Customer Success Playbook Podcast S3 E66 - Jake McKee - AI Experience Design

Send us a text

The final episode of this transformative series tackles the ultimate challenge: scaling AI experiences without sacrificing empathy. Jake McKee reveals why most companies approach AI transformation backwards—focusing on tools instead of relationships, replacement instead of enhancement. This customer success playbook episode demonstrates how successful AI transformation mirrors the digital transformation of the past decade, requiring fundamental changes to business processes, not just technology adoption. McKee's framework for maintaining authentic human connections while scaling AI across enterprise environments provides practical guardrails for companies navigating the complex balance between efficiency and empathy. From addressing AI hallucinations transparently to designing trust through micro-moments, this conversation offers a roadmap for AI implementations that enhance rather than diminish human relationships.

Detailed Analysis

McKee's perspective on AI transformation represents a sophisticated understanding of organizational change management applied to emerging technology. His comparison to digital transformation provides crucial context—just as companies had to fundamentally rethink business processes when moving from analog to digital, AI transformation requires reimagining workflows, decision-making processes, and human-machine collaboration models.

The conversation reveals critical insights about trust-building in AI systems, emphasizing that trust develops through consistent micro-moments rather than singular grand gestures. This mirrors human relationship dynamics and provides a practical framework for designing AI experiences that build confidence over time. McKee's examples of internal process failures—particularly the 13-screen system requiring hours of work before allowing saves—illustrate how poor experience design destroys trust regardless of underlying functionality.

Perhaps most valuable is McKee's nuanced approach to AI transparency and hallucination management. Rather than attempting to eliminate AI limitations, he advocates for honest communication about system capabilities and uncertainties. This customer success playbook approach recognizes that users can develop healthy relationships with imperfect AI systems when expectations are properly set and limitations are communicated clearly.

The discussion also addresses the critical challenge of scaling empathetic AI across large organizations. McKee's emphasis on relationship design over feature development provides a sustainable framework for maintaining human-centric experiences even as AI implementations grow in scope and complexity. His insights about contextual AI behavior—understanding when users need speed versus thoughtful interaction—offer practical guidance for enterprise AI strategy.

Now you can interact with us directly by leaving a voice message at https://www.speakpipe.com/CustomerSuccessPlaybook

Kevin's offering

Please Like, Comment, Share and Subscribe.

You can also find the CS Playbook Podcast:
YouTube - @CustomerSuccessPlaybookPodcast
Twitter - @CS_Playbook

You can find Kevin at:
Metzgerbusiness.com - Kevin's person web site
Kevin Metzger on Linked In.

You can find Roman at:
Roman Trebon on Linked In.

Episode Transcript

Kevin Metzger (00:05):
Customer success.

Roman Trebon (00:10):
Welcome back everyone to the Customer Success
Playbook podcast.
I'm your host, Roman Reon.
Here with me as always is myco-host Kevin Metzker.
Kevin, we've made it to Fridaythe end of the week.
I know you're excited for AIFriday as we wrap up our three
part series with Jake McKee, theCommunity Guy.
You ready to go, Kev?

Kevin Metzger (00:30):
Yeah, let's get going.
It's AI Friday.
Jake's walked us through trustin AI and community driven
product builds.
Today we're gonna talk about howdo we break bake empathy and
authenticity into AI as wescale.
So Jake, when companies roll AIpilots into full production.
As, or when they plan to anyway,what guardrails do we need to

(00:53):
think about to keep theexperience empathetic instead of
robotic?
And how do we ensure that the AIapplications are gonna work,
work at scale?
In other words, how are theygonna make sure that they stay
within those guardrails?
What are the things thatcompanies should be thinking
about?

