Latest / The Diary of a CFO / The AI Danger Triangle: What Finance Leaders Must Check Before Using Any AI Tool (with Lauren Elizabeth Pearl)

Transcript

The AI Danger Triangle: What Finance Leaders Must Check Before Using Any AI Tool (with Lauren Elizabeth Pearl) · The Diary of a CFO

0:00
Today's episode is brought to you by Lineos.
0:03
If you're planning in one system, closing another,
0:05
reporting in a third, that is not a data problem.
0:08
It is a fragmentation problem. Lineos brings
0:11
it all into one line of sight, one login, one
0:15
platform AI built into the workflow. With Lineos,
0:18
your finance has never looked so good. And so
0:21
I invite you to learn more at lineos .com. That's
0:25
L -I -N -E -O -S .com. I will drop the link in
0:28
the show notes. Welcome back to the Diary of
0:31
a CFO podcast. I'm your host Wassia Kamon,
0:33
a sitting CFO with a background in accounting
0:35
and FP &A. I started this show because I believe
0:37
that the way company build and lead their finance
0:39
function determines whether they scale or fall
0:42
apart. Each week, I sit down with CFOs, CEOs,
0:45
and strategic business partners to talk about
0:47
how that actually looks like in real life, building
0:50
the right teams, getting systems and controls
0:52
in place, partnering with the business, and doing
0:55
it all hopefully without burning out. Today I'm
0:57
here with Lauren Elizabeth Pearl. Lauren Pearl
0:59
is a CFO, advisor, educator, and three -time
1:02
founder who believes finance is the most under
1:05
thought skill in entrepreneurship. and she spent
1:08
her career fixing that. A former software developer,
1:11
she has a particular knack for working with engineers,
1:13
founders, and translating finance into a language
1:16
they can actually use. Through her firm, she
1:19
has helped more than 300 great companies build
1:22
a financial foundation to scale. Currently, she
1:25
serves as Fractional Strategy CFO at North Pole
1:27
Security, where her technical background informs
1:30
a growing body of work on AI security for finance
1:33
leaders. and teaches financial modeling at NYU
1:36
Stern's Berkeley Center for Entrepreneurship.
1:39
Lauren also writes a newsletter called The Daily
1:42
CFO and co -hosts the Growth Money CFO podcast,
1:45
which had a pleasure of being a guest recently.
1:48
Welcome to the show, Lauren. Thank you, Melzia,
1:51
for having me. I'm so psyched to be on your podcast
1:54
now that you were such a wonderful guest on ours.
1:57
Oh, don't make me turn red. So glad to have you
2:06
on I'm so glad we get into you know, actually
2:09
finally have this Yeah months I'm so grateful
2:15
especially for your career right you started
2:17
as a developer So I'm curious to hear how did
2:20
you go from developer to accounting and finance?
2:24
Yeah, it was kind of a wild ride. I must say
2:27
initially like I had no idea I was going to end
2:31
up in finance. I started out, as you said, my
2:35
first job was on the tech side of things. I was
2:38
developing internal web applications for big
2:41
companies. And I was coming from a little bit
2:44
of a business background because I grew up in
2:45
a family business. So that engineering focus
2:49
was a bit of rebellion years. And I was working
2:53
on the tech side of things. And I liked that,
2:56
but I kept feeling pulled to get back into sort
2:59
of the business side. I was, uh, in a consulting
3:01
environment and I wanted to work with code and
3:04
they kept putting me in client meetings because
3:06
I was, I was good at kind of the people aspect.
3:08
And eventually, um, I got actually pulled back
3:11
into my family's business because my father wanted
3:14
to retire. And given that I was kind of feeling
3:18
a knack for the business side of things, I went
3:20
and I did it and I got there and. I had worked
3:24
there sort of on and off my entire life. But
3:26
what struck me is that things were going wrong
3:28
and I had no idea like what was happening, how
3:32
to fix it. I just felt really under qualified.
3:35
So it was then that I applied to business school
3:38
figuring like, okay, if I'm going to take this
3:40
business over from my father, I have to know
3:42
how to run a business. I'll go to business school
3:43
and I'll figure it out and I'll come back. And
3:45
when at business school, that was sort of the
3:48
moment that I started to get the inkling that
3:50
I was going to move into finance. Cause I sat
3:52
down and as a fun exercise while learning how
3:56
to read financial models, I figured I would read
3:59
my family businesses, financial statements to
4:02
see like what was going on with the business.
4:04
And I had my friend Brian sitting down with me.
4:07
He had more experience helping walk me through.
4:09
And he looked at me after looking at the statement
4:11
and said, Oh my god, Lauren, this business has
4:14
to close. And he walked me through just how much
4:18
money it was losing, how the inventory was going
4:20
out of control, all of these things. And there
4:23
was this like aha moment of I have worked alongside
4:28
this business for 27 years and I didn't know
4:33
as much about it as this person was able to for
4:36
looking at the financial statements for one minute.
4:39
And it was just this moment of like, oh my gosh,
4:44
this is where all the information is. Like if
4:46
I want to run a business, this is the language
4:48
I have to learn to speak. And so like from that
4:51
moment on, I think I started heading more towards
4:54
the finance function because I realized the power
4:57
that came from sitting with those dashboards,
5:00
sitting with the data and understanding what
5:02
was actually going on. From there, I had a couple
5:05
of other roles. I founded three companies. I
5:07
was a management consultant, you know, yada,
5:09
yada, yada. But that's really what got me started
5:12
in that finance journey. Wow. Amazing. And I
5:15
like how you said engineer was like the rebellious
5:17
thing. I wouldn't think about engineering for
5:20
being rebellious, but okay, very smart woman.
