20.810 --> 00:35:26.950
you know, AI has a potential for, you know, insanity like you guys mentioned with it. Yeah.
00:35:27.850 --> 00:35:32.570
Well, I mean, the FSI customers were seeing that as a use case as well that
00:35:32.570 --> 00:35:36.710
they're looking at, which is can they do screening or it's kind of the AI on
00:35:36.710 --> 00:35:40.070
the shoulder, you know. saying that, you know, something's been said or it shouldn't
00:35:40.070 --> 00:35:42.350
have been said, or he's, it's for that person.
00:35:43.050 --> 00:35:46.350
Based off regulatory compliance during those calls. Yeah.
00:35:46.730 --> 00:35:52.150
Do you think, do you think Microsoft will get to a point where now that we've
00:35:52.150 --> 00:35:55.550
released sort of agent-to-agent communications for MCP,
00:35:56.070 --> 00:35:59.890
do you think that we'll get to a point where I'll kind of have an agent,
00:36:00.050 --> 00:36:02.090
which is my, almost like a personal assistant,
00:36:02.690 --> 00:36:06.670
and, you know, I've got an Azure VM down and I just go, can you just ping Microsoft
00:36:06.670 --> 00:36:11.410
and resolve it, and then that agent will connect to the contact center agent,
00:36:11.610 --> 00:36:14.050
and they'll just sort it out. Yes.
00:36:14.950 --> 00:36:18.070
Do you think that's possible? Mind-blowing. I know. That will happen.
00:36:18.230 --> 00:36:22.030
That will happen. In fact, I think there will be also this point where –,
00:36:22.593 --> 00:36:27.053
your agent doesn't even have to do it. Like there's somebody monitoring that
00:36:27.053 --> 00:36:31.153
something has gone wrong and they reach out to me or they reach out to your
00:36:31.153 --> 00:36:36.013
agent, you know, you know, this, this guy agent and like, they all sort it out for you.
00:36:36.133 --> 00:36:38.813
And they let you know and say, Hey, we saw some, you know, like,
00:36:39.073 --> 00:36:41.813
you know, power cycling on your VM, we solved it.
00:36:42.313 --> 00:36:45.493
And like, you're, you know, you're, you're informed, but yes,
00:36:45.533 --> 00:36:49.793
I think we will get there. And I think we will get there faster than we think that we will.
00:36:49.973 --> 00:36:52.293
And again, given how fast everything is moving.
00:36:53.133 --> 00:36:56.573
Well, yeah, if you think about like what we were talking about earlier about,
00:36:56.733 --> 00:36:58.753
you know, what are all the processes in the contact center?
00:36:58.933 --> 00:37:03.593
Well, what are all the inquiries that are coming in that are now solvable with an agent?
00:37:03.973 --> 00:37:08.213
And then, yeah, rather than have anyone even ring the contact center or chat
00:37:08.213 --> 00:37:10.913
or anything, just have your own agent, just ping it.
00:37:11.213 --> 00:37:13.953
And just go and resolve it. So that'd be great.
00:37:14.113 --> 00:37:18.373
Microsoft could release a support agent that I can install and it just sits
00:37:18.373 --> 00:37:23.133
there and monitors everything and then things supportive it needs to. That's probably that.
00:37:24.013 --> 00:37:27.413
I would say that would be the ultimate in a similar experience. I know.
00:37:27.553 --> 00:37:32.933
That would be the pinnacle. It would be the pinnacle. And it would be a vastly superior experience.
00:37:33.113 --> 00:37:35.733
And remember when Clippy was trying to help us right over?
00:37:36.113 --> 00:37:39.973
I love Clippy. I love Clippy. I love Clippy. So yeah, yeah, yeah.
00:37:40.013 --> 00:37:44.573
I've got my team kickoff in a week, and I'm going to go on to Vivo and see if
00:37:44.573 --> 00:37:46.653
I can create a Clippy something. Yeah.
00:37:47.793 --> 00:37:50.513
That video software that Google released a few weeks ago.
00:37:51.347 --> 00:37:54.267
I know i know well yeah just speaking
00:37:54.267 --> 00:37:57.127
of the like your your personal use of you know how you
00:37:57.127 --> 00:38:00.207
are leveraging you know ai to make yourself more sophisticated you
00:38:00.207 --> 00:38:03.267
talked about your upcoming off-site so how are you guys
00:38:03.267 --> 00:38:09.187
using ai in cool ways you know i either professionally or personally yeah yeah
00:38:09.187 --> 00:38:12.867
well it's timely because we're going through it now so maybe i'll go first matt
00:38:12.867 --> 00:38:19.267
so i use it personally i use across chat grok and gemini for different things
00:38:19.267 --> 00:38:21.387
and to see you know So for most things,
00:38:21.567 --> 00:38:25.127
if I've got to ask a question or even learning, I use it a lot for learning.
