Latest / The Prompt / Ep. 557 | How to harness AI decisioning without losing trust or control

Transcript

Ep. 557 | How to harness AI decisioning without losing trust or control · The Prompt

0:01
This is on base your one stop shop for solving your most
0:04
pressing B2B go to market challenges.
0:07
Each week, host Chris Moody and Paul Gibson talk to sales and
0:09
marketing leaders to get you in depth insights and creative
0:12
strategies to set your revenue teams up for success.
0:18
Hey, everyone. Welcome back to the Home Base
0:19
Podcast. I'm your host, Chris Moody, and
0:22
today I'm very excited to be joined by John Moran from SAS.
0:25
And we're talking about reimagining decisioning in the
0:28
AI era. John, thanks for joining us
0:30
today and welcome to the show. Yeah.
0:32
Thanks, Chris. Thanks so much for having me on.
0:34
Super excited to talk about one of my favorite topics,
0:37
decisioning in the era of AI. Yeah, I am excited.
0:41
And I realized now we're so busy with our virtual meetings, but
0:45
you and I are only about 22 minutes away from each other as
0:49
we're talking. So next time we'll have to get
0:51
an in person one set up. For sure, for sure.
0:54
Yeah, it's, it's awesome that we've shared some educational
0:57
credentials and overlaps and careers and whatnot.
1:01
But yeah, so happy to be connected.
1:03
Awesome. Yeah, me too.
1:04
Well, let's start there. We always like to understand the
1:07
journey. It's never linear, usually.
1:09
But what led you to what you're doing today?
1:12
Yeah. So I've been in the martech
1:14
space my whole career. Went to undergraduate at North
1:17
Carolina State University in Raleigh, came out of there,
1:22
started working for a company by the name of Teradata, which does
1:25
analytics and data warehousing. And then from there I've moved
1:28
on through several companies and actually landed at SAS.
1:32
So I've been in the Raleigh area my whole career.
1:35
I'm married, I have a wife and two kids, and we live South of
1:40
downtown Raleigh in the in the Apex area.
1:42
So for about a dozen years now, I've been with SAS.
1:45
SAS is most commonly known for being a pioneer in analytics,
1:49
and SAS is actually celebrating its 50th birthday this year.
1:53
But SAS also has a lot of solution lines, including a big
1:56
martech line of business, which is known in the in the market as
2:00
SAS Customer Intelligence. I'm responsible for the
2:04
marketing and advertising work that surrounds that solution,
2:07
doing all of the things you might imagine, right?
2:10
Webinars, podcast messaging, writing interviews, presenting
2:14
that sort of thing. So yeah, a little bit about me.
2:17
Yeah, that's awesome. And for anyone who doesn't know
2:19
SAS, look it out. Such a cool story.
2:22
And being in our backyard is really advantageous too.
2:25
But I have friends, you know, some folks early on at SAS were
2:29
at GameStop, but understood how to work and data and then jump
2:32
in. Just some really cool stuff
2:34
that's in the background and history of SAS.
2:36
But AI decisioning, I, I'd love to talk about that more.
2:40
I think many are embracing AI somewhat blindly and just
2:44
jumping right in. But as as AI decisioning is
2:47
rising in prominence, why is it so important to strike the right
2:52
balance between the adaptive AI driven decisions and structured
2:56
governance of enterprise decision?
2:58
Yeah. So I think there's a a few main
3:01
reasons that organizations really need to negotiate that
3:03
balance before between more traditional enterprise
3:07
decisioning and these new AI decisioning initiative.
3:10
So first, for those that aren't familiar, a quick definition to
3:13
differentiate the two, right? Enterprise decisioning combines
3:17
rule based decisioning with more advanced analytics.
3:20
You know, business rules, workflow automation, you know it
3:23
it relies on those predefined rules and conditional kind of if
3:27
then logic and other behavioral and customer based conditions to
3:31
guide the decision making. Often these rules are are
3:35
simple, they can be complex, but they usually cover areas like
3:38
approvals, work flows, and what we what we call in martech next
3:43
best offer or next best action. Enterprise decisioning
3:47
centralizes control and oversight, making it possible
3:50
to, you know, define, test, refine organizational decision
3:55
models. Whereas AI decisioning, which is
3:58
kind of a a newer technology, uses artificial intelligence to
4:02
make automated business decisions without as many
4:05
guardrails as what enterprise decisioning might have.
4:08
So it it follows A progression from analytics to artificial
4:12
intelligence to machine learning, then down into
4:15
reinforcement learning using large data sets, both historical
4:19
and current data to inform decisions.