5:24
But I'm now I'm curious to hear because saying
5:27
that you grew up in a family business already
5:30
shaped. how you view the world and then you go
5:33
into engineering and then you add actual schooling
5:37
on top of it and experience. How would you say
5:40
those experiences have really helped you being
5:42
the CFO you are today? That's a great question.
5:46
I think it's complicated because that's a, you
5:49
know, that's, that's sort of a very multifaceted
5:52
place that I'm coming from. I think that coming
5:57
from a very like multi -discipline nary background
6:02
in general as a CFO gives me a very broad view.
6:06
I think I get little extra things from each aspect,
6:09
right? Like from that sort of brief stint working
6:14
on the tech side, I think I got a ton of like
6:17
first principles thinking and sort of thinking
6:20
about how things are built from the ground up.
6:23
And also in working with a lot of engineering
6:26
CEOs now, got a little bit of a shared language
6:29
that I can now use when explaining how finance
6:33
works and why it's useful. Because in fact, one
6:36
of the things that's funny about working, especially
6:38
with CEOs coming from engineering background
6:40
is finances is kind of a crazy system when you
6:44
think about it, like the way that accounting
6:45
works. It can be, it can feel really wild. for
6:49
a CEO to have financial data not be up to date
6:54
minute by minute. Like most databases, most modern
6:56
databases, you can like know the actuals immediately.
7:00
So it's very odd to have a system that you can
7:02
only know accurate data like once, like 10 days
7:06
after the month, right? So that doesn't make
7:08
any sense to most engineering CEOs. But if you
7:11
explain that like, oh, this is not just like
7:14
a normal database, this is the high assurance
7:16
system of record. then they get it, right? Because
7:19
it's just a different database type and a different
7:22
informational system. And then they can work
7:24
within that structure. Yeah, and coming from
7:26
a family business, I think it just kind of gives
7:29
that sort of ownership mentality. Like that's
7:32
something that I've spoken with other CFOs on
7:35
the podcast about how they learn to be a CFO.
7:38
And many of them, like a surprising number, have
7:41
mentioned startup experience as that moment that
7:44
They transitioned from someone in accounting,
7:47
someone in FP &A into CFO mindset because it
7:51
gives you the perspective of like running the
7:54
whole business, which is so valuable if you're
7:56
going to be the finance leader. Yeah. So I don't
7:59
know. I think, I think there's lots of advantages
8:01
to having those experiences. Oh, I can, and I
8:04
can tell how you care, right? Because you've
8:07
seen and you've seen a business, I mean, for
8:11
over 20 years, like you said, and you wanted
8:13
to succeed. at a different level, right? That's
8:18
very amazing. And also when you think about the
8:21
startup experience, for those who haven't worked
8:25
in a startup, like what can other finance leaders
8:28
or CFOs learn from that startup experience? So,
8:32
so much. I was actually super surprised when
8:36
I started hearing this trend on the Growth Minded
8:38
podcast because I kind of figured that a lot
8:43
of CFOs at larger companies would have this very
8:46
specific path, that they would start out maybe
8:51
in banking and then transition to the FP &A team
8:54
and then get promoted internally or come from
8:58
accounting. And you kind of stay at this big
9:01
company space in finance with a capital F function
9:05
and eventually become CFOs. And it was so surprising
9:10
to hear that a huge number of the CFOs we were
9:13
interviewing took a stint at a startup, either
9:16
started a company themselves, which is really
9:19
cool, or were the CFO of a smaller company. I
9:24
think there's a lot of different things that
9:25
you get. Ownership of that company, that sense
9:28
of what it means to truly be the guy holding
9:32
the bag at the end of the day is huge. I think
9:35
that's probably like the foundational piece of
9:37
it, but there's other stuff too. Like one of
9:40
the things I think is really cool about having
9:42
a startup background is getting comfortable in
9:46
very uncomfortable situations and learning how
9:49
to just like make your way through it. So as
9:51
a startup CFO, you're put in situations you would
9:55
never face as a large company CFO, right? When
9:57
you come on in day one as a startup CFO and you're
10:00
making your financial model, you have no data.
10:03
Like, none, right? You have to make it. People,
10:08
when they build financial models, will sometimes
10:11
poo -poo at a model that has too many assumptions
10:13
that are like, no, these are estimates, and these
10:16
aren't real. And in a startup, you don't have
10:18
any real numbers. You don't have real assumptions
10:21
or historicals. You just have to deal with the
10:24
best guesses you have in this moment and move
10:26
forward. So learning how to make decisions despite
10:31
that huge data gap, That's a really big deal.
10:35
And then when you're working in a larger organization,
10:38
you'll have much more data. But when you have
10:40
a gap like that, it doesn't just stop you in
10:42
your tracks because you have that experience
10:44
to say, yeah, I know we don't have perfect data
10:47
on what's going to happen in two years with this
10:49
metric, but you know what? We can kind of extrapolate
10:52
it. We can take a guess and correct it later.
