Latest / The Diary of a CFO / A CFO Who Led Finance in 6 Countries on What AI Cannot Fix (with Johan Dovander)
Transcript
- 0:00Today's episode is brought to you by Rillion.
- 0:02They recently released their 2026 AI in Finance
- 0:05report based on the survey of 250 US finance
- 0:09leaders. It explores how finance teams are using
- 0:12AI, where they're seeing opportunities and what's
- 0:15holding adoption back. I was glad to contribute
- 0:17my perspective alongside other finance leaders.
- 0:20So if you are thinking about AI's role in your
- 0:22own finance team, download the report at Rillion
- 0:24.com slash AI report. You will find the link
- 0:28in the show notes. Let's get to it. Welcome back
- 0:30to the Diary of a CFO podcast. I'm your host
- 0:32Wassia Kamon, a sitting CFO with a background
- 0:35in accounting and FP &A. I started this show
- 0:37because I believe the way company build and lead
- 0:39their finance function determines whether they
- 0:41scale or fall apart. Each week, I sit down with
- 0:43CFOs, CEOs, and strategic business partners to
- 0:46talk about what that actually looks like in reality.
- 0:50Getting systems and controls in place, building
- 0:52the right teams, partnering with the business,
- 0:54and doing it all hopefully without burning out.
- 0:56My guest today is Johan Dovander. Johan is a
- 0:59CFO of Rillion, an AI -driven finance automation
- 1:03company. It has more than 20 years of experience
- 1:06in senior finance roles at international technology
- 1:09and SaaS companies. Previously, he served as
- 1:12group CFO and interim CEO at European restaurant
- 1:15tech company Trivac and held CFO roles at Sidegate
- 1:19and Microsoft Sweden. Throughout his career,
- 1:21he has led transformation in financial management
- 1:23and organizational development, working closely
- 1:26with leadership teams in both corporate and private
- 1:28equity environments. Drawing on experience with
- 1:31global companies and growing businesses, Joanne
- 1:34focuses on developing the finance function into
- 1:37a more strategic partner to the business. Today,
- 1:40we'll dig into how AI is changing the finance
- 1:43function, what it really takes to get value from
- 1:46it, and what Johan has learned from leading finance
- 1:48across six countries. Welcome to the show, Johan.
- 1:52Thank you. Pleasure to be here, Basir. Oh, thank
- 1:55you. Curious to hear. First off, what drew you
- 1:59into the world of finance and accounting? So
- 2:02take us back to your early career. Yeah, no,
- 2:04that's a great, great question. Thanks for asking
- 2:07it. So I started my career in the oil and gas
- 2:11industry for the lovely company BP. And in those
- 2:15early days, the first job I got landed a transformation
- 2:20project. It was a business process re -engineering.
- 2:23project is about changing business processes,
- 2:26implementing ERP systems, BI systems, et cetera.
- 2:30And very quickly into the job, I got landed the
- 2:33responsibility to implement the BI system part
- 2:36of the change business processes. And that kind
- 2:39of got me hooked on data structures, presenting
- 2:42data, driving insights. That's really how it
- 2:46all started. So working with data and business
- 2:50insights from data. That's how I started my career.
- 2:53That's like you went straight into the fire almost.
- 2:57Absolutely. And in that time, I mean, I've been
- 2:59to business school, but to be honest, was I sure
- 3:03I wanted to work in kind of business finance?
- 3:06No, I didn't. So I landed that project thrown
- 3:10into the deep end of a business process re -engineering
- 3:14project. I mean, I knew I liked numbers, but
- 3:17that kind of that role really got me more clearly
- 3:21refined on what I wanted to do. So after I'd
- 3:23done that role is, okay, I think this is kind
- 3:26of where I want to try to focus my career. But
- 3:29having said that, you know, my career has spanned
- 3:32different areas working from finance at the core,
- 3:37but then with other functions in the past 20.
- 3:40Wow. And you stepped as interim CEO twice. So
- 3:44I'm curious to see how that changed, how you
- 3:49saw the business since you were in that chair.