4:22
So reinforcement learning uses these algorithms that
4:26
essentially learn from from data, right?
4:29
So the idea is that adaptive decisions can improve outcomes
4:35
over time and humans are very minimally involved.
4:40
And because of that analytical path, these systems are used to
4:44
predict future outcomes and behaviors, such as likelihood of
4:48
making a purchase or cancelling a service.
4:50
And because AI is involved, a natural characteristic is that
4:54
decisions can be made automatically without needing
4:57
that constant human oversight that enterprise decisioning
5:02
might need. So we talked a lot about a
5:04
balance, right? And I think for businesses,
5:06
there is a balance between enterprise decisioning and AI
5:09
decisioning. But you know, ultimately whether
5:12
the scale is tipped in One Direction or the other depends
5:15
on several factors. Right now, I would say the first
5:19
factor is really risk tolerance, right?
5:22
A lack of balance can expose the enterprise to risk such as
5:27
regulatory penalties, reputational damage, unintended
5:31
consequences of autonomous decision making.
5:35
So structured oversight, as you know, or really acts as a safety
5:40
net, ensuring that AI operates within clearly defined bounds,
5:45
right? And so AI decisioning, it
5:48
delivers responsiveness and a competitive edge while, you
5:51
know, providing the governance and the guardrails that are
5:54
needed to use AI responsibly and sustainably at scale.
6:00
I think that, you know, organizations that balance both
6:02
will drive better outcomes, right?
6:05
Faster, safer and better decisions aligned with strategic
6:09
goals. So risk is is the first thing
6:12
that should be should be accounted for risk tolerance.
6:15
I would say the second thing is really this appetite for agility
6:19
versus accountability, right? And so AI decisioning brings
6:24
speed and adaptability, you know, allowing enterprises to
6:28
respond in real time to to data and changing customer behaviors
6:32
and conditions. However, you know, without the
6:35
level of structured governance that enterprise decisioning
6:38
typically has, agility can lead to inconsistent decisions if
6:44
you're not careful, compliance breaches, ethical lapses, that
6:48
sort of thing. So you got to make sure the
6:51
governance is there to ensure that accountability,
6:54
auditability is very important as well and regulatory
6:58
alignment. And so, you know, adaptive AI,
7:01
which is, you know, what AI decisioning is, encourages
7:05
experimentation and learning from data, which fosters, you
7:08
know, more rapid innovation. But ungoverned decisioning can
7:12
introduce risks around bias, opacity and black box decisions
7:18
or conflicting outcomes across business units.
7:21
So a governance framework is, is really important there, which
7:24
we'll we'll probably talk about a little bit later.
7:27
And then finally, I would say, you know, organizations have to
7:30
look at balancing trust with brand integrity and standards,
7:35
right? So, so poorly governed AI
7:37
decisions, especially those in involving certain topic areas in
7:42
marketing and advertising, personalization, pricing, that
7:47
sort of thing, can, you know, damage trust if they appear
7:50
unfair, erratic or invasive in nature.
7:54
So enterprise governance enables scalable deployment of AI by
7:59
providing reusable policies, data, standards and those sorts
8:03
of things. So, you know, those are a couple
8:05
things that I would really consider if I were an
8:07
organization, right? What what's the, what's the risk
8:09
tolerance? How agile do you feel that the
8:11
organization can be and how much do you want to balance trust
8:16
with kind of brand standards, brand integrity, that sort of
8:18
thing? That was a great breakdown.
8:21
I love the three things you had on there and the explanation.
8:24
I think you covered a little bit of the next question, which is
8:27
extremely helpful. The black box comment.
8:29
I was smiling when you said that.
8:31
I mean, you never want a seller to say why did you pick this
8:34
account and then not be able to answer the question.
8:36
Just a hi jazz fingers. That's what I've always said
8:39
about that one. We actually did a video shoot
8:41
several years ago, which you will know where this is,
8:43
Dorothea Dix Park, and it was me smashing a black box.
8:47
So we recreated the office space scenes.
8:50
Just with you being local, you know exactly where we were
8:54
before the park opened and all of that, the new park.
8:56
But I'd love to understand as everything has changed.
9:00
We're moving from if then decisioning that still exists,
9:03
right? But also the AI decisioning, if
9:06
we think about customer experience, if we're the end
9:09
user, if we're the customer, how is that balance between
9:13
enterprise decisioning and AI decisioning changing an area of
9:17
like customer experience? Yeah.