00:38:25.327 --> 00:38:27.827
For work, we're using it for account planning.
00:38:28.087 --> 00:38:32.167
So researcher, which is our agent internally, go and, you know.
00:38:32.827 --> 00:38:36.227
Look at these accounts, look at the data and come back and tell us where we
00:38:36.227 --> 00:38:37.387
think the opportunities are.
00:38:37.647 --> 00:38:40.167
We use it for summarization all the time.
00:38:40.727 --> 00:38:45.987
Rewording emails is nearly every email now. I think I run Copilot over it first
00:38:45.987 --> 00:38:50.467
and just a whole bunch of admin kind of tasks and things like that.
00:38:50.627 --> 00:38:55.847
So I would say I use AI a lot for my day at the moment.
00:38:56.267 --> 00:38:59.187
And there's a huge saving in time, huge.
00:38:59.627 --> 00:39:04.007
I don't think we could do the kind of transformation that we're going through
00:39:04.007 --> 00:39:06.487
right now without it. That's for sure. Yeah, absolutely.
00:39:06.947 --> 00:39:10.507
Certainly couldn't live without it now. It's like a mobile.
00:39:10.967 --> 00:39:14.027
How about you, Matt? I've been using it in some really curious ways.
00:39:14.427 --> 00:39:17.247
I can't even remember the day we didn't have AI. I'm at that point now where
00:39:17.247 --> 00:39:21.607
I'm using it so much that I just can't fathom the fact that we didn't have this
00:39:21.607 --> 00:39:24.207
technology before. It's actually scary.
00:39:25.249 --> 00:39:28.649
Personally at home, I've been playing a lot. Did you play that Wordle game through
00:39:28.649 --> 00:39:31.149
the New York Times or something like that? I've tried to avoid it.
00:39:31.149 --> 00:39:33.129
My husband plays it. Yeah.
00:39:33.989 --> 00:39:38.809
So my girlfriend got me into it, and I'm super competitive, and I was in like a 10-day Wordle.
00:39:38.929 --> 00:39:41.909
I couldn't figure this thing out, so I screenshotted it because I didn't want
00:39:41.909 --> 00:39:43.849
to lose that day. I was up to my last turn.
00:39:43.989 --> 00:39:47.509
I put that into AI, and it gave me the final word so I could keep going.
00:39:47.809 --> 00:39:51.029
So, yeah, if she's listening, I'm sorry. I cheated.
00:39:51.589 --> 00:39:54.809
Yeah, I was going to say, that's not good. That's what you call cheating, man.
00:39:56.969 --> 00:40:00.649
I was on like a 10-day streak. I didn't want to lose the streak.
00:40:00.869 --> 00:40:04.049
I was pulling my hair out and it was ruining my day. You let the secrets out.
00:40:04.609 --> 00:40:07.729
Yeah. That's just the answer in the back of the book, you know.
00:40:09.149 --> 00:40:14.009
I'm just shocked. I am shocked at just how good the models are getting as we're going.
00:40:14.009 --> 00:40:20.709
But I'm also shocked at how quickly my technical team will grab the new technology.
00:40:20.709 --> 00:40:27.649
But basically, MCP released the next day we build a multi-agent demo and we're
00:40:27.649 --> 00:40:30.209
hitting the product team and they're going, no one is doing this.
00:40:30.269 --> 00:40:31.849
What are you doing? It's quite crazy.
00:40:32.109 --> 00:40:38.609
So like it's just running so fast to adopt the new technology and then get it out to customers.
00:40:38.609 --> 00:40:42.329
And if any partners are listening to this, I think there's a huge opportunity
00:40:42.329 --> 00:40:48.489
but to sync with what we're doing because I think the gap between what partners
00:40:48.489 --> 00:40:52.249
traditionally delivered and what we're actually delivering now is very different and Matt,
00:40:52.389 --> 00:40:55.329
you would have seen that over the last 12 months, yeah.
00:40:56.029 --> 00:40:57.369
Yeah, we're building a lot of stuff.
00:40:57.549 --> 00:41:01.609
We've kind of re-visioned how we were designing solutions in the past.
00:41:01.789 --> 00:41:04.409
So yeah, it's a really exciting time and like you said before,
00:41:04.509 --> 00:41:05.749
Jen, there's so much happening at the moment.
00:41:05.889 --> 00:41:09.069
You're just consistently learning and evolving and changing because this AI
00:41:09.069 --> 00:41:14.969
stuff is changing by the weeks almost, not even months, it's weeks. So it is super cool.