10:54
We can make a bunch of scenarios based off of
10:57
this and plan around it. You know what the options
10:59
are and you don't freak out. So I'm sure there's
11:02
lots of other things, but those are the two big
11:04
ones that come to mind immediately. Yeah. Being
11:06
comfortable in uncomfortable situation, like
11:09
that's the whole summary of being a CFO these
11:11
days. Yeah. We're all uncomfortable. Yeah. So
11:17
people look up to you for direction, for clarity,
11:20
and you yourself are looking for direction and
11:23
clarity. Right. It's so true. This is like a
11:27
very weird time to be the chief risk officer
11:31
of a company because what do we have right now
11:33
other than unknown? So it's like a very relevant
11:36
time to really understand how to operate without
11:39
a lot of information and just like move the company
11:41
forward anyway. Because I really want to dive
11:46
a little deeper on the, you know, getting started.
11:49
Because I think if it feels, whether you start
11:51
up a smaller company, like, how do you get on
11:54
board it, right? Because there is usually no
11:56
reference point. There is no template to refresh.
12:00
There is no, let me take the schedule that so
12:02
and so did before me. Like how, like, what is
12:06
your routine? Especially because you're a fractional
12:08
CFO to different companies. Like, how do you
12:11
get started? Actually, the consulting work I've
12:13
done with startups has really helped in this
12:16
regard, because you do. You get a lot of when
12:19
you work with 30 companies in a year, you really
12:24
have to get good at quickly getting understanding
12:26
of the situation. In my consulting practice,
12:30
I developed a bit of a diagnostic process that
12:33
I really like to start with each time. And it's
12:37
a pretty fast paced, just like data dump and
12:42
information gathering that then results in a
12:44
report. So again, this is very consulting -y.
12:48
So I don't necessarily recommend that a full
12:50
-time CFO coming onto the role takes this exact
12:52
process, because you might get looked at a little
12:54
funny. But what I've done is I'll come in. I
12:58
ask for standard reporting, all of the financial
13:01
statements, any model that they currently have.
13:04
But I'll also ask for um, and review a lot of
13:07
documents that have to do with the operations
13:09
of the company. Like what are the processes you
13:11
have documented? Uh, what does your org chart
13:14
look like? Um, what are the large, like your,
13:18
if you are comfortable sharing like your customer
13:21
list of like who you're serving, any of your
13:23
kind of key marketing docs that help people who
13:26
are trying to understand how you go to market,
13:29
like what do you, what do you give them to ramp
13:31
up on that profile target customer? And then
13:34
also who you are to them. So kind of every document
13:37
I can get around like, who is this company? But
13:40
then the place I get the most information is
13:42
I ask to do interviews with key folks. So I want
13:45
to interview. Again, it's a startup. So advantage
13:47
of coming on board to a startup is you can talk
13:50
to five people and that might be half the company.
13:52
So that's great. So I'll try to speak to somewhere
13:55
between three to five folks and have like an
13:57
hour to two hour interview with them where I'm
13:59
asking them questions about their role, about
14:02
the company, about the stuff that's going on.
14:04
great, the stuff that's going wrong, and just
14:07
really get a view into their specific world as
14:10
an individual. From there, I put together, essentially,
14:16
this report. And the report will first talk about
14:19
what things I heard and found. I've seen a lot
14:23
of different companies, so I'm kind of pattern
14:25
matching to what stuff sounds like. You're doing
14:28
best practices. What stuff sounds like, oh, this
14:31
is really funky. What stuff sounds like, oh,
14:33
this is really interesting, like different and
14:35
good, and kind of putting that stuff into that
14:37
bucket. And then I try to come up with a thesis,
14:40
sort of a recommendation given everything that
14:42
I've heard on if I could only do three to five
14:46
things in this company, like over the next year,
14:50
what would be the most high impact stuff to focus
14:53
on for this company? And then, you know, this
14:55
is a report for consulting. So I also try to
14:57
give them like a step -by -step guide and like,
14:59
okay, if you want to do these three to five things,
15:00
here's exactly the steps you should take. So
15:02
they could take that report and work with me
15:03
or work with another consultant to get those
15:06
things done or their internal team or what have
15:08
you. I do think this is kind of an interesting
15:10
idea for folks coming on full time. I've spoken
15:12
with a couple of CFOs on the podcast who've done
15:15
something actually kind of similar, like CFO.
15:18
Carla Cooper, I think came from an investor relations
15:21
background and she wrote a prospectus on a company
15:24
when she was coming in to help her get her head
15:27
around it. So I think having enough information
15:30
and trying to use that to form an opinion is
15:33
really, really, really powerful when coming into
15:37
an org, if you can do it. But the final thing
15:39
I'll say about just context and coming into an
15:42
org that I think is also worth mentioning at
15:44
the same time is This has been really powerful
15:48
for me and it sounds like for Carla. The other
15:50
thing I keep hearing again and again from CFOs
15:53
is when coming into an organization, the importance
15:56
of patience. I've heard this from, I remember
15:59
Rita Colley was one of the folks who mentioned
16:01
that she was a regional CFO for Roche Pharmaceuticals
16:04
and would move often from region to region. And
16:06
she had talked about her way that she came on
16:08
board in any given location. And she said her
16:12
kind of lesson learned over and over again was
16:15
to wait before taking any action or making any
16:19
changes if she could, because however much she
16:22
thought she knew, something would always surprise
16:25
her. If it is just something about the local
16:28
context or the particular unit or the team, like
16:31
something was not as she expected. And so she
16:33
always made a rule to try and not take any big
16:36
actions like for a month end and to spend that
16:39
first month exclusively listening if she could.
16:42
which I think is also a counterpoint to the go
16:45
write a thesis and figure out all the things
16:47
you want to change. That is so hard though, but
16:50
it's a very fine balance because the reason you
16:54
hired as a CFO is usually because we want you
16:57
to fix something. We want to go to the next level.