- 3:52You were in that chair. Yeah. No, it's a great
- 3:54question. I think that as a CFO, you have a great
- 3:59partner to bounce ideas off with your CEO, even
- 4:02though that you kind of responsibility for your
- 4:04function. I think what changed it for me stepping
- 4:06into the interim CEO role twice is that the buck
- 4:10stops with you. The accountability stops with
- 4:14you. And it's no longer about, you know, and
- 4:17I'm not saying that's the role of the CFO, making
- 4:19recommendations, driving decision support, you
- 4:21have to make a lot of decisions in the CFO role
- 4:23as well. But there is there's nobody else behind
- 4:26you in the CEO role. And you can have great conversations
- 4:30with the chairman of the board. But somehow,
- 4:33you're the most knowledgeable to kind of set
- 4:35a strategy, drive the whole people agenda for
- 4:37the company. And you need to be kind of comfortable
- 4:41with that accountability. So I think that was
- 4:43a huge, huge difference compared to being in
- 4:46the CFO role. And it kind of makes you also value
- 4:49the role of CFO a lot more, having had that experience.
- 4:53Yes. So what was it like to actually step into
- 4:56the CEO seat versus stepping into the CFO seat?
- 5:00Like, what was your first three months, let's
- 5:04say, when you stepped into each role? I think,
- 5:06to be honest, at first, scary, you know, I'll
- 5:11be brutally honest, scary in both of the situations.
- 5:14I was kind of in the leadership team already
- 5:16asked to see you for working with my peers. And
- 5:19when you're asked to step up, all of a sudden
- 5:22you're supposed to manage your peers. So, you
- 5:24know, how will they think about me having kind
- 5:27of been on part of the same team for a couple
- 5:30of years? So it's like, you know, a lot of things
- 5:32kind of going through your head, even though
- 5:35in both situations, I've worked with a lot of
- 5:37the individuals at least for two years, you know,
- 5:40so you kind of knew a lot of the environment.
- 5:43you came into. The other thing that was kind
- 5:47of new from both of my CFO roles is, you know,
- 5:51in the role of CEO, there will be a lot more
- 5:54relationship building with customers and partners
- 5:56that, you know, yes, I've met a few customers
- 6:00and partners in my role as CFO, but here it's
- 6:02kind of at a different level. You're expected
- 6:04to perform at that different level. So I think
- 6:07working through that, but also being humble of
- 6:10thinking, OK, where can I make an impact? and
- 6:13where do I better leave the table, leave the
- 6:16team who are more experienced in this to manage
- 6:20it. So, kind of gauging those different scenarios
- 6:24and then really thinking about what is the tone
- 6:28I'm setting with the right of. with the organization,
- 6:32because as a CEO, you're part of the same leadership
- 6:35team as everybody else. You have to think about
- 6:40what messages you want to send from a leadership
- 6:42standpoint. But when you are the CEO, you need
- 6:46to set that whole tone. And you need to be considerate
- 6:50and cautious about how you communicate internally
- 6:53and externally, about goals, about people, about
- 6:57organizations. So a little bit more... careful,
- 7:00I would say, and more considerate in how you
- 7:03drive and communicate things. If that, let's
- 7:08say, if you're driving a business transformation,
- 7:12a cost reduction, think about the impact on the
- 7:15people and the organization in a broader context
- 7:17than from the CFO chair. Those were a couple
- 7:21of things that I worked through anyway in both
- 7:24of those situations. Thanks for sharing. How
- 7:27would you say having stepped into the interim
- 7:30CEO role changed the way you now partner as a
- 7:34CFO with the CEO today? Yeah, great question.
- 7:38I think, you know, it's rarely the numbers themselves
- 7:42that keep CEOs awake at night, but rather the
- 7:46kind of the impact that those numbers will have
- 7:49over the organization. And what I think, I hope,
- 7:53I do better now than I did 10 years ago, is thinking
- 7:56through those scenarios. What would the CEO have
- 8:00to work through? What decisions will he or she
- 8:03have to make? And when making recommendations
- 8:07that they're maybe not so black and white, because
- 8:10the world is never that black and white, that
- 8:12we like maybe things square in finance. So when
- 8:17making recommendations about a particular org
- 8:20change, for example, that really partnering with
- 8:24the CEO on those issues instead of just simply
- 8:27saying, okay, we just need to make a cost reduction
- 8:30of X. And here are some broader thoughts, really
- 8:34being that business partner in terms of organizational
- 8:37change and setup is something I think I do a
- 8:41lot more today than I would have 10 years ago.
- 8:43Yeah, and it goes with what you said earlier
- 8:47about you have to be more considerate because
- 8:48you have to talk to so many partners about this
- 8:51one thing, this one initiative, and make sure
- 8:55you deliver to those audiences the right message
- 8:57and the right tone as a CEO. So now it makes
- 9:01another CFO that you're now thinking broader.