Jake McKee (01:09):
Well, like we talked about on Monday, this design
practice that I've been workingon called A IX AI AI Experience
design is really focused aroundthat sort of, you know, working
principle.
That, that the tools are notreally the end result, the end
results, the relationship we'rebuilding up with, um, our
creative and critical partner,the, the AI system.
Right.
But I think, you know, to answeryour question, I would actually

(01:31):
step back a second and.
Think about what's happeningright now, right?
Instead of thinking about justwhen we build a tool, how do we
make the tool good?
I think it's really important,um, that what we're seeing
happen right now is, is a, isthe AI business transformation,
right?
Very much like we saw, you knowwhat I.
10, 15, 20 years ago, thedigital transformation where we

(01:53):
did, we saw bookstores turn intoe-commerce, right?
And, and all what that required.
That moving away from, uh, youknow, it driven, uh, tech teams
inside companies and moving intohow do we really trans
transition the way that ourteams work, that we share
information, that we work withpartners, the way we work with
customers through much, muchmore digital means than analog

(02:15):
means.
So I think we're, we're in thisstage where we're really
thinking about how does AItransform the way we do
business?
And we have to start there,right?
Because you can't build trustand empathy if you're not really
starting from what are we reallydoing here?
And we shouldn't be anyway, justtrying to create a chat bot to
offload basic sales questionsthat we think if we put a fairly

(02:38):
scripted user experience journeyin place, that that.
Customers will get on the chatbot, they'll ask the questions,
and then they won't call us andwe'll have call deflection and
it'll be saving so much money,right?
Because we're looking at AItransformation as a way to
really rethink how we do basicbusiness processes, how we
automate a activities inside thethe company.
And that means thinking aboutinternal processes and thinking

(03:02):
about things like the, the dataethics and, and responsible
innovation, uh, the overalloperating model that we're using
around, uh.
How we create with ai, what arethose guardrails?
What are our upskilling andtraining and requirements for
approvals and that sort ofthing.
All this stuff that goes intomaking AI more successful for

(03:24):
the business as part oftransformation.
That's where we start.
Right?
And if we start there, I thinkwhat, what naturally starts to
happen is we stop thinking aboutan AI tool as a point solution
to replace a very specificsingular process.
That.
Uh, and we see this a lot rightnow where, you know, marketing
teams are saying, geez, we love,we write a lot of Facebook ad

(03:47):
copy and this would be great tojust outsource that all to JGBT
and we'll be done.
I think it changes theworkflows, I think it changes
the ability for us to create, I.
Uh, in that example, you know,various AB testing, uh, copy
that we can do a whole lot moreexperimentation, which is
awesome, right?
And, and, but that's a change tothe workflow as much as it is

(04:07):
anything.
And so back to this idea of howdo we make it more humanistic?
It is like we talked about onMonday, designing that
relationship with the AI systemand the humans that are using
it.
That is happening within thisgeneral, uh, AI transformation.
So if we're trying to replaceinternal processes or add an
ability to, to connectassociated, uh, data sources and

(04:27):
make better, more interestingconclusions for our, uh, data
analysis teams, these are thingsthat are really meant to do to,
to, to have outcomes that, thatwill help us to transfer,
translate, transform.
That's the word I was lookingfor.
Transform the business.
What do we then do with that?
Right?
So, you know, trust is, uh, uh,one of the A IX principles is,

(04:47):
is trust happens in micromoments as much as big moments,
right?
That how we build trust with,with other humans, right?
Back to this relationshipbuilding construct.
How we build trust happens.
Sure.
That we're not committing fraud,that not being violent, but of
course those big, huge things.
But also that if I say I'm gonnatake out the trash, I take out

(05:08):
the trash.
If I say, Hey, I, I keepforgetting to lock the front
door.
I'll work on it, that I actuallywork on it.
Whatever it is, you know, reallymaking sure that some of those
little moments are continuallybuilt up over time.
That when I hear from the ITteam that this process is going,
going to improve something I'mdoing internally to track my
hours or whatever it might be.

(05:29):
That it really does, and thatthey're coming to me with not
just something that they've donein a vacuum like it teams often
do, and then roll it out andsay, you gotta use it, and we're
all like, this is terrible.
My favorite example of thatbeing the 13th screen internal
process for a company that Iworked with, literally 13th
screens.
It wasn't until the 13th screen.
It took like two to three hoursto do all this work in this, but

(05:52):
it wasn't until the 13th screenthat they had a save button.
If your browser crashed on onscreen four.
You were starting over, right?
Well, of course that gotfeedback pretty quickly, but
that development team hadalready moved on.
Gotta stop what they were doing,get back and re refresh their
brains on what was going on.
Trust happens in a bunch ofdifferent ways over time.
That is all focused on.