16:59
There is a reason why you were hired and it wasn't
17:01
just to stand still. The pressure, whether it's
17:05
from the board or the CEO, the expectations is
17:09
for you. to do stuff but yes once you get in
17:13
I learned it in my own journey um first off I
17:16
learned that the by the time the information
17:19
comes to you it is often filtered yeah so you
17:22
don't always know you have that big picture you
17:26
don't always know all the details all the duct
17:28
tape that's around a process like the foundation
17:31
has a lot of duct tape you try to rip it off
17:34
we may all fall right there for a reason. So
17:40
I'm also curious, given your developer background,
17:44
where, you know, you get data, you get a quote,
17:46
a code, you there, how do you balance that need
17:49
for speed, but also the need to deeply understand
17:54
things before you change anything? Yeah. I think
17:58
when I come into a new space, especially in a
18:02
full -time role, I. I'm doing that analysis in
18:06
my head of like building a hypothesis. And it's
18:08
funny, actually consulting background is really
18:11
helpful for building kind of a framework around
18:13
this, like the management consulting, like big
18:16
management consulting background. Because when
18:18
you're first learning how to get into a firm,
18:21
they give you case interviews. In case interviews,
18:24
you essentially have to solve like a gigantic
18:27
global crazy problem with very little information.
18:30
And you need to kind of do it. quickly. And so
18:35
the same principles, I think, from this kind
18:38
of case interview end up being really applicable
18:41
when you're coming into a new role where you're
18:43
sort of, you get kind of the problem with like
18:47
the baseline information and immediately you
18:50
start to form a hypothesis, but it is not your
18:52
conclusion, right? So you from day one are trying
18:56
to like build the hypothesis and then you're
18:58
pulling on information and trying to gather information
19:01
or deduce information to either prove your hypothesis
19:06
or disprove your hypothesis and change it as
19:11
you go. And it means that you can get to that
19:14
solution more quickly and kind of progress in
19:18
and in and in until you've like narrowed in on
19:20
it. I think that when I'm coming into an organization,
19:23
I'm kind of thinking about things the same way.
19:26
I'm not preventing myself from having any gut
19:29
instincts around what's going on. I am trying
19:31
to guess at it and just build confidence in that
19:36
guess as I go along. And I think if you take
19:39
that skill and then you pair it with identifying
19:42
things that feel like easy, quick wins, That's
19:45
the key. So as you're going along, you've got
19:46
your hypothesis of sort of like what's going
19:48
on, where the value is, where the big problems
19:51
are. And there's going to be gigantic stuff you
19:53
can change. That's going to be huge risk. It
19:56
might be the duct tape you're ripping off, but
19:58
then there's probably other stuff that you could
20:00
both do that could be like a quick win and also
20:03
further could help confirm. like disprove your
20:06
hypothesis. So like, could help you learn more,
20:08
but also could kind of give you that feel of
20:11
like, we're taking actions, right? Something
20:13
like, oh, we're going to implement a temporary,
20:16
like let's, we're going to do a deep dive of
20:18
like one -to -ones with the whole department
20:19
for two weeks and do like an info gathering thing.
20:22
You're doing something, you're building information.
20:24
It's temporary. So if everyone hates it, it's
20:26
fine. It's two weeks of everyone's life where
20:29
they have like a one hour meeting once a week.
20:31
Or if you're going to. I'm trying to think of
20:35
another like on the fly good example. Just looking
20:37
for something that's a little that you can kind
20:40
of like fix. You're not ripping off any duct
20:42
tape, but you're maybe just like adding your
20:44
own little duct tape in a place where there's
20:46
been a leak forever that isn't permanent, that
20:50
isn't a huge lift, just makes someone's life
20:54
a little bit easier like along the way as you
20:56
plan for your big moves that you're going to
20:58
make later on when you have full context and
21:00
info and trust. Nice. Nice. I like, I like that
21:04
framework, um, um, idea because, and habit, if
21:08
I should say, because it allows you to manage
21:11
risk because that's what really the CFO comes
21:13
in to do. It's not the no person, but you, the
21:17
why. Yeah. We do this, um, because you don't
21:22
want to have, you don't want to have all these
21:24
great ideas become like go fund me campaign because
21:27
it doesn't make sense for the balance sheet.
21:29
Right? So what are some of the things that can
21:32
tell your business is actually sustainable versus
21:35
just surviving? Like how, what is your approach
21:37
to managing risk? Managing risk. You mean, so.
21:42
Sustainable so whether the company is taking
21:45
on too much risk you mean either too much risk
21:47
because you as a CFO coming in Helping them understand
21:51
what how is the business doing? Where are the
21:53
risks like where should we be careful? What are
21:56
the things we're trying to mitigate like what
21:58
how do you manage that like build your your risk
22:01
profile for a company? Great question. Now let's
22:05
take a quick break here so I can tell you more
22:07
about today's sponsor Lineos. Finance teams don't
22:10
have a data problem. They have a fragmentation
22:12
problem. Planning runs in one system, closing
22:15
another, reporting in a third. So your best people
22:18
spend their time reconciling systems instead
22:21
of running the business. Lineos brings a full
22:24
finance workflow into one line of sight. One
22:28
platform, one login with AI built into the workflow.