- 9:03Your CEO is probably so grateful to have you
- 9:05on board with that perspective. a lot of pressure
- 9:11on CFOs to use AI, especially from their CEOs
- 9:15and boards. What do you wish more people understood
- 9:18about AI in finance? Good. That there are no
- 9:23shortcuts is what I'd love for people to understand
- 9:27more. You can put AI on top of an existing business
- 9:32process, whether that be an AP or AR or forecasting.
- 9:37but really the true value you're going to get
- 9:40from challenging your existing processes. I mean
- 9:44I mentioned I started my career in business process
- 9:46redesign and business process re -engineering
- 9:48and working with AI it's kind of the same thing
- 9:51with a different tool set that yes you will drive
- 9:55some sort of impact, positive impact with applying
- 9:58AI tools on top. But to drive full impact, and
- 10:02if you kind of want to take, for example, you
- 10:04want to drive better value, you want to take
- 10:05cost out whatever from transformation, you need
- 10:08to do some more work. and working through your
- 10:12processes. To take one example, so in the Rillian,
- 10:16we're focused on AP automation, and now part
- 10:19of that process starts with capturing the invoice,
- 10:23an AP invoice, and you can capture it all right
- 10:26with AI, and you can drive coding from AI, which
- 10:30we do with our application, but if you've got
- 10:33a very complicated setup with lots of different
- 10:37coding across lots of different dimensions, Yeah,
- 10:40AI can help to some extent, but it also find
- 10:44it hard to get things right. So challenging yourself
- 10:48and your business processes is going to drive
- 10:50a lot more meaningful impact with AI. I wish
- 10:54people knew that more. Everything is not so simple
- 10:58just because you've got an AI tool. Yeah, there
- 11:00is no shortcut with AI. That's a great line.
- 11:04So why are you seeing AI have the most practical
- 11:07impact inside finance? Yeah, I see it in a couple
- 11:11of different areas. Where we see meaningful impact
- 11:14is a lot around the transactional processing
- 11:17element where you have relatively large volumes.
- 11:20such an AP in AR and where decision -making is
- 11:24sometimes more repetitive. There I see a lot
- 11:29of a meaningful impact. That being, for example,
- 11:32capture coding. I see a lot of impact in reconsolations,
- 11:36for example, bank reconsolation, account reconsolation,
- 11:38lots of value that you can drive there. Also
- 11:41kind of drafting your first management report,
- 11:44doing cost variance analysis. You can see a lot
- 11:47of meaningful initial impact. So that's to name
- 11:49a few. different areas where I think all of today
- 11:52we do see a lot of impact from AI. Okay. And
- 11:55I know really I recently surveyed like 250 CFOs
- 11:59about AI in finance. What was the finding that
- 12:02surprised you the most and why? Yeah, no, that's
- 12:05a great question. We surveyed 250 CFOs across
- 12:09the US. We also did a similar survey in Europe.
- 12:15Can I name more than one? Okay. Yeah, of course.
- 12:18Go ahead. I think that that stuck out to me.
- 12:23One of the things that we asked them is, you
- 12:26know, what are the things that you're concerned
- 12:27about, about AI? Yes, we also asked them how
- 12:31many people, you know, are you using it a lot?
- 12:33And no surprises there. A lot of people do use
- 12:35it. And also people trust AI. I didn't don't
- 12:39think that was a huge surprise, but a couple
- 12:41of things that surprised me. So a lot of people
- 12:43see concerns with AI. I think CFOs are not the
- 12:46only ones. a broad spectrum in society. But 34
- 12:50% also said that they were afraid of lack of
- 12:54skills and expertise would come from AI. And
- 12:58it surprised me in the way, because when coming
- 13:00to new technology shift, it would be to think
- 13:05that people are gonna keep the same knowledge
- 13:07of the old ways of doing things. and also adding
- 13:11the new, I think would be a little bit naive
- 13:13to think. And it's almost, to use an analogy,
- 13:17using the pen and the paper compared to today
- 13:19when you're typing things on a PC or tablet.