(06:14):
Building that relationship andthat connection.
There's faith from the user thatsomebody was thinking about me,
somebody was trying to dosomething specific.
They obviously understood my, myproblem, not just, uh, their
problem.
Right Back to everybody goeshome.
Happy mantra from fromWednesday.
So I think that's a, a, a, it'sa very long-winded starting
point, but that's the startingpoint is really thinking about

(06:36):
this relationship building and,and how, as you've transformed
your business, are you reallytransforming all the parts of
the business that go into thatexperience?
You really understand theexperience, you're able to add
value, not just replace it withsome, yet another change.
I think all of us that work incorporate America are
overwhelmed by change, constantchange, constant reorgs,
constant restructures, constantnew bosses, constant, everything

(07:00):
can, that can whack our trust aswell.
So you know, that's a huge partof how we design the systems.
Less and, and, and yeah, therewas ways in the system you can
design trust.
But you know, I always like totalk about this from the very
beginning of when trust startsand that that is telling me,
Hey, I've got a new tool foryou, but I understand your
problem well enough that I knowyou're gonna be excited.
Not just frustrated that it'syet another tool just because

(07:23):
somebody got a new contract overin it.
Right?

Roman Trebon (07:26):
Well, I, I love Jake, your examples, because I
think sometimes you hearrelationship and, and you know,
we're the Customer SuccessPlaybook podcast, so a lot of
it's customer focused, right?
Like in terms of therelationship between the, the
organization and the customerand the journey mapping and all
that.
But you're so many times forgetai, any new tool internally,

(07:46):
there's still relationshipsthere too, right?
Like, uh, like you gave theexample, like the IT department
says, Hey, we've done this andgo use it.
And it's like, whoa, whoa.
Has anyone understood like how,how this impacts me, how I use
it and Right.
But as relationships, it, it'sthe same thing, right?
Like, and keep me honest here,Jake, external internal
relationships are all the same.
I mean, that those relationshipsare just as important.

(08:06):
Mm-hmm.

Jake McKee (08:07):
Agreed.
And, you know, back to this,the, to the, to the front end,
the experience piece.
So.
I'm sure that that, um, some ofthe listeners, when Kevin, when
you asked your question and theywere like, okay, cool.
Tell me all about the things Ineed to do to make a good AI
tool.
I think we know a lot of thosealready.
We're learning a whole lot moreabout those, but they're,
they're really, they're still inthis vein of relationship,

(08:27):
right?
That you're making things quickwhen I need it to be quick.
You're making things slower.
When I could use a moment toslow down.
If you're not, and we hear aboutthis one all the time, if you're
gonna hallucinate, tell me, orat least be clear about the
nomenclature for what'shappening for a project I'm
working on right now.
I did this earlier this morningwhere I asked Chad CPT for
something and it, and it gave mean unsourced thing, and I had to

(08:47):
ask, is this.
Based on anything.
Sometimes I do and it's, it isbased on something they just
didn't tell me.
Right.
And other times it's not basedon anything.
It made it up, but it's able totell me and being able to bring
some of that up earlier, youknow, telling your wife to be
home late is, is one thing, butnot showing up on time and then
just getting home late andtelling her, oops, I didn't tell
you, is a whole different story.

(09:08):
Right.
There's a lot of those littlehumanistic behaviors.
I think we can start to pullinto that.
But, but again, I, I just, I sodesperately don't wanna forget
that.
It's the context in which wehave these tools that's as
important as the tool itself.
Mm-hmm.
You know, I could have a great,uh, AI interviewing function,
uh, that collects, you know, ifyou're doing something with a
candidate, you have an AI toolthat's interviewing them as the

(09:30):
first clearinghouse of, ofpeople.
Okay, fine.
No problem.
If a, I believe that that'sactually gonna be paid attention
to, it's not gonna get turnedinto an automated transcript
that then goes through an AIfilter and no human ever sees
it.
And so what's the point of medoing this?
Really?
I.
Issue one.
Issue two is.
Help me warm up to be my bestself.
Don't just start asking mequestions and now I've got a, in

(09:52):
a robotically human voiceinstead of a robot or a human
voice.
Can't really tell who I'mtalking to, and it feels
uncomfortable.
It takes me a minute to settleinto the experience.
Gimme a few minutes to warm up.
Yes, it's those humanistictraits of I'm gonna have good
conversation.
What does that look like in reallife?

