Tracy Kerrins, Senior Executive Vice President, Head of Technology, Wells Fargo & Company KC McClure, Chief Financial Officer, Accenture Moderator: Sheryl Estrada, Senior Writer, CFO Daily, Fortune; Co-chair, Fortune Future of Finance
Category
🤖
TechTranscript
00:00 Good afternoon, everyone.
00:02 Thank you for joining us.
00:04 Thank you so much.
00:04 We have Finance and Technology United on stage,
00:07 which is awesome.
00:08 Always.
00:09 Exactly.
00:11 So I'm going to kick things off with you, Casey.
00:13 You lead Accenture's global finance organization,
00:16 which includes strategic planning, financial reporting,
00:20 a ton of things.
00:21 Right.
00:22 What are the best use cases that you have for AI
00:26 in the finance function?
00:27 That's a great question.
00:29 We have a number of different use cases for AI.
00:31 And I think the key first thing that we do
00:34 is we look at what is the right technology for the process
00:37 that we're talking about.
00:38 So in some cases, I have a predictive analytics tool
00:42 that does revenue forecasting.
00:44 I've had that for years.
00:46 We're going to enhance that with Gen AI,
00:47 but we use that today, and that works.
00:50 I have AI machine learning-- again, so good old fashioned
00:53 AI, not necessarily Gen AI-- that we use for Treasury.
00:57 That helps us with--
00:59 we call it the Treasury AI application or platform.
01:02 And that has helped us identify $600 million of vital cash
01:06 that I can put to better use.
01:10 And then now we're looking at Gen AI for different functions.
01:14 I have a function where we've looked
01:16 at this for a controllership.
01:17 Again, we're working with clients and internal use cases
01:21 because I want to be our own best credential.
01:23 So I'm privileged that we offer this work to clients,
01:26 so we do it for myself.
01:27 So I have a unique position.
01:29 And one of the areas we're working on, for example,
01:31 is controllership.
01:33 Everybody closes their books every month and every quarter.
01:35 Doesn't sound so exciting.
01:37 But we are able to free up tens of thousands of hours annually
01:42 based on using Gen AI.
01:44 Now, that's great, and that's amazing.
01:47 But the key part is that we also have to then--
01:49 to really get the most benefit out of that cost savings
01:52 and efficiency is we're also looking at the processes
01:56 that we need to change and the organizational structures
01:59 that we need to change.
02:01 So again, it's technology around the continuum
02:06 of what you need for each tool.
02:08 That's how we're evaluating it.
02:09 And it's not just technology.
02:12 You have to bring the people along along
02:14 with the process and potentially organizational changes.
02:18 Now, AI uses large amounts of data.
02:22 How do you determine the data process in the finance world?
02:26 Yeah, so I think the first thing is all companies really
02:31 need to be focused on their digital core.
02:33 And so if I talk about us and then
02:35 what we're seeing with our clients,
02:37 so the first thing is really what's your data readiness?
02:40 What is your data readiness?
02:43 We actually have decent--
02:44 no one's ever done.
02:45 That's a continued movement, as we know.
02:50 And then with our clients, they're
02:51 at different places with their data.
02:54 So some of them have a strong digital core,
02:57 and they're getting ready to move faster.
02:59 Others now are just really reckoning with the fact
03:01 that they need to really focus more
03:04 on their data in their readiness.
03:07 So we work on that in terms of data readiness.
03:10 And then we also look at what is the capital allocation
03:15 that we have, and where do we want to put capital?
03:18 And I think that's really where we collaborate
03:20 with my technology partners, both for our clients
03:24 as well as for us in terms of how
03:26 it is that we're going to approach the different use
03:29 cases that we have.
03:31 Now, Tracy, in terms of data that's in your wheelhouse,
03:35 what is the importance of having the right data structures?
03:38 I mean, you're going to invest in AI technology,
03:41 but if you have an ancient system that isn't compatible,
03:47 how do you manage that within your team?
03:49 Well, I mean, when we're talking about artificial intelligence,
03:52 like whether it's machine learning, kind of more
03:55 kind of AI, small AI, and then you move into gen AI,
03:58 data is the most important thing.
04:00 But good data in equals good data out.
04:03 Bad data in equals bad data out, I think, as you know.
04:05 So it's really important that you're looking at your data.
04:08 And so one of the things we've been doing--
04:10 because we've been doing AI for the last 10-plus years.
04:13 We're a heavy AI shop.
04:15 Generative AI is the newer piece.
04:17 But we have really robust data and model risk governance.
04:21 And it's taking the time to do all of that analysis upfront.