9:19
So right now enterprise decisioning or this if then
9:23
rules based decisioning and AI decisioning, right.
9:26
Right now I think they serve two different functions within an
9:29
organization. The fact of the matter is you
9:31
can't really have one without the other.
9:34
AI decisioning and Enterprise decisioning and rules based
9:38
decisioning are eventually going to merge.
9:41
You know it's possible in the future, but it hasn't fully
9:43
happened yet. Could enterprise decisioning act
9:46
as a governing body for AI decisioning?
9:50
Absolutely. I think that the key difference
9:52
lies in how each approach uses technology and, and data, right?
9:59
And so I think that, you know, as these two come more and more
10:03
together, right, we're, we're going to see evolution occur,
10:07
right? Both enterprise decision and AI
10:10
decision are moving to being more dynamic.
10:12
They're both integrating real time data, more contextual
10:16
inputs, predictive signals to guide those decisions
10:20
dynamically. Both are becoming more
10:22
centralized and interoperable, working across marketing,
10:25
operations, service and risk domains.
10:28
I think this unification is going to help deliver more
10:32
seamless and consistent customer journeys and experiences, right.
10:36
And finally, I think both are becoming more well governed,
10:40
providing transparency, traceability, and, and we know
10:45
that these are essential for regulatory compliance,
10:48
auditability and risk management, especially in highly
10:51
regulated industries, which is where SAS often often play.
10:56
So you know, what we're seeing is that right now brands want to
10:59
have both, right? They've got that enterprise
11:01
decisioning kind of framework in place.
11:03
They've got a library of rules that they're using, but they're
11:06
thinking about, if they haven't already, you know, deployed, how
11:09
can I integrate AI decisioning to take, you know, the adaptive
11:13
decision making making that I have as an organization even
11:16
further. And so I think with, you know,
11:18
regards to CX, right, enterprise decisioning puts that structure
11:22
in place, ensures a customer's treated consistently across
11:25
channels, receives offers aligned with corporate policies
11:29
and con and constraints and that sort of thing and interacts
11:33
within governed boundaries. And then AI decisioning right
11:36
now is acting as that layer or that flavor on top of enterprise
11:40
decisioning that refines the moment, selects the next best
11:43
experience or interaction based on behaviors, preferences and
11:48
pattern. So that's, that's how I I really
11:49
see those coming together to impact the customer experience.
11:54
I love that. I think it's such an opportunity
11:56
to be more proactive too. We've covered this on a podcast
12:00
before, and it was more about pricing and packaging.
12:03
And you know, someone was talking about Internet services
12:06
and it's, hey, you usually only call when something's wrong or
12:10
when your price goes up. But what if we knew you've got 4
12:14
kids, you probably have 50 devices connected, you have a
12:17
movie room beside you, your kids play Fortnite.
12:20
Your needs are very different. So what if we could look at all
12:23
that data and then proactively make you an offer?
12:25
So I think that's the part for us as customers too, not just
12:29
where we work. There's such an opportunity to
12:32
be more tailored to what we need.
12:34
And you talked about seamless. I'm curious if we start to think
12:38
about silos and I know there are no silos at SAS or demand base.
12:43
So we don't have to worry about that where we work.
12:45
But for organizations that struggle with the siloed systems
12:49
or legacy architectures or maybe not even people talking, that's
12:53
usually the problem with sales and marketing.
12:55
They just don't talk enough, not speaking the same language.
12:58
What are some practical strategies that leaders can
13:00
employ to move towards a more seamless enterprise wide
13:04
decisioning framework? Yeah.
13:05
So Chris, this is easy to say, but hard to do.
13:09
But I'm going to say it anyways. And that is you've got to view
13:13
everything with an enterprise wide versus departmental lens,
13:17
right? I, I often talk about martech
13:20
stacks and martech ecosystems and what they look like and how
13:23
they should all be, you know, interlocked and that sort of
13:25
thing. And you know, I always put in
13:29
you've got to have kind of that enterprise decisioning layer,
13:32
right? And it's got to be across sales,
13:35
service risk that so forth and so on.
13:37
But the problem is the way that businesses and organizations
13:42
have been structured from their inception.
13:44
It's almost like they're siloed by default, right?
13:47
And so I think that when we think about, you know, the
13:51
enterprise wide decisioning layer, we have to think about
13:54
people, what are the org structures, processes, right?
13:57
What, what operational model shifts need to take place,
14:01
organizational model shifts and then technologies, right?
14:04
What, what technologies do you need to put in place that you
14:07
might not have in place to help you do decisioning from, from
14:11
truly an enterprise level, right?