Kevin Metzger (10:07):
It's funny, I, I was doing a interview.
I had an AI interview merecently on something and it
conflated.
Some of the numbers I gave itand got it wrong.
I tried to correct it, still gotit wrong and I was like, all,
you've got it good enough.
You get to a point where from arelationship standpoint, it's
like, okay, I'm talking to anai, you're not going to

(10:29):
understand it.
Yeah.
I'm not gonna explain it 15times because you think you
understand it a certain way.
It's never really understanding.
Right.
It.
Only a predicting tool, so don'tknow what garbage bin that went
to, but, well, and one of thereasons

Jake McKee (10:44):
why, you know, when we talked on, on Wednesday about
the community driven productdevelopment work that I do,
where you bring in passionatecustomers from, from your
community base into the productof design.
Cycle itself.
One of the values of thatprocess is that you're not just
asking people once for a thing,but you're seeing over time how
they experience and learn andgrow from the thing.

(11:04):
So if I put a, you know, a newproduct in front of somebody,
the feedback that they'll giveme upfront, maybe a little
colored by what they want.
To say to me based on what theythink I want to hear as a
development team, right?
As a product manager.
And then that kind of fadesaway, but then they get excited
about using it'cause it's newand they're being invited to
this experience.
But then two weeks later theymay actually get to the point

(11:25):
of, okay, this is the realexperience.
But that took a minute, right?
That took a, a second to, tomake that connection.
And I think that that's, youknow, thinking about stuff like
that with, with AI experiencesas well, where.
The robotic human voice insteadof the robot or the human voice.
Right?
Either I need a, a veryhumanistic voice to get into
this dynamic.

(11:46):
And then what we're promptingthe, the interview questions,
how we're going about, do wesubmit them in advance and let
the AI read it to us, and thenextrapolate from there?
Or is it creating its own paths,right?
And really understanding, uh, aswe develop tools like that.
Processes, how does this fitinto the process?
How does a person feel usingthis?

(12:07):
How does the system come off tothem from an emotional
standpoint?
You know, when they're nervousand doing a job interview with
an ai and it doesn't, you know,if it started off saying, I'm
the robot, I'm here topre-filter a human will
absolutely look at this becauseI can't be trusted as my own.
Robot experience, you mightkinda laugh for a second, then
it asks a few warmup questionsand you're like, oh, this is
kind of funny.

(12:28):
A, a robot's asking me how myday goes.
That's a very differentexperience than a fake human
trying to pretend like it'sasking its own questions.
Even though you may have fed itthose questions in the beginning
or it's, it's responding just tocreate conversation with no
particular point.
Right.
Um, these are process questionsas much as they're technology
questions.
I think

Roman Trebon (12:48):
come, I think gentlemen to the end of our
three part series.
Jake, awesome stuff all week.
I appreciate all your, all yourinsights and, and for joining
the show.
Can you give us, uh, tell ouraudience where they can find you
at?
Very

Jake McKee (13:00):
simple.
So my name's Jake McKee.
My website is jake mckee.com.
Jake mckee.com/ix for, for moreinformation on the, the AI
experience, design work and, andpractice that I'm doing.
But, uh, yeah, that's it for me,jake mckee.com.
Awesome.

Roman Trebon (13:15):
Easy enough to remember.
Audience.
We will have it in our shownotes as well, so you can find
it there as well.
If you love these episodes,please subscribe.
Rate'em.
Share'em with your team.
It helps us bring actionableinsights to a broader audience.
Kev, we're back next week withmore strategies for our
audience's, customer successplaybooks.
I'm excited for what we havelined up over the next couple

(13:36):
months.
Until next time, Kevin, keep onplan.