- 13:23In that context, some expertise and skill will
- 13:27change and will be not needed anymore. And yeah,
- 13:32okay, there could be that your AI tools are not
- 13:34working and you need to find other ways to solve
- 13:37the task. But I think it's kind of natural. I
- 13:40mean, I have the benefit of having worked in
- 13:42a few different countries, as you say, and I
- 13:45speak a few different languages as well. And
- 13:47I've had great pleasure. of that, interacting
- 13:50and learning the culture better. But there is
- 13:52also a world today where most of that you can
- 13:55do with AI. And does that mean, for example,
- 13:58that speaking six, seven languages tomorrow is
- 14:01going to be equally important as yesterday? Probably
- 14:03not. So I think that that's kind of a natural
- 14:07change. The other thing that surprised me a bit,
- 14:10we asked the C4 leaders in terms of priorities
- 14:13of making AI investments and finance. Where do
- 14:16you see them? Where do you stack rank them? And
- 14:19the 40, I think 45 % or 43 % to be exact said,
- 14:24you know, we're going to invest a lot in FP &A.
- 14:27So financial reporting, forecasting and analysis,
- 14:29but only 30%. or 34 % for AR looking at transaction
- 14:34processing. So significantly more thinking that
- 14:37they're going to invest more in AI in FP &A.
- 14:40And for me, it kind of surprised me from two
- 14:44perspectives. A, transaction processing, that's
- 14:46where you have a lot of people, a lot of repetitive
- 14:49tasks. What I would think here is lots of potential
- 14:53for actually driving meaningful impact with AI
- 14:56and bringing people further up to strategic finance.
- 15:02But kind of thinking about that, is it because
- 15:04organizations think they've already done it?
- 15:06My experience during meeting a lot of customers
- 15:08is there is a lot more to be done in that. territory.
- 15:11People might have administration but not automation.
- 15:15The second thing is I think that the FP &A agenda
- 15:17perhaps is closer to the CFO transactional processing
- 15:21and hence it more inclined to invest and prioritize
- 15:25FP &A, which I don't think I don't say it's not
- 15:28important. It is important. But I would think
- 15:30that actually you have more opportunities to
- 15:32work on the transactional processing part of
- 15:35it. So that's two kind of small little surprises.
- 15:38And if I can mention the biggest surprise of
- 15:42them all from the study, it is we asked people
- 15:45to think three to five years ahead. And how do
- 15:48you think the final staffing of the organization
- 15:51is going to change? Do you believe it's going
- 15:52to be roughly the same number? Is it going to
- 15:54be more? Is it going to be less? And I think
- 15:5663 % said it's going to be at the same level
- 16:01or more. So actually 33 % said it would be more
- 16:04headcount and only 34 % said less. And given
- 16:07everything else we hear in society today about
- 16:10AI and what it will do to to labor, that was
- 16:14a surprising insight that so many thought that
- 16:17there would be no change or an increased staffing.
- 16:22So there you have it. Three different. You asked
- 16:24me for one, you got three. Thank you. Buy one,
- 16:26get two free. We love deals in the U .S. There
- 16:29you go. There you go. Um, two things though,
- 16:33like let's take a quick break to talk about invoice
- 16:36processing. Maybe your finance team still enters
- 16:39invoice data by hand, or maybe you've moved to
- 16:42OCR scanning, but someone still has to build
- 16:45templates, maintain them and fix what the system
- 16:48misses. Today's sponsor Rellian takes a different
- 16:51approach with AI invoice capture. Think chat
- 16:54GPT, but for invoices. It reads the information
- 16:57on an invoice without needing a template or upfront
- 17:00training. From there, William can route approvals,
- 17:03match invoices to purchase orders, and help manage
- 17:06payments all connected to your ERP. This will
- 17:10allow you to give your AP team less data entry
- 17:13and more room to focus on exceptions and decisions.
- 17:16See it in action at realign .com slash AP dash
- 17:19automation dash software. I'll drop the link
- 17:21in the show notes. Now let's get back to the
- 17:24episode. The last one, I'm surprised that, you
- 17:27know, I'm also surprised that people will think
- 17:30that there will be even more people. in department,
- 17:35but also the FP &A part. I'm not too surprised
- 17:38that there is more attention to FP &A. For me,
- 17:40it's more because it's almost like a show. It's
- 17:44something you can kind of do show and tell. And
- 17:46it's beautiful to have beautiful dashboards and
- 17:50things and graphs and scenarios that moves easily
- 17:53and commentary. I feel like that's the quote
- 17:56unquote sexy part of finance. And I think that
- 18:00that's probably also why, in addition to what
- 18:02you mentioned, because it's close to the CFO,
- 18:05that it may be getting the attention. But I do
- 18:07agree that the actual work, the run of the work
- 18:10is in the transactions. actions are not right.