04:25 Make sure you're complying to compliance, law, rule, reg.
04:29 You have ethical checks.
04:31 For us, it's like fair lending, all of that.
04:34 And making sure that the data, we
04:36 understand where it's coming from, the lineage of it,
04:38 the quality.
04:38 And that has to be proven before we'll even
04:41 approve a use case to be deployed or used in production.
04:45 And we like to go slow to go fast.
04:47 So we do a lot of piloting.
04:49 We do a lot of checks.
04:50 So based on the risk profile of the model that we're using
04:53 or the use case that we're doing, we go back
04:56 and we look at the output.
04:58 Because the other thing you have to be careful on with AI
05:00 is model, it's drift.
05:01 As you add additional information into the model,
05:04 are the decisions coming out of that model changing in a way
05:07 that you don't agree with or that weren't originally
05:09 approved?
05:10 And you have to course correct.
05:12 When it comes to generative AI specifically
05:15 in financial services, it's a highly regulated field.
05:20 So are there extra measures that you
05:23 take in regards to safety of information?
05:26 Or is it something that you've already been doing?
05:28 Well, we've layered it on top of what we're already
05:31 doing with AI.
05:32 But we have set up a generative AI council
05:34 that includes legal, risk, compliance.
05:38 Myself as the head of tech, we're all in there
05:40 and we review use cases.
05:42 And we review it for a lot of different ways.
05:44 And because generative AI is still so new
05:46 and we want to make sure that we're making the right risk
05:49 based decisions, we do a lot of piloting first.
05:53 And then again, we go and we compare results
05:55 that we thought we would get to the results that we're seeing.
05:58 And we come back and we look at that a lot.
06:00 So yeah, we have a lot of extra scrutiny that we've put on it.
06:05 We also evaluate the use cases themselves.
06:07 So not just the data, but what is the use case?
06:10 What are you trying to drive?
06:11 What models are you using?
06:13 So when you look at generative AI,
06:15 it's all about data, large language models,
06:17 and then the computing power needed
06:19 to run the large language models.
06:20 Your riskier use case scenarios are
06:22 when you need to start using some
06:24 of these external large language models
06:26 that you're not controlling all the data input into it.
06:29 And so that's a big point of discussion and review
06:33 in that generative AI group that we have that reviews everything.
06:36 Thank you, Tracy.
06:37 Casey, in terms of risk, I know in the finance function
06:41 with generative AI, there's been some kind of hesitancy
06:45 in some cases.
06:46 So I was wondering, are people still
06:48 kind of feeling out their way, finance leaders,
06:51 in terms of generative AI?
06:54 Yeah, I think that it's--
06:56 I talk to a lot of CFOs because I talk
06:59 to a lot of our client CFOs.
07:00 And there's no question that AI, gen AI,
07:04 is viewed as being very transformative in technology.
07:09 So I think everyone has embraced that.
07:12 And as I said earlier, really then
07:14 is figuring, how do I get started, and where do I go?
07:18 And with that, we have designed what we call an AI navigator
07:23 for enterprises.
07:25 And what does that do?
07:26 That helps you take a look at your different industries.
07:29 And now we're doing it by function,
07:30 so like for the finance function.
07:32 We have an AI navigator for the finance function
07:35 that goes through all the different core processes
07:37 that every finance organization would do.
07:40 And you would customize it a little bit for industry, too.
07:42 But there's some core things that all CFOs deal with.
07:45 And we really look at three things, at least to start.
07:48 What's the value case?
07:50 What then, based on that value case,
07:52 what is the AI architecture?
07:55 And then what is the AI solution?
07:57 And then, as we were just talking
07:59 about how things will get more complicated
08:01 with large language models and having to switch back and decide
08:05 which avenue do you take for each of these pilots
08:09 or points of view, we then have what
08:11 we call an AI switchboard that helps us determine,
08:14 then, how would you switch over and make sure
08:16 that you use the right solution for the problem that you have.
08:21 And so I think that is a big part
08:23 of how we're helping ourselves.
08:25 I spend a lot of time on that.
08:27 And our view is that, for me, I'm taking--
08:32 I would not take risk in terms of how I run Accenture.
08:35 But we want to be the leading company in finance,
08:39 be our own best credential, since we do this as a company.
08:42 So I am pushing kind of boundaries
08:44 with the teams that deliver the work to the clients to say,
08:48 let's see what we can do together.
08:50 Because we can all--
08:51 they have access to all of our data.
08:53 And so it's all we as me.
08:56 And so that's really very--
08:58 I feel very privileged to be able to kind of move and go
09:01 fast on Gen AI use cases, which we have dozens
09:04 that we're working on right now in the finance function.