00:41:15.229 --> 00:41:19.069
So before we sort of wrap up, oh, sorry, you go. No, the thing I was going to
00:41:19.069 --> 00:41:22.669
tell you is my personally, just kind of a plus one to your point. Yeah, yeah.
00:41:23.143 --> 00:41:25.863
Of you know all that stuff is moving so fast and it's nothing
00:41:25.863 --> 00:41:28.943
that you have to compile right it's all low code no code
00:41:28.943 --> 00:41:31.643
you know clickable user enemies vibe coding is
00:41:31.643 --> 00:41:36.443
huge at the moment everyone vibe coding yeah and
00:41:36.443 --> 00:41:41.003
so the point of that like the vibe coding but the microsoft researcher so let
00:41:41.003 --> 00:41:46.663
me tell you you know my recent example you know one thing that we need to do
00:41:46.663 --> 00:41:50.923
to make sure that you know our businesses are solid is we need that red teaming
00:41:50.923 --> 00:41:53.283
kind of persona people that will tell us,
00:41:53.423 --> 00:41:57.183
here is why your business venture is going to fail.
00:41:57.623 --> 00:42:02.843
And so I did exactly that. So I took Microsoft Researcher and,
00:42:02.843 --> 00:42:08.643
you know, I shared with it some, you know, business concepts that we've been working hard on.
00:42:08.703 --> 00:42:12.983
And these are high visibility projects that we want to make sure that we are successful at.
00:42:13.323 --> 00:42:18.163
And I told it to go through and tell me all the reasons for why it will fail.
00:42:18.858 --> 00:42:21.978
And how to build out a mitigation plan. Because if you think about this,
00:42:22.178 --> 00:42:26.658
if you sent that, if you brought in somebody to do that, for you to act as a
00:42:26.658 --> 00:42:29.978
red team, to act as that devil's advocate, you'd hate them, right?
00:42:30.058 --> 00:42:31.898
Because they're basically tearing apart your idea.
00:42:32.718 --> 00:42:37.538
But having this researcher do that for you, it saved me two weeks.
00:42:38.078 --> 00:42:44.478
And it's just amazing. And then had it turned it into a podcast so I could listen
00:42:44.478 --> 00:42:48.218
to it while I was walking to pick up my daughter from camp. So it's just,
00:42:48.218 --> 00:42:50.238
it's incredible what it's doing now.
00:42:51.278 --> 00:42:56.558
That's great. Well, on the topic of podcasts, you've recently launched one as
00:42:56.558 --> 00:42:59.098
well. Do you want to talk to me a bit about that?
00:42:59.538 --> 00:43:02.958
Yeah, yeah, yeah. Thank you. Yeah, thanks for the shout out.
00:43:03.078 --> 00:43:04.858
Yeah, it's been, you guys inspired me.
00:43:05.138 --> 00:43:09.358
So, you know, one of my very dear friends who's got three decades in personal
00:43:09.358 --> 00:43:14.278
finance, and a few years ago I wrote a book on finance, and I started it on
00:43:14.278 --> 00:43:17.578
basically how to be money well, how to be healthy in personal finances.
00:43:18.018 --> 00:43:19.578
So, yeah, thanks for the shout-out to it.
00:43:20.078 --> 00:43:25.298
That's awesome. We'll put a link in the show notes so people can listen in and into the podcast.
00:43:25.538 --> 00:43:28.398
I'd love to come on and learn a thing or two and be picked apart.
00:43:28.558 --> 00:43:31.098
So please invite me on. I'm sure it'll be great to see you. What about you?
00:43:31.158 --> 00:43:32.918
We'll talk about your meme coins.
00:43:35.318 --> 00:43:38.098
I'll tell you what my crypto portfolio is by the time I come on.
00:43:38.178 --> 00:43:41.918
Hopefully millions, but we'll see. It's a bit volatile, as we know.
00:43:46.038 --> 00:43:48.898
That was an awesome chat thanks again for coming on
00:43:48.898 --> 00:43:52.478
hopefully in a year's time we can come back for another update but
00:43:52.478 --> 00:43:55.378
any final words how can people get in
00:43:55.378 --> 00:43:58.298
contact if they need to yeah any any final
00:43:58.298 --> 00:44:02.838
words yeah so so one you know we are learning we're moving fast um if anybody
00:44:02.838 --> 00:44:08.438
has any questions for me you can reach out at jelowry at microsoft.com and i
00:44:08.438 --> 00:44:12.918
think scott and that we'll also post my LinkedIn profile so yeah we will do
00:44:12.918 --> 00:44:16.958
that awesome it was fantastic catching up with you and we always enjoy the conversation,
00:44:18.093 --> 00:44:20.713
We'll keep on to next time. Cheers. G'day.
00:44:24.533 --> 00:44:32.053
Thanks for tuning in to PowerTime, your source for practical insights across Microsoft AI.