- 18:14Doesn't matter how cute the dashboard is, is
- 18:17it really useful? That's right. And I would say,
- 18:19you know, one of the other things that came out
- 18:22of the report is that I think more than 80 %
- 18:25of the CFOs believe that the data in their organizations
- 18:29are so good that they can fully use AI on it
- 18:33for, for example, those forecasts or analytics
- 18:37around it. And I think, I mean, if that was the
- 18:39case, that is great indeed. My impression is
- 18:44that the data, at least what I've seen so far
- 18:47working with AI is that now we make numbers at
- 18:52democracy and everybody can run numbers through
- 18:55AI. But if you don't got your data in order,
- 18:59basically people are going to look at the same
- 19:00data set and come up with lots of different results.
- 19:03So there needs to be a fundamental investment
- 19:08into data platform and the structures around
- 19:10it so that you can trust what's in it and people
- 19:13with a number similar to thing you can interpret
- 19:16in many different ways for sure that you can
- 19:19spin the story but you don't necessarily want
- 19:22to argue what is the churn rate is it you know
- 19:2410 5 15 depending on how people have prompted
- 19:28the data then it becomes a problem in itself
- 19:32and I think that yes a lot of time has been kind
- 19:36of spent on driving dashboards. We've done that
- 19:38in our organization as well. But sometimes the
- 19:42investment to automate a dashboard that was already
- 19:46kind of semi -automated, the payback on that.
- 19:49is relatively little. True. So how are you using
- 19:52AI in your own finance team today? Yeah, thanks
- 19:55for asking. Maybe kind of starting with the transaction
- 19:59processing part and then we can kind of work
- 20:01through the whole spectrum. So obviously we use
- 20:05our own tool. for AP automation that kind of
- 20:09helps us with driving, capture, driving the invoices
- 20:12in with AI and the coding and workflows. So there
- 20:15we use it ourselves, which makes sense. We use
- 20:19from a kind of a transaction AR perspective,
- 20:22we use it in the deal flow. Not in all processes.
- 20:26I mean, we're now checking how we can do Dunningham.
- 20:29payment reminder letters, which we outsourced
- 20:32today. But now we're looking to bring it in -house
- 20:34and to drive AI through that. But we're doing,
- 20:37for example, sales commissions calculations.
- 20:40We're running that in AI with verification. from
- 20:43people in the end, because it's about paying
- 20:45money. We're doing it for partner kickbacks in
- 20:48our SaaS business. We use it when it comes to
- 20:51accounting. In the month -end close, we got lovely
- 20:54dashboards showing where we are with different
- 20:56steps in the process. We use bank reconciliation
- 20:58and other reconciliation with AI. We use for
- 21:02the monthly reporting pack. We use first various
- 21:07analysis to pick up the differences. We ask AI
- 21:10to do the first draft of the report. after we
- 21:14validate. And we also do it for our commercial
- 21:17analytics that we run AI. We also use it for
- 21:21cash flow reporting and forecasting for the next
- 21:2530 to 60 days. So there's lots of different usage
- 21:28areas, but more to come. More to come, I think.
- 21:31But so far, we've actually managed to touch a
- 21:33lot of the different processes in finance. When
- 21:36it comes to forecasting, Not so much yet. Certainly,
- 21:41we're using, we're running a budget process at
- 21:43the moment and using AI to kind of build some
- 21:47of the framework, asking AI to prompt and ask
- 21:51for replies, refining some of the analytics in
- 21:54it. But for sure, we can do more of AI usage
- 22:00in all areas. But that's a couple of examples.
- 22:03That's awesome. So how did you, like, actually
- 22:07get it inside your workflows? Like, what type
- 22:10of governance or boundaries or buy -in or, you
- 22:15know, how did you get people on board? How did
- 22:17you validate the tools you were using before,
- 22:20you know, knowing that it was the right time
- 22:22to bring, for example, AI in a process? Just
- 22:24curious to see what was your thought process
- 22:28and the steps you took to get there. Yeah, I
- 22:30know, great. Great question. For us, and for
- 22:33me, it started with us having a huge focus on
- 22:36on AI. So we've had an ambitious agenda for the
- 22:39past two years. And I remember before I joined
- 22:43this company, I was two and a half years ago
- 22:46at the kind of a big CFO event where they asked
- 22:49roughly 50 CFOs how much AI they use. At that
- 22:52time, there was one person raising their hands,
- 22:54you know, one of them. So things have obviously
- 22:56moved very, very fast. And some people haven't
- 22:59started yet either. And it's good to get started.