22:32
not bolted on top. So finance gets to be brilliant,
22:36
not just busy. And with Lineos, your finance
22:38
has never looked so good. So learn more at lineos
22:41
.com. That's L -I -N -E -O -S .com. I will drop
22:45
the link in the show notes. Now let's get back
22:47
to the episode. I, again, coming from a startup
22:51
background, I am always prioritizing, which I
22:55
think is helpful for a risk conversation, especially
22:58
in a context where you need to grow, grow, grow,
23:01
grow, grow. Um, and it's funny, actually risk
23:04
management when you're at like a big company
23:06
that just needs to sustain for years and years
23:08
and years as a really different thing than if
23:10
you're working for like a VC backed company where
23:13
we got a shoe for the moon. And if we don't hit
23:15
the moon, that's failure. So it's like, okay,
23:18
what, what risk can you take? The whole company
23:19
is a risk. So step one, I think maybe it's like
23:23
know where you are and, and, and, and what the
23:27
tolerance should be and not being like. too cautious,
23:31
just because that's your role, but instead recognizing
23:34
the prioritization of any degree of risk management
23:39
within what the company's trying to do. I think
23:42
as CFO, sometimes we can get hung up working
23:44
for a high risk company, feeling like high risk
23:48
companies are doing something like bad, like
23:50
almost like it's morally wrong, right? But it's,
23:53
it's not, it's just a strategy. It's like just
23:55
a mode and you kind of need to calibrate that
23:57
to that appropriately. So it's like one is like,
24:00
I'm always making sure I'm calibrating to like,
24:03
well, what are their goals? Cause if it's be
24:05
a billion dollar business from 10 employees or
24:09
failure, like, yeah, we got to take a lot of
24:11
risks to get there. And then I think it's about
24:14
again, prioritization, but like within the aspects
24:17
of the firm thinking about for this company.
24:21
What are the biggest risks? What does stop business?
24:24
What is something we couldn't come back from
24:27
versus what is something that's like, it's going
24:30
to be annoying. It's going to be a slog. It must
24:31
cost a lot of money, but like taking that risk
24:34
is worth it in order to get the thing that we
24:37
want to get. So you're almost kind of doing like
24:39
a risk taxonomy of the different aspects of the
24:43
business and thinking through like, where is
24:45
the like golden egg value here that we cannot
24:47
destroy? and that where we need to be like extra
24:51
careful and where are the aspects where it's
24:53
like, yeah, we can kind of push that off because
24:55
if something goes wrong there, it's not a huge
24:58
deal. So I think in both aspects, it's just prioritization
25:02
and figuring out what's important and what mode
25:04
we're operating. And it's interesting because
25:07
I'm curious to also hear because the CFO role
25:09
compared to when I started in. accounting and
25:13
finance is very different than the one today.
25:16
So I'm curious to hear what you've been hearing
25:17
from your guests on the Growth Mining CFO. Like,
25:22
what else is coming under the umbrella of the
25:24
CFO? And how can we just keep up? That's the
25:29
question we all want to answer, right? How can
25:32
we calm ourselves? I don't know. What are you
25:33
hearing? I think the two trends that it feels
25:38
like CFOs currently are feeling the most pressure
25:42
with in terms of like falling behind or trying
25:44
to keep up. I think the major one we all are
25:47
thinking about is AI and how are we going to
25:49
deal with this new technology in our organization,
25:52
which we could talk about at length. The other
25:54
one that I think is interesting that I've seen,
25:57
at least in the CFOs I've been interviewing,
26:00
is this diversification of the CFO role and what's
26:03
expected of us. I think this one I think is driven
26:10
by, there's a lot more folks that I've spoken
26:14
with who are both CFOs and COO. There's a lot
26:18
of CFOs who are running the IT team. There's
26:21
a lot of CFOs who are simply taking on way more,
26:25
right? Me too. It's just not. I think there's
26:33
this old school notion of a CFO as being kind
26:36
of like the like ultimate transformation, like
26:40
the ultimate form of an accountant, right? Like
26:42
you go up all the way from like bookkeeper to
26:44
controller to like staff accountant all the way.
26:46
And then finally your CFO. And now the CFO is
26:50
just expected to do so much play outside the
26:53
finance organization that you really need a greater
26:56
diversity of skills. Yes, it's those two. Let's
27:00
dive into the first one because it's all over
27:02
my feed AI. Yeah. How do we deal with this AI
27:07
thing? Is it going away? Is it here to stay?
27:10
I don't think it's going away. No, it's not.
27:13
I don't think it is. But I also think there is
27:16
a lot of hype around it because it's like everything
27:19
is with a sprinkle of AI. It's like ketchup and
27:21
fries. all over everything. So I'm curious to
27:25
hear from a CFO standpoint, what are your thoughts
27:28
on what AI means for businesses and also for
27:31
how the finance function will evolve eventually?