14:14
And so if we break each one of those areas down, right, those
14:17
three areas of people, process and technology.
14:20
First, let's start with people, right?
14:21
So people have to be equipped to understand the value of, of
14:27
seamless versus siloed decisioning.
14:30
And if we think about it from a customer experience perspective
14:32
and that Omni channel orchestration perspective,
14:35
knowing what decisions are being delivered across departments
14:38
from a, from an interaction point of view is critical,
14:42
right? And we know that a lot of brands
14:44
don't do that today, whether it's Internet providers,
14:48
financial institutions, insurance organizations, so
14:51
forth and and and so on. So knowing what decisions are
14:55
being delivered, right. And so the success there isn't
14:58
just technical, it's really cultural as well, right?
15:01
You've got to be able to equip teams with training, clear
15:04
communication strategies and and champions to foster kind of buy
15:08
in into enterprise decision across the organization.
15:12
Emphasizing things like your long term strategic value of
15:16
knowing what was said to a customer in the contact center
15:19
versus what you just sent them an e-mail.
15:21
And you know, how far off are they or how close are they or
15:24
are they sequential in nature? So, you know, from a process
15:27
perspective, right, modernizing and phases becomes the most
15:31
logical approach to accomplishing these business
15:34
objectives. So an easy way to modernize is
15:36
to wrap decision rules in your library with, you know, services
15:40
or micro services that can start to interact with new AI based
15:45
decisioning engines, right. So that's as about as deep as
15:49
I'm going to go into architecture on on this series
15:51
as as if I want to. But and then finally, from a
15:55
technical perspective, right from a tech point of view, you
15:58
know, creating a unified decision layer that blends more
16:02
static enterprise decisioning rules, those if then rules with
16:06
AI enabled rules, that's imperative.
16:09
So if you've got a technology that can only do one but not the
16:11
other, you might want to look at either, you know, how things
16:14
start to integrate or how you modernize in the future.
16:17
So this blends those AI driven insights like predictions and
16:21
classifications with those rule based business constraints.
16:26
So the hybrid approach really, you know, enables
16:29
personalization and adaptability with the guardrails that we were
16:32
talking about earlier set by organizational policies and and
16:36
regulations. So if you start with the library
16:39
of rules that are use case specific and reusable across
16:43
departments and then prioritize high impact use cases, you can
16:46
start to demonstrate that ROI early.
16:50
And so I, I think that, you know, from a tech perspective,
16:54
tech has to be governed, right? Tech on its own is can be
16:57
dangerous, as we've seen multiple Times Now.
17:00
I think one of my most favorite examples is some of the
17:03
struggles that Waymo is having. For those who are familiar with
17:07
Waymo, look up what's happening with some of the self driving
17:09
cars on the West Coast. Anyhow, wrapping this answer up,
17:13
right? Cross functional governance,
17:15
right? This governance body includes
17:17
IT, data science, marketing, compliance, legal operations, so
17:20
forth and so on. So this team can, you know,
17:23
share decisioning rules, define them, you know, what should be
17:28
shared across departments from a decisioning perspective, What
17:31
are our risk thresholds? What are our ethical standards?
17:34
KP is that sort of thing, making sure that decisions are both
17:37
agile and accountable. And so, you know, from a data
17:41
process point of view, I think that organizations should build
17:45
or upgrade data pipelines to enable real time or near real
17:49
time data flows into this decisioning framework that we're
17:53
talking about, this hybrid decisioning framework.
17:56
Timely unified data is critical for making these these relevant
18:00
and adaptive decisions at the moment of interaction.
18:03
So that's what I would consider start looking at things from a
18:06
people processing and tech point of view.
18:09
Awesome. Yeah.
18:10
I've had some great experiences in Waymo.
18:12
I have seen the news. But it was nice not talking to
18:14
someone. Sometimes with the work, it's
18:17
just I just need to get there. I want some quiet time.
18:20
But my wife was a little angry. She was saying what if this
18:23
happened? What if this happened?
18:24
And yeah, I think 1 area that you were talking about when we
18:29
hit on ethics and policies and regulatory, there's this
18:33
tendency for a lot of us to do the Drake meme, you know, the
18:37
kind of stiff arm thing or the Heisman for those who don't know
18:41
the meme I'm talking about. And discount that, move on to
18:44
the next topic. Rosemary here at Demand Base is
18:48
really good at elevating privacy and a lot of the things we need
18:51
to pay attention to and making it a sexier topic.
18:54
But there's one thing I'd love to hit on there.