09:08 Are there best practices in kind of measuring
09:11 the ROI of AI within--
09:15 Yeah, I think-- and again, I think it does--
09:17 there are.
09:18 And I think it really depends on the organization.
09:20 I do think that the key thing that we look at
09:23 is what is the no regrets moves?
09:27 What am I looking at right now?
09:29 What are the things that we should do?
09:31 What are the things that we should
09:32 do that will pay for themselves very quickly?
09:35 But then I call it the second box
09:37 that we look at, which is, what are the more transformative
09:41 things at scale?
09:43 So they may not be identified right now as a quick win.
09:46 But really, what I'm looking for is at scale efficiencies,
09:51 risk mitigation, cost savings.
09:54 And I think that's really where it becomes even more
09:58 interesting as we're all working through our different use
10:00 cases.
10:01 Thank you, Casey.
10:02 In regard to AI strategy, AI is moving quickly.
10:07 It can change.
10:09 So that means maybe data structures need to change.
10:12 How do you scale for growth, Tracy,
10:14 in this type of environment?
10:16 Yeah, I would say it's not--
10:18 it can change.
10:19 It will change.
10:19 It changes almost weekly.
10:20 I mean, we're seeing--
10:22 it's just an ever-evolving industry.
10:23 So from a data perspective, what we're doing
10:26 is we want to design everything fit for purpose.
10:28 And that's based off the use cases
10:30 that Casey and I have both talked about.
10:32 So when a use case comes in, you evaluate it.
10:35 And you say, here's what we think the need is.
10:37 And that could be the large language model
10:40 that you're using.
10:41 Does it require human-in-the-loop kind
10:43 of review?
10:43 Do we want to run processes in parallel
10:46 before we go and deploy things?
10:48 We're making the right risk-based decision.
10:51 We're also making the right cost-based decision.
10:53 So one of the things I mentioned before was compute.
10:56 So when you talk about these large language models,
10:58 we have to make a decision.
11:00 Is that something we want to run internally?
11:02 Or is that something that we want
11:03 to run on a public cloud, leveraging one of our partners,
11:06 in our case?
11:07 So that's really what we're watching.
11:09 Every time you move data to one of these models--
11:13 and if you're doing it not in-house,
11:14 and you're moving it to a public cloud provider,
11:17 there is a cost associated with that.
11:19 Sometimes that's needed.
11:20 But it definitely always has to be a point of consideration,
11:23 is how I would look at it.
11:25 Yeah, and maybe I'd add on one other thing,
11:27 I think, as we talk about value is you really
11:30 need to look at responsible and ethical AI,
11:33 because that really is a value element, too, of it.
11:36 And so as everyone is embracing AI, what they're also realizing
11:40 is the additional responsibility that
11:41 comes from the potential ethical implications of the way
11:45 they choose to do their AI.
11:47 And really, to unleash your value in AI,
11:50 you really need-- in full innovation,
11:52 you have to have guardrails around that,
11:55 that also serve to make sure that you
11:57 have the trust of your consumers, your employees,
12:00 and your other stakeholders.
12:02 And so at Accenture, we really start
12:06 with responsible and ethical AI.
12:08 We just named a chief responsible AI officer
12:12 this week, this position.
12:13 And we are able to help clients, because we've
12:16 done it to ourselves.
12:17 When we're global, we have 740,000 plus people.
12:21 And it is an immense, ongoing effort and task
12:26 to stay along with that.
12:29 And so we anchor to our Accenture core values
12:33 and our code of business ethics, and then
12:35 have a compliance framework, and also training
12:38 that goes along with that.
12:39 So each company has to do that, though,
12:42 themselves, to define what is their ethical framework, what
12:45 are their core values.
12:46 Because it could vary by industry.
12:48 It's different by company.
12:49 And I think everyone, to really unlock
12:51 the value and the innovation, and to do it
12:54 in a way that is taking care of people as well,
12:58 you need to make sure that that's really at the heart
13:00 of everything that you do.
13:02 I just want to mention, because we've
13:04 talked about a lot of use cases.
13:06 And you were talking about use cases even for finance.
13:09 But I would be remiss if we didn't talk about AI and even
13:11 generative AI, and how it's transforming the technology
13:14 industry itself, and the technologists that
13:17 develop the solutions, which is a big part of that.
13:19 That's something I worry about and focus on every day.
13:23 There's a lot of tools out there that help developers generate
13:26 code better, faster.