27:34
I'm actually super excited. I feel like in this
27:37
AI era, I oscillate between depression and huge
27:41
excitement when it comes to AI. I don't know
27:43
if that's also how you feel about it. I think
27:48
it's been very interesting that AI technology
27:53
or kind of innovation has been really happening
27:57
in the finance function of all the functions
27:59
because I've seen a lot of this hype coming from
28:03
different companies around like, let's vibe code
28:06
everything. And I have AI bots that are doing
28:08
my reconciliations and like, let's open -claw
28:11
yellow everything. watching that and I as a technologist
28:16
and like excited that like, oh, I can like whip
28:19
together an application without having to like,
28:22
get into code. And I can, I can think about all
28:25
these different solutions to problems that used
28:28
to be so manual as like making so many more things
28:30
automated. And at the same time, like we're the
28:33
risk management department, like why are we so
28:38
cavalier? And I've had this amazing opportunity
28:42
recently working so much again in security. I
28:45
was a security CFO before I became a fractional
28:48
CFO for a security research company and I'm in
28:52
security again. And I recently went to a couple
28:55
of different security conferences and they're
28:58
talking about AI and the vibe in the room is
29:01
not like excitement and oh my gosh, and this
29:03
is crazy. It's a bunch of engineers. So they're
29:05
all using AI to do coding, but their vibe is
29:08
Oh no, it's the apocalypse. Like they are freaking
29:10
out because with every new kind of huge emergent
29:14
technology that gets adopted, there's a whole
29:17
new attack vector that's coming. online and people
29:21
are going to get burned and they have to figure
29:24
out a whole slew of new security practices to
29:26
keep people safe when they're using this tech.
29:29
You know, the same thing happened when the internet
29:31
became a thing where people just had like, you
29:34
know, private information dangling out there
29:36
for you went to find and they were getting, you
29:38
know, hacked and attacked and, and, and it took
29:41
years to put those like security things into
29:43
place. Um, and so, you know, I think for security
29:48
or for finance professionals, we're really going
29:51
to have to contend with balancing our excitement
29:54
of all the things that we can now do with AI,
29:57
which is super exciting and very helpful, and
30:00
our education around what's risky, going back
30:04
to your risk conversation. It's hard to understand
30:09
whether working with AI is risky or not right
30:12
now. So we really have to look at that more carefully.
30:16
I agree because AI can make something bad look
30:20
so good, right? Yeah. So the kind of emails I
30:24
get, they're so slick. Like before you can, you
30:27
could tell easily that this is a scam. Like this
30:30
is, this is not real. Like the request for invoice,
30:33
request for payments, like they're becoming slicker
30:36
and slicker. And I can understand how, you know,
30:38
being a in cybersecurity space, like, okay, wow.
30:42
It means they can really deepfake a person, a
30:46
voice, so many things. And that to me, yes, is
30:49
scary. I do like to see reconciliations being
30:51
done like this, getting closer to zero -day close,
30:56
becoming a standard. That's exciting. But in
30:59
terms of security, it's scary. It is scary. It
31:03
is scary. I've been working, trying to put together
31:05
some frameworks to help other finance professionals.
31:10
understand the types of risks that could happen.
31:13
Because I feel like when we're learning about
31:16
new technology, we can build an instinct for
31:19
what's safe and what's dangerous. And the way
31:22
you build that is you have to have a sense of
31:24
the buckets of stuff. The risk that you just
31:26
described with phishing emails, I think, goes
31:29
under the bucket of just new attacks. And so
31:32
what's happening with the typical attacks that
31:35
you might find, especially like phishing attacks,
31:38
the quality is getting better because of the
31:40
AI and the volume is going up because it doesn't
31:43
take work to build more of them. And so you're
31:46
just going to see this gigantic volume, those
31:49
attacks coming in and looking really good. And
31:52
that is one attack vector. But there's like other
31:56
stuff where it's like instead there's another
31:57
attack vector around prompt injection, which
32:00
has to do with if you have an AI model that you're
32:03
using for work and it has exposure to the internet,
32:07
like it can take in information from the internet.
32:10
A prompt injection attack is a little piece of
32:14
code or even just text that an attacker will
32:17
hide in something that that AI is going to ingest,
32:20
but then will cause that AI bot to like dump
32:22
information to a website that they own so they
32:25
can steal it, right? And again, this is like
32:27
just a new kind of attack that people are going
32:31
to encounter that's not produced by AI, but it's
32:35
because they're using AI. And again, that's like
32:38
just two of them. There's so many other things
32:40
that - Oh, I want to hear all the other examples
32:42
you have. Yeah, give me more so I can be scared
32:45
after this. More examples of reasons to be scared?
32:48
Cool, cool, cool. Again, I'm still working on
32:56
an exhaustive list, but the other things that
32:58
when I am working with AI that I'm worried about,
33:02
actually, the better way I think to talk about
33:04
this is thinking about a structure that someone's
33:06
already created. Someone very clever whose name
33:08
I can't remember right now, unfortunately, came
33:10
up with this idea of the danger triangle. Again,
33:14
I even might be saying it wrong, but it's this
33:16
idea when you're using AI, when you're like building
33:19
with AI, you're working with an LLM, there's
33:21
sort of three different things that should give
33:25
you little alarm bells on like, uh -oh, I'm reaching
33:27
unsafe territory. And those three little things
33:30
are, if it can take in untrusted inputs, which
33:36
means like strangers can put stuff into the LLM,
33:40
and if it has access to the internet, that applies,
33:44
right? So that's one. The other aspect is if
33:47
it has access to private information. So it has
33:52
some data that's like very valuable that you
33:54
wouldn't want to share externally. And then the
33:57
third is Can it change stuff? And this one is
34:00
toggling between now it used to be, can it like
34:04
send data other places? Like, can it communicate
34:07
with the outside world, like send email on your
34:09
behalf? But now that OpenClaus become a thing
34:12
and agents are able to have right access, it's,
34:15
I think, widening up to like, can it like change
34:18
things on your machine or others? In this triangle,
34:21
it's like, if you have any one of them, you're
34:24
kind of okay, not too much risk. your LLM can
34:27
do two of them, you're in a risky situation.
34:30
If it can do three, if there's an attack, you
34:33
will get burned. So you're in a bad posture.