- 23:02But for us, it really started with a very clear
- 23:05agenda. of using AI in our product, getting AI
- 23:09usage across the different teams. I joined at
- 23:13the beginning of this year, and then we started
- 23:16to use AI in a few different parts of the team.
- 23:19But for me, it started with a clear focus on
- 23:22making everyone use it, including myself. And
- 23:26so you kind of have to lead by example using
- 23:29it. And number two, for some people who haven't
- 23:33used it, start small. but making sure that there
- 23:37are AI use cases in their OKRs or in their goals
- 23:40for the year and make sure to follow them up
- 23:43and creating a culture of it's okay to be at
- 23:47different levels, it's okay to test, it's okay
- 23:49to make mistakes, but really setting that culture
- 23:52of working with AI. That's what's been my focus
- 23:56this year and so far I think we've made great
- 23:58inroads. Okay. And at some point you were mentioning
- 24:01on how you were using AI in your team, you said
- 24:03when it comes to paying people, you let AI draft,
- 24:08but you validate first. Before you send your
- 24:11commentaries going to executive leadership, you
- 24:13let AI draft and then you validate it first.
- 24:16Can you speak to the importance of doing that
- 24:18even as AI gets better and better at what it
- 24:22does? Yeah, I can. And I think in the end, the
- 24:27accountability for finance stops with you as
- 24:30the CFO, right? So you can't come into a board
- 24:33meeting and say, oops, the number was wrong because
- 24:36of Claude. or because of Copilot or because of
- 24:41any other tool, the accountability is there with
- 24:46you. And sometimes the challenge with AI, we
- 24:48take a little bit of shortcuts, but when it comes
- 24:51to actually paying money. out to employees or
- 24:55even paying invoices for that matter, you want
- 24:58to make sure that it is right. And it's similar
- 25:02to approving an invoice. If you set up POs, yeah,
- 25:07you can fully automate it, all right, but replacing
- 25:10the full approval of big spend, it's problematic.
- 25:13sometimes because the box stops. And if you're
- 25:16sending a message and into a board forecast,
- 25:21you kind of need to know what it is that you're
- 25:24sending. It's the same as going into a board
- 25:26meeting and there's 40 slides and you've had
- 25:28your team prepare them. And you've basically
- 25:31looked at one slide before going into that meeting.
- 25:34That doesn't work. You still need to own the
- 25:37message you need to own what's there. So regardless
- 25:40who prepared them, if it was AI, if it was your
- 25:42team, you need... you represented, right? So
- 25:45I think for those reasons, you want a human in
- 25:49the loop, if we use that expression, to validate
- 25:52some of those important messages. Yeah, it definitely
- 25:56won't work because you always have that one or
- 25:58two board members that will be like, on slide
- 26:0023, on the third bullet. Of course, that will
- 26:06still happen. I mean, I remember also part of
- 26:09my career. I remember going into one of those
- 26:11big business reviews for the budget, and it was
- 26:14400 slides or whatever that they'd received.
- 26:16And then it was exactly on slide 150. The number
- 26:19is wrong. Yeah, you know, we all make mistakes.
- 26:23Yes, we do. But some commas and decimals, you
- 26:26know, they can always be wrong. The core messaging,
- 26:28though, if you haven't read them through, you
- 26:30kind of need to get that right. And so if you
- 26:33let AI do the work for you and you haven't been
- 26:36through it. You haven't done your homework in
- 26:39reality. True. Now AI isn't cheap, right? So
- 26:43AI isn't cheap. A lot of CFOs are getting hit
- 26:46by big invoices for AI token usage. How do you
- 26:49make sure that the cost of using AI does not
- 26:52get out of hand? Yes, and we have also seen that
- 26:55this is increasing and we've used multiple tools.
- 26:59So we've done a couple of things. Now we made
- 27:02some choices around what application, I'm not
- 27:05going to name the application, but we're going
- 27:08for one for most of the people. And that's also,
- 27:12I think it's helpful for people because again,
- 27:14when I came into this organization where we're
- 27:16using multiples, people didn't really know where
- 27:18to start. And so let's not get confused. This
- 27:21is the one we're going to pick as an organization
- 27:23and we're going to run with that. until further
- 27:24notice. But back to the costing. So in our case,
- 27:27we have a spend limit for every individual, which
- 27:31is up until that limit, you can essentially use
- 27:34tokens. So first, this is the seed -based costing.