18:57
And as we think about AI decisioning, how do we make sure
19:01
that it's ethical and trustworthy, but still allowing
19:04
for innovation? And I'm sure that there's a role
19:07
governance plays in this. And to your point, I mean, just
19:10
like the Waymo example, we've probably all seen sometimes
19:14
where Grok has gone a little wild on X or Twitter and seen
19:18
some of what can happen if we don't have everything buttoned
19:23
up and figured out for all the possible scenarios.
19:25
And I'm sure everyone's seen Terminator 2.
19:27
So not to say that not having ethics and trustworthy policies
19:32
and rules and regulations is going to take us into Terminator
19:34
land, but how do we do that balance of being ethical,
19:39
scaling for innovation, and where does governance fit in?
19:43
Yeah, I think, Chris, we're kind of dating ourselves because
19:45
we're talking about Terminator 2 and Office Space, which I looked
19:48
the other day just because I was curious.
19:50
I'm ready for the Office Space remake, but Office Space was
19:53
released in like 99 or something like that.
19:56
So sad there are multiple people on our team that haven't seen it
20:00
and just they're going, what do you mean?
20:03
And I'm like the Bobs, Yeah. Yeah, it's anyways.
20:06
Yeah, governance is is critical, right?
20:09
Got to make sure that decisions that are delivered to customer
20:13
interaction channels remain ethical, trustworthy and really
20:17
governance comes in a in a few different forms, right.
20:20
So SAS recently did some research with an organization
20:23
called Coleman Parks and we looked at organizations
20:26
readiness for AI initiatives and it really highlighted some
20:29
interesting data points. So as you might imagine, 80% of
20:34
all personas inside of organizations say that they use
20:38
generative AI daily. And when I say generative AI,
20:41
I'm speaking LLMS, not synthetic data generation and and you
20:45
know, digital twins and that sort of thing.
20:46
So we see very broad usage across organizations.
20:50
But then the next question was how many of you have well
20:53
established governance frameworks in place?
20:57
7%. How many of you organizations
21:00
provide a high level of training around generative AI governance
21:04
and monitoring? 5% How many of you organizations
21:09
out there feel like you're fully prepared to comply with
21:12
generative AI regulations? 9% So we've got 8090% of people
21:19
using it, but we've got single digits when it comes to having
21:21
governance initiatives and frameworks in place.
21:24
That's obviously a very large gap.
21:27
So the gap exists between usage and actual readiness.
21:30
And what does that create? That creates concern, that
21:33
creates risk, and it ultimately creates danger, right, from an
21:37
organizational perspective. And, you know, we see that show
21:40
up when we ask the question, you know, what are your concerns
21:44
regarding the usage of generative AI and AI more
21:47
broadly in your organization? The top three concerns in that
21:51
same research report from respondents, data privacy,
21:55
security and governance, which are all very closely related to
21:58
each other. I mean, they're, they're
21:59
slightly different, but they all fall under that same umbrella.
22:02
And this is above other options for their answers, which could
22:06
be ethics, accuracy, bias, cost, trust.
22:10
So this screams enablement education and a framework that's
22:14
needed to ensure that AI and generative AI and any emerging
22:18
technology is used in, in the proper way.
22:21
And now As for what should be in that enabling framework, it's
22:26
got to have a lot of components and it's, it's going to take
22:29
time to develop and it's, it's got to be holistic in nature.
22:32
And I've got 10 main things that should be in that framework.
22:37
I'll hit on them all at a very quickly because if I go into
22:41
detail around all 10 points, we might be here till it gets dark
22:44
outside. So First off, use cases, start
22:47
with use cases and goals, right? Of AI.
22:50
What do you want AI to do and not do?
22:52
Ethical guidelines implement measures to make sure that you
22:56
can detect and mitigate bias in the data that you are using.
23:01
For AI, that is extremely important.
23:03
Privacy and data protection. The way the data is handled
23:07
ensure compliance with data protection laws and regulations,
23:11
including how it's collected, stored, used, moved, encrypted,
23:16
so forth and so on. And oh by the way, make sure
23:19
that you obtain consent from users when collecting data and
23:23
using it, right? That's why we see things like
23:26
these cookie notices on websites and what not.
23:29
Next up #4 security. Strong security measures to
23:32
protect the AI system from all unauthorized access and misuse
23:38
#5 is QA established protocols for testing and validating the
23:44
accuracy, reliability, and outputs of the AI system.
23:48
Please test it in development before you put it into
23:51
production. We've seen what happens when
23:53
that's not done. Legal and compliance.