34:37
So when I'm working in my own environment, working
34:42
with cloud, I'm always thinking about those three
34:44
things in mind, knowing that I want to avoid
34:48
all three. So if I'm working with private data,
34:51
I have to make sure I'm trying to think like,
34:52
okay, I can work with private data, but it can't
34:54
have access to the internet. And it certainly,
34:56
you know, can't have like the ability to communicate
34:59
anything outside my own environment, right? or
35:02
I'm always thinking about if I'm using Claude
35:06
to build something to do with productivity, I'm
35:08
always a little worried and wary because if that
35:11
means I have to give it my data, that's one of
35:14
the areas of the trifecta. That to me has been
35:18
the most useful thing to keep in mind to prevent
35:21
from getting safe. One more dangerous scenario
35:24
that I'll mention because I think it's the other
35:26
category that's really important. is we've seen
35:30
a lot of data destruction. We hear a lot about
35:33
data leaking from AI attacks, which is scary.
35:37
But the other scary thing is if we're using OpenClaw,
35:40
OpenClaw can go ahead and do a whole bunch of
35:43
things on your behalf, right? And it doesn't
35:46
know right from wrong. So you might tell it to
35:49
go get data from some database somewhere, and
35:52
it might struggle to get the data, trying to
35:54
scan it, and then it might find a SQL injection
35:56
and just hack the database, right? We've seen
35:59
that happen. It might want to change something
36:03
about your repo and then discover that it didn't
36:05
do it right. So it wants to wipe and start over
36:07
and then delete your entire database. We've seen
36:10
that happen. So we just want to have this sense
36:13
of like AI agents are this weird combination
36:17
of like a super, super, super junior employee
36:20
who does not yet have wisdom, but they're also
36:22
an expert hacker. They can do a lot of very powerful.
36:27
That's right. They're very powerful, but not
36:29
very smart. It sounds like a very powerful toddler.
36:36
I'm a big Marvel fan, so I'm imagining the Avengers
36:40
suit of a toddler that have all the power. And
36:46
here's Jarvis. Jarvis is a toddler. Wasn't it
36:51
like the baby from The Incredibles had this issue
36:53
too? Baby Jack Jack. Yeah. Oh, yeah. I love that
37:00
show. Yeah, it was great. Yeah, because I feel
37:04
like CFOs now somehow are expected to be superheroes
37:09
at some point, so we may as well. There's a blog
37:13
post there. We're all meant to be the Incredibles,
37:15
but no one's accounting for Baby Jack. No, we're
37:20
just hoping and praying Baby Jack doesn't happen
37:22
because you don't know what you do with it, right?
37:25
Because that's always the thing with AI is how
37:28
can you tell when it's wrong? Because it can
37:30
make good very look. make bad look very good.
37:35
But also, how do you fix when it's wrong? Like
37:37
you mentioned, if it destroys your data, how
37:40
do you bring it back? Because I heard the story
37:43
about how the agent deleted the backup because
37:48
it saw that there was no first version to backup.
37:52
So hey, you probably don't need a backup. There
37:54
is no first instance. Those things are definitely
37:57
scary. But I'm curious to hear, what's the one
37:59
thing you're most excited about AI? Coolest thing
38:03
right now about AI and functions. Development
38:07
being that much closer to the user, I think is
38:11
just so interesting. Like we're looking at a
38:14
world where we are all going to have the ability
38:18
to sort of be an engineer, right? We're all going
38:21
to have this capability of seeing something we
38:24
want to have exist in the world and then bringing
38:28
it in to existence. I'm so interested in what
38:33
that world will look like, because there's just
38:36
a whole fleet of problems we won't have to deal
38:39
with. And then we have other problems to deal
38:43
with that I think we will. I think that also
38:48
it's going to really shift around what things
38:51
are valuable in someone in an organization. I
38:56
think that even though I think there's a lot
38:59
of fear, I think around AI that it's going to
39:01
take everyone's jobs and that, you know, unless
39:03
you're building with AI, you're going to be left
39:05
behind and you're going to have no value to an
39:06
organization. I think that's pretty silly. I
39:09
think that's throwing out the baby with the bathwater
39:11
because AI has the ability to do a lot of stuff
39:22
to like that 80 % marker, right? Like it can
39:25
write. But it's not engaging writing. It can
39:28
do like financial modeling, but the modeling
39:31
is a little off. It can do things to like the
39:35
degree where someone who's like pretty okay at
39:38
it can do. And it will get smarter, but in a
39:43
world where everybody has that capability. There's
39:46
going to be other stuff that's rare. Those like
39:48
spiky skills. I think there was a lot of fear
39:51
when Chachi P .T. first kind of came out that
39:53
like the writing skill was going to go away.
39:56
Like, oh, writers are dead. We don't need them
39:58
anymore. And what I've seen happen in that area
40:00
is like writers have never been more valuable
40:03
because when there's so much content that's just
40:06
mediocre, the standard for what you have to be
40:10
able to achieve as a writer to be engaging and
40:12
have people be excited about it has gone up.
40:15
And now if you have that like 1 % skill set,
40:19
you are very rare indeed and very valuable. So
40:23
yeah, I'm also interested for that shakeup. Okay.