- 27:38And then that's, let's say, one for a certain
- 27:41dollar. And then you have the limits above. Let's
- 27:44say for the sake of usage, say $50. And up until
- 27:48that, you basically can use the tokens and you
- 27:52can override. But once you come past a certain
- 27:55threshold, let's say $50, then only very few
- 27:58people can approve the additional token usage
- 28:01in the organization. So that's kind of a way
- 28:04to track that. And on a regular basis, reviewing
- 28:07also those, are there some individuals actually,
- 28:11maybe somebody coding and engineering, well,
- 28:14$50 a month isn't gonna be enough. We need to
- 28:16increase that limit, but you do it deliberately.
- 28:19And then I would say, you know, with AI, you
- 28:22need to experiment, but there are different LLMs
- 28:25that you can use. I mean, I come across this
- 28:28every day. Oh, for some reason, I've used this
- 28:30higher level LLM for tasks. Maybe I don't need
- 28:33it. So I hit my token ceiling. And then you go
- 28:36back and change that LLM. And maybe, you know,
- 28:40one of the most frustrating things is let's say
- 28:42that you're preparing a presentation for tomorrow.
- 28:44And then all of a sudden you use your tokens
- 28:47and it's in the evening. You can't reach in them
- 28:49on and you get frustrated and you look at maybe
- 28:52there are other AI tools I can use. But sometimes
- 28:55when you hit the ceiling, it's also maybe time
- 28:58for reflection. Perhaps you can wait, you know,
- 29:0330 minutes or an hour. There are other things
- 29:04you can do and actually maybe think about what
- 29:07it is that I need from the AI tool. So sometimes
- 29:09it can actually, I found, be beneficial that
- 29:12you hit the ceiling. I said, okay, are there
- 29:15any other ways I can solve this task? Maybe change
- 29:19the LMM. Maybe it isn't just by prompting AI
- 29:23another 15 questions to get to my answer. We
- 29:27do a combination of things. Basically trying
- 29:31to streamline the AI tools that we do use, setting
- 29:34limits and also token levels and monitoring them
- 29:38closely with a controlled approval path for them.
- 29:41Makes sense, because that cost is probably going
- 29:46to get out of hands for many people, especially
- 29:48as you said, if you don't hit your limit, you'll
- 29:50keep going, going, going, going. It's almost
- 29:52addictive like me on Netflix. Like you start
- 29:55watching, watching, watching. You just want more
- 29:57and more and more. And, you know, it's like in
- 30:00the old days, you were using your mobile phone
- 30:03and roaming and traveling across the Atlantic.
- 30:04And all of a sudden you got this massive phone
- 30:06bill. How did that happen? Well, you know, or
- 30:10using, as you say, data using Netflix or any
- 30:14other streaming to oops, my data is gone. So
- 30:17you just need to be a little bit considerate
- 30:19and planned in that's nothing strange. Yep, very
- 30:24true. So how do you balance like experimentation
- 30:28like getting people to try it. I know you mentioned
- 30:32the limit, but how do you balance experimenting
- 30:34with AI versus having the right governance versus
- 30:37also the need to demonstrate value? Because I'm
- 30:40giving you a limit every day or every month,
- 30:42and you are reaching the limit. Like, what are
- 30:44you using it for? How is it useful for the organization?
- 30:47Very, very good. Very good question. Because
- 30:50in the end, people need to be curious, need to
- 30:55experiment with the tool to drive the value.
- 30:58But I would argue you can try and test also with
- 31:02different LLM models. It doesn't have to be the
- 31:05kind of top spec models. So there is actually
- 31:08a way that you can handle that within existing
- 31:10limits. But then when you kind of go beyond,
- 31:13and if it's going to be a real big usage, then
- 31:17I would think like in any other scenario in a
- 31:19company, It's an investment you make of sorts,
- 31:22right? So you need to demonstrate what's the
- 31:24business impact, what's the business case for
- 31:27having a huge increase in tokens in one part
- 31:32of the organization or a function, for example.
- 31:34And if you can demonstrate through the use of
- 31:37a business case, okay, this is why we're gonna
- 31:39have need to use these level of tokens for six
- 31:43months or whatever. It's because we're running
- 31:45this project. that will get these benefits. So
- 31:47it's kind of doing it in any other way you would
- 31:51make any other investment in reality. And of
- 31:54course, there's also the case of, you were back
- 31:57to Netflix, how many people are using their own
- 32:01mobile phone usage or AI for personal use, for
- 32:04example, designing, making drawings of their
- 32:07house project with Claude at home. driving tokens.