23:56
Make sure that you're, of course, adhering to relevant
23:58
laws. Human in the loop or the human
24:01
component, you know, make sure that ethical review and impact
24:06
assessments are done. You know, have that ethics
24:09
review board in place. This is the human in the loop
24:11
component to assess potential ethical issues related to the AI
24:15
system, train people who are interfacing with it.
24:18
That's, you know, the eighth thing training education is of
24:22
course, very important oversight.
24:24
So governance and oversight. This is define the governance
24:28
structure for overseeing the development, deployment, usage,
24:32
modernization of the AI system. And then finally close fully
24:37
could close that loop, right? So create a process for
24:40
incorporating feedback and lessons learned from the, you
24:44
know, in order to enable ongoing improvement of the AI system.
24:48
So there's a lot of things that should go into a framework, but
24:52
all of those things are are slightly different but very
24:55
important. Yeah, those ten were great.
24:58
And we didn't talk about this in the lead end, but I was sued
25:01
over a blog post way back in the day and used to rank number one
25:05
for sued over a blog post. And I think many of us have a
25:09
tendency to not think about legal repercussions and some
25:15
things like that in content creation or what we do.
25:18
And you know, Sam Altman just came out this week and said
25:22
there are people that are admitting to crimes and ChatGPT
25:26
and talking about problems they have.
25:28
And all of this could be given to a lawyer like you're, you're
25:32
signing that away. So the, the folks who think that
25:35
you can get by without these policies, I, I would beg to
25:39
differ. I think we're going to see
25:41
ground breaking legal cases in the future where all this stuff
25:46
is going to come from ChatGPT, just like search history and
25:49
everything we've seen over the last 20-30 years.
25:53
It's out there. So we need to be thinking about
25:55
that. I know we have policies here.
25:57
And the first thing everyone does when you see policies or a
26:00
pop up like why are you using this is, oh, come on.
26:03
Right. Like there are arms up Tim
26:04
Robinson. If anyone keeps up with Tim
26:06
Robinson, like you sure about that kind of thing every single
26:09
time. So I, I do think we need to pay
26:11
attention to that. And I'm curious before we jump
26:14
into some of the quicker questions and close things out,
26:18
if we think about modular architectures and some of the
26:21
nimble agile ways we need to integrate AI decisioning across
26:26
enterprises. You talked about use cases as
26:29
one of your 10 things. What are your favorite use cases
26:32
that show how that approach of having composable modular
26:36
architectures will deliver real value?
26:38
Yeah. So as far as kind of, you know,
26:41
tips right for implementing composable modular
26:45
architectures, integrating AI decision in that sort of thing.
26:48
Got a few tips there and then we can kind of talk a little bit
26:51
about use cases. So start with the end of mine.
26:54
Same with the use cases. Start small architect for scale,
26:58
right. So the end is, you know, you've
27:00
got this scalable enterprise widely used AI system, but start
27:05
with simple use cases and then work over time and and put those
27:08
frameworks in place where you can grow into your big bad AI
27:13
decisioning framework. That's, you know, combined with
27:15
those enterprise decisioning libraries.
27:18
Make sure you can infuse, you know, plug and play agents,
27:21
right? So you're going to want to
27:22
obviously kind of bolt on and integrate technology over time.
27:26
So these are, you know, both internal and external models and
27:29
rules, that sort of thing. We know we haven't talked about
27:32
a lot, but AI decisioning is kind of one of the foundations
27:36
that's going to be used or the Gentic AI, which is, and that's
27:41
going to be our next big marketing topic over the next
27:45
couple years just as generative AI has been.
27:47
And then we'll get into quantum computing and all of that stuff.
27:50
But yeah, agents are going to need to be incorporated there.
27:53
So, you know, that requires an AP API first design approach,
27:58
openness, extensibility, and you've got to be able to, you
28:01
know, infuse the, the governance, right, Whether it's,
28:04
you know, internal governance systems or it's, you know,
28:07
external technology providers you're using to infuse
28:10
governance to make sure that you've got explain ability,
28:14
you've got bias mitigation, right?
28:16
Because just as you mentioned, right, other organizations might
28:19
come to your organization and say, hey, why did you deliver
28:22
that decision? And you know, this was offensive
28:25
and so forth and so on. And there could be implications
28:28
to what you're actually delivering.
28:30
And then finally, make sure you're sharing ownership for the
28:33
system across the business, right?
28:36
And we've talked about this IT, data science, engineering, so
28:38
forth and so on. So As for examples of how this
28:41
delivers value, right? Implementing AI decisioning and
28:46
evolving beyond enterprise decisioning really achieves one
28:49
of the first goals of digital transformation.