13:29 Hopefully, it reduces risk by improving
13:31 the quality of the code, making sure there's
13:33 no gaps in test scripts, test cases, et cetera.
13:37 We're also seeing the potential for value
13:40 when you're developing user stories or business
13:42 requirements, and the quality of those,
13:44 and how we can help improve the quality of those
13:46 that we get from business partners.
13:48 So for me, it's an imperative, is how
13:50 I talk about it with the team.
13:52 I say, the best engineers in the world
13:54 won't want to work here if we don't give them
13:56 these tools, because they're available in the industry,
13:59 and it makes them better.
14:00 And it allows them to do what they'd like to do well,
14:03 which is develop high-quality code and solutions.
14:06 So we have to give them those solutions.
14:08 So it's not just about technology delivering
14:10 solutions for our customers and clients, which
14:12 is, of course, top of mind.
14:14 But the way we do that better is looking at ourselves
14:16 as a technology organization, and how
14:18 we use AI and Gen AI to transform
14:20 the way we do our own work.
14:22 - Speaking-- that kind of leads to talent, in my mind.
14:27 So what type of talent are you looking for your organization
14:30 now that we have all of these advancements to keep up?
14:34 - Well, we hire a lot of quants on the model side, of course.
14:37 That's not as new.
14:38 But I mean, AI and Gen AI technology
14:39 hasn't been out there that long.
14:41 So you can go and hire talent, but what they have
14:45 is expertise maybe in a certain technology or a tool, which
14:48 is great.
14:49 But then when we put them into our environment,
14:51 they still need to understand how to deploy a solution
14:55 or deploy a solution for a use case in our environment that's
14:58 aligned to the business, and honestly,
15:00 at the scale and complexity that we operate in.
15:03 So we found a lot of success just hiring great technology
15:06 talent.
15:07 We, of course, have a robust training program.
15:10 And we're bringing up our own talent, people
15:12 who are interested in it, just good technologists.
15:14 It doesn't matter what technology you know.
15:16 We can train you on this, and you're kind of learning
15:20 with us, right?
15:21 And that's part of the fun.
15:24 We're a pretty big--
15:25 I like to think-- we're a pretty big talent draw.
15:27 Because we have so many different opportunities,
15:30 we use a lot of different technology in the space.
15:32 And our scale and complexity is sexy to kind of top talent
15:36 who like working in AI and Gen AI.
15:40 And I think on the talent, if you look at just the finance
15:42 function itself, we did a survey that
15:44 said over 70% of CFOs who've worked
15:47 on transformational efforts, of which all of what we're talking
15:50 about is clearly transformation, they
15:52 have to collaborate across three plus parts of their business.
15:57 So obviously, with technology, it's obviously a no-brainer.
16:00 That's one of them.
16:01 So you really need collaborative skills in finance and people
16:04 who like to collaborate.
16:05 And we've also shown that the CFOs and their entities who
16:08 are able to collaborate actually outperform
16:12 their peers in their industry.
16:14 So the way that I organize our team,
16:17 and I talk a lot about to my CFO peers,
16:20 is I have a strategy and enablement function, which
16:22 means I have a group of people who wake up every day
16:26 and think about what we're talking about.
16:28 And the lead of that has, I call it,
16:30 a foot in finance and a foot in technology, right?
16:33 And really just bridges that--
16:36 there's not a divide, because maybe that
16:37 was a divide a decade ago, but just is part of both worlds.
16:42 Because this is so big that you need someone who does wake up.
16:47 You need people who, in strong leadership,
16:49 with the ability to make decisions,
16:51 collaborating across your enterprise,
16:54 to really drive this change.
16:57 Because the pace of change, as we say,
16:59 will never be as slow as it is today, right?
17:01 Everything's just getting faster.
17:03 And so to do that, you have to empower a team.
17:05 And I find it highly valuable to have a strategy and enablement
17:09 function within my finance team that
17:11 serves as a key linchpin across the organization collaborating.
17:16 And it helps the team to say no, too.
17:18 Yes.
17:19 You have to focus your efforts.
17:20 Things are moving so fast, and the technology
17:22 is moving so fast.
17:23 You can't have that squirrel mentality
17:25 where everybody runs and tries to work on all the cool things.
17:28 You really need to be focused.
17:29 Pick what aligns to your strategic priorities
17:31 and your strategy, and do them well, or try to.
17:34 Right.
17:34 Well, that's a great note to end on.
17:36 Pick what aligns to your strategic priorities.
17:39 Thank you both so much for joining us here today.
17:42 Thank you.
17:42 Thanks for having us.
17:43 Thank you.
17:45 [BLANK_AUDIO]