40:28
I'm curious to hear what you think that that
40:30
skill will be for finance people. I would think
40:37
experience is going to be one of the biggest
40:40
things. I feel like in finance, you so often
40:43
get into these nuanced situations where you have
40:47
to make a decision. And there's so many factors
40:51
at play that to really get to the right call,
40:55
you're going to need so many reps, so many past
41:01
experiences, so much extra relevant information,
41:04
this kind of like judgment and AI, again, I think
41:09
it's it's sort of that 1 % like everyone can
41:12
sort of learn how to do Excel, but not everyone
41:17
is like the best, right? The person who really
41:20
gets it, people are going to be AI is going to
41:22
know the 10 tenants that you have to follow if
41:25
you're doing an M &A deal. But that doesn't mean
41:28
you're then going to be able to successfully
41:29
close a deal. Right? These things require so
41:33
much nuance and experience. And AI skills are
41:37
something that you can build weekend, right?
41:39
You can take a course on AI and suddenly be coding
41:42
quickly. And that learning curve is just going
41:44
to go shallower, right? AI companies want us
41:47
to adopt the software. So it's going to be easier
41:48
and easier and easier to catch up building with
41:52
AI and doing AI things. But everyone's focused
41:56
there. If you instead are honing your leadership
41:59
skills, your deal -making skills, those kinds
42:02
of complicated, messy problems that involve people
42:07
and nuance and numbers and all this sort of thing,
42:10
and you're getting your reps there, you're going
42:12
to be super rare. And you're of course going
42:15
to be better than AI because where is it going
42:17
to get its data on that from an individual going
42:20
through it? There aren't enough people blogging
42:21
about that for it to be good. Yes. Yes. And that
42:25
goes to what we started a conversation on when
42:27
you start a new, when you start on, you become
42:31
a new CFO at a new place. And how do you onboard
42:35
yourself? Like I can tell from speaking with
42:37
you, you've done it so well. You have a system
42:40
now. So yes. It's like definitely that experience
42:44
and judgment to know, OK, this is where I need
42:46
to to go look for. This is the these are the
42:49
questions I need to ask because AI is only as
42:52
good as what he gets, what he gets fed with.
42:55
Right. Yeah. Yeah. And it's also I feel like
42:59
it's been working with AI has been the most helpful
43:03
in areas where I either know nothing or I know
43:06
a ton. Because if I, and I'm using it for different
43:09
things in those regards, if I'm using AI in a
43:12
place where I already have expertise, I'm using
43:14
it like a junior employee. I'm using it to like
43:17
do the grunt work of the thing that I already
43:19
know how to build from all the experiences I've
43:21
had as a senior employee for years and years.
43:25
And that in that way, it's just like exciting.
43:26
And I'm trying to get to that 1 % product, but
43:29
I might be using AI's help to get there. Or if
43:34
I know nothing about an area, I'm using AI to
43:37
get to like prototype level because my standards
43:39
are so low that it doesn't matter, right? If
43:41
I'm not trying to get something production level
43:44
ready, but it does let me get to that like first
43:46
MVP version. But everywhere in the middle, it
43:50
kind of falls short, right? Like if you want
43:52
to have like a production ready software and
43:54
you don't know how to engineer something, like
43:56
AI is not going to get you to the same point
43:59
that someone who's really skilled in the area
44:01
is going to get. Drew. Good stuff ahead. I'm
44:06
curious outside of AI and all that and work,
44:09
what is your favorite thing to do outside of
44:11
work? I always like to ask my guests. Oh, well,
44:14
I can show you because I have all of these pots
44:18
behind me. My pottery is a thing that I really
44:23
love to do outside of work. You made that yourself?
44:26
Yeah. Wow. This is what I do on the weekends.
44:35
That is so neat. Wow. I started doing pottery
44:40
actually when I first broke off on my own. I
44:43
was at a full -time CFO role and I rage quit
44:47
that role and started a startup and a consulting
44:49
firm at the same time, which was very stressful.
44:52
And my partner encouraged me to take a pottery
44:56
class to deal with the stress and it is how I
44:59
survived as a CFO for hire for six years. Very
45:04
centering. No pun intended. That is so cool.
45:12
That is so cool. Yeah. I wish I knew how to do
45:16
something else with my hands, except for cooking
45:18
and eating. I mean, the two activities go together.
45:21
Yeah. I wish I could cook because I can't eat
45:24
those pots. I need something to put in them.
45:27
So you and I should have some collaborate. We
45:30
need to do some collaboration. We need another
45:34
collab. Thank you. Thank you so, so much, Lauren,
45:38
for being on the show. You're so brilliant. It's
45:41
such a joy to also connect with another lady,
45:43
powerful lady in finance. Thank you. Thank you
45:46
so much. Thank you so much for having me. This
45:49
was so much fun, Wassia. It was so, so nice to
45:52
get to chat with you. Thank you. Thank you. And
45:55
that's it for today's episode for the Diary of
45:56
a CFO podcast. I really hope you enjoyed it because
45:59
I know I did. Don't forget to leave a review
46:02
and subscribe on whatever platform you're listening
46:04
on as it helps this conversation reach other
46:06
finance leaders who need them. As always, if
46:09
you want to go deeper on anything, feel free
46:11
to reach out at AskAtTheDiaryOfACFO .com. Make
46:15
sure you follow Lauren Pearl on LinkedIn and
46:19
also listen to her podcast. As I mentioned during
46:22
our episode, I was recently a guest and I spoke
46:25
on how to build a personal brand as a senior
46:28
finance leader and how it has helped me land
46:30
my first VP role and the CFO role I'm currently
46:33
on. So thanks for tuning in. Definitely check
46:35
out the podcast, The Growth Money CF. and I'll
46:38
see you in the next one.