28:52
We talked a little bit about some of the analyst firms, but
28:55
really, you know, one of the analyst firms out there,
28:57
Forrester says that there are two main goals to, to digital
29:00
transformation. 1 is to elevate the customer experience and the
29:04
other one is to improve or increase operational
29:08
efficiencies, right? And so evading the customer
29:10
experience requires proactive, contextually relevant decision
29:16
offer message delivery. AI decisioning enables that,
29:19
right? So the second goal of digital
29:22
transformation is to improve organizational efficiencies.
29:25
Implementing composable modular hybrid architecture lowers key
29:30
metrics around customer service and infrastructure costs while
29:34
improving staff productivity and process efficiency.
29:39
And so, you know, some of the, the results that we've seen,
29:42
some of the early results working with brands includes,
29:45
you know, maintenance cost, major decreases in, in overall
29:49
maintenance cost of, of technology modernizing that
29:52
technical debt. We've seen a lot of examples of
29:55
technical debt failures, whether it's via, you know, airlines or,
29:59
or other industries where they're, they're really
30:02
struggling by waiting to make the decision to modernize some
30:06
of their technology. You know, reduction in, in total
30:09
cost of ownership, millions of dollars in reduction of that by
30:13
moving to a more automated and adaptive decisioning framework
30:19
improvement and marketing OPS, right, freeing up your marketers
30:22
not to have to do so much tactical work.
30:25
I mean, I, I think that you and I can both attest, you know, the
30:28
tactical and marketing can be overwhelming at times.
30:32
So being able to be strategic and think about things strategic
30:35
is, is certainly welcomed. And then finally just process
30:38
sufficiencies, right. We've seen huge process
30:40
efficiencies because of that automation and improvement of
30:45
the continual improvement of the process.
30:47
So those are just some of the the things that that we've seen.
30:51
Yeah, those are great. And the combo of the last two,
30:54
that's I think that's probably my favorite thing about where we
30:58
are in our industry. I think it the more we can
31:00
embrace AI to take care of plenty of the tactical things
31:05
and to help us with better decisions and next best actions
31:08
and all the things that we can do.
31:10
That frees up time for us to use AI and humans to be more
31:14
creative and be more strategic. And if I'm betting on anything,
31:19
if I'm ACMO and I'm betting on anything right now, it's going
31:22
to be creativity. So tying to relevance and
31:25
personalization and really breaking through and standing
31:28
out. And I think many are stuck in
31:31
that efficiency loop where it's just efficiency for efficiency
31:34
sake, which is very important. I'm not discounting that, but
31:39
with efficiency that gives us the time and the ability to be
31:44
more strategic. So that that's where I get
31:46
really excited about this. And I'm curious, this is very
31:50
related. You could hit on that too.
31:51
But how do you stay ahead today? Everything changes on a daily,
31:56
weekly basis and this space. What is your strategy for
32:00
staying ahead? Yeah, I do not claim to be the
32:03
smartest person out there, that's for sure.
32:06
And I love to find interesting takes in the space and latch on
32:10
to, you know, what thought leaders and and influencers and
32:14
even vendors are are putting out, right.
32:16
I think that some people that focus on AI decisioning are
32:20
ahead of the pack in terms of how they are thinking about the
32:24
future and I think vendors are as well.
32:26
You know, analyst firms will tell you that what vendors talk
32:30
about and where brands are, there's about a five year, three
32:34
to five year gap there, right. So there are thought leaders
32:38
within brands, but also that operate as kind of their
32:40
independent operate on their own as independent kind of agencies
32:44
and consulting firms that have a lot of knowledge in this area
32:48
and where things are going. I think we talked about Chris
32:50
Penn. Chris Penn comes to mind.
32:52
Certainly a great mind in the space, Ronald Van Loon, Bernard
32:57
Mahr, these guys are are great to just read and consume what
33:01
they have to say. And you can do that with AI,
33:04
right? Because you have more time to
33:06
read and learn, because you're not less focused on the
33:08
tactical. Yeah, perfect.
33:11
I have one more question for you before we go into true lightning
33:14
round very fast, very fast. What do you think is the single
33:19
biggest challenge in B2B marketing today against single?
33:22
You can pick one thing. What's the toughest part today?
33:26
I've still got to go with the topic that's been a challenge
33:29
for a number of years now, measuring our live via
33:33
attribution. In today's world, there are so
33:36
many marketing channels and there and there's so many
33:38
different forms of marketing. Multi touch attribution is hard
33:43
media mix modeling, incrementality testing above the
33:46
line mediums below the line mediums.
33:49
Getting that completely right is, is difficult and there's a
33:52
lot of challenges when it comes to, you know, data, external
33:56
factors, non monetary goals, you know, how do you, how do you
33:59
measure trust? How do you measure brand
34:03
awareness, those sorts of things?
34:05
They're, they're hard to measure.
34:06
So getting that right and having a a good handle on that is I
34:10
think still very hard to do in in marketing.
34:13
True, totally agree. Now I'm going to fight you on
34:16
that one. One of my favorites for that was
34:18
Argyle Social back in the day, which is out of Durham, NC.
34:22
So RIP Argyle Social but OK quick hitters.
34:26
Few things you hit on people already too.
34:28
You may have some other ones, but favorite tools or technology
34:31
that you're using today. Yeah.
34:33
So got to say this because I work for SAS, right?
34:36
SAS Customer intelligence 360, enterprise grade AI powered
34:40
decisioning and journey orchestration solution with
34:43
strong governance all built in. But other than that, I like any
34:47
tool that enables marketers like myself to get better at AI
34:51
automation and decisioning. This could be, you know, tools
34:54
like Dynamic yield, Zapier I use a lot, or even generative AI
34:58
tools like Sora. So yeah.
35:01
Awesome. What about Goodreads?
35:03
Now we're generous with reads. It can be book, blog,
35:05
newsletter, podcast, video. What are some of your favorite
35:08
sources of information? For sure.
35:10
So recent podcast that I was on that I think is a Gray one is
35:14
called Markaggy. And Leanne Dalweimer is the host
35:18
of that podcast she runs that has some great folks on.
35:21
They talk about all kinds of cool things.
35:23
So certainly recommend that. Markaggy is the name of that.
35:26
And then anything that is written or presented by Scott
35:30
Brinker. I mean this guy is doing things
35:33
from martech perspective that are light years ahead.
35:36
So I love reading his stuff and and how he just thinks about the
35:39
future of martech. That's awesome and I want to see
35:42
what he's doing next. Did you see the LinkedIn post
35:44
this week? What is it now?
35:46
Yeah, I think he is leaving HubSpot.
35:49
So I I don't know what he's going to do next.
35:52
I I read the whole post and then there were some great comments,
35:56
but he is just an incredible work.
35:58
I think he's probably going to do the same thing even if he
36:02
goes out on his own because he's built such a reputation for
36:05
that. But yeah, I mean, how many years
36:07
have we used that slide and presentations of how crazy the
36:11
martech and sells tech space is so great one there.
36:14
What about other folks who inspire you and B to B?
36:17
You hit on Christopher Penn, you talked about Scott Brinker.
36:20
You've hit on a couple others, too.
36:22
Any other folks you'd like to shout out?
36:24
Yeah, I've got, I got a few more.
36:25
I would give some shout outs here.
36:26
So Brian Velmer, he's more than a more tech mind.
36:30
He's a futurist. So if you don't follow him on
36:33
LinkedIn, he's thinking about things that he and I have
36:37
discussed that again, are, are are well ahead of where we are
36:40
currently. So I, I love to keep up with
36:42
him. Keanu Taylor at the Martech
36:45
Weekly, love talking to him recently did something with him
36:48
on the eras of martech and how they're, they're changing in the
36:52
air that we're in now, which is the autonomous era as he pens
36:56
it. He's great.
36:57
So definitely follow him. And then Greg Killstrom of the
37:00
Agile brand, he's another great one to follow.
37:03
Awesome. We have nailed it.
37:05
I'm looking at the clock. I mean we are right within the
37:08
last minute. So the very last question, most
37:11
important one of all, how can everyone get in touch with you
37:14
after this episode? Yeah, Linkedin's a great space.
37:18
Jonathan Moran is my name there on LinkedIn.
37:21
Additionally, if you want to e-mail me, feel free
37:23
jonathan.moran@sas.com. Outstanding.
37:27
Well, John, you crushed it, dude.
37:28
Thanks for joining us. Thanks so much for having me.
37:30
Thanks a lot Chris. This has been great.
37:33
Thanks for joining us for On Base.
37:35
If you enjoyed this episode, please be sure to share, rate,
37:38
review and subscribe on Spotify, Apple, YouTube, or wherever you
37:42
get your podcasts. The show is brought to you by
37:45
Demand Base, the only account based experience platform built
37:48
for sales and marketing alignment.