- 32:12And you know, that's the same as any other thing
- 32:14that, you know, you bring your PC at home, you
- 32:17bring your phone at home, you still have to act
- 32:19responsibly with this company property. And the
- 32:23same goes for AI. And that's why you also need
- 32:26to monitor the usage and when it's being used.
- 32:30I don't like to control society, but as with
- 32:35any other cost items, you need to kind of be
- 32:37on top of things. True. True. Now I'm curious
- 32:41to hear, since you've led finance in several
- 32:43countries, what would you say changed from one
- 32:46country to the next, and what stays the same?
- 32:48So first, what are all the countries? One, and
- 32:51all the languages you speak. But curious to hear
- 32:53some things that you notice in one versus another,
- 32:55and some things that you can say are universal
- 32:58to every finance team. Yeah, no, that's good.
- 33:01Yeah, no, to mention a few countries. So I'm
- 33:05by birth, I'm Swedish. So I've spent part of
- 33:09my career here. Ireland, I worked for a couple
- 33:13of years. France, I worked in Belgium, I worked
- 33:16in the Netherlands. I have worked in Germany,
- 33:21Switzerland and Russia. So I haven't worked in
- 33:24your lovely, lovely country. I've traveled there
- 33:26many times, but never worked there. Okay. So,
- 33:29you know, what is different and what stays the
- 33:33same? So obviously, you know, regulations are
- 33:35going to be different in every country you're
- 33:37in. Tax systems are going to be different. There
- 33:39are similarities. You know, there's VAT value
- 33:41added tax in a lot of parts of Europe. There's
- 33:44sales tax in the US, GST over in Australia. So
- 33:47of course, those things will change. I think
- 33:50also, you know, obviously, the culture, the leadership
- 33:53style is going to be different. Although I think
- 33:55in some contexts, I've worked in the same kind
- 33:58of corporation in different parts of the globe.
- 34:01And then you kind of might have leadership traits
- 34:03from the corporation as such. But otherwise that
- 34:07change. I think also the way you kind of look
- 34:09at risk and assess risk might be a little bit
- 34:12different. The risk appetite might be different.
- 34:15Now, I think what doesn't change is I think what
- 34:19true good finance is. An organization with high
- 34:23integrity. I think that goes across the board.
- 34:27An organization that partners with the rest of
- 34:29the organization. An organization that understands
- 34:34the business and contributes to the business.
- 34:37That doesn't change. And an organization that
- 34:40is transparent with its assumptions. Doesn't
- 34:44change either. regardless of what culture and
- 34:48country you work in. So the true fundamentals
- 34:51of finance stays the same in my view. So the
- 34:56more things change, the more they stay the same
- 34:58in reality. And I think having worked in all
- 35:02these different countries, people always say,
- 35:04what's changed from all of these? And I think
- 35:07sometimes you stay true to finance and what finance
- 35:11is good at, and you deliver that in different
- 35:14countries. And that's why I say some things definitely
- 35:18don't change. I think the finest professionals
- 35:21that I've met in these countries, they have all
- 35:24those characteristics that I've just listed.
- 35:26And do you feel like we also all have the same
- 35:29struggles, like maybe lack of capacity or resources
- 35:34or time? Absolutely. We're always short -staffed.
- 35:40We're always struggling with poor data, struggling
- 35:43how to work with a business partner, how to convince
- 35:46the business partner of a certain message, deadlines
- 35:50being short, month -end close is going to be
- 35:52tighter and tighter, forecasting budgets. Yes,
- 35:56all of that, the same. Oh, yeah, yeah, that's
- 36:00depressing. Do you think AI is going to help
- 36:02with that part? Is there hope? No, no. So if,
- 36:06you know, if we have worked, you know, I remember
- 36:08starting my career, maybe we closed on day 15,
- 36:10day 20, the day after the month after. And then,
- 36:15you know, we successfully brought it down. I
- 36:18guess you can close before day one. Maybe you
- 36:20can close at minus three. It all has consequences.
- 36:23So, yeah, I mean, I think AI will help for us
- 36:27to close a little bit earlier. Will AI help to?
- 36:31for us to do more forecasting? Yeah, potentially,
- 36:34but you also need to keep track of all the forecast
- 36:36you do. Maybe I can do that for us. Will it remove
- 36:39the need for a budget? I personally don't think
- 36:42so. I think the budget is setting targets. We're
- 36:45always going to need to set targets for an organization,
- 36:48maybe more rapidly in the future than in the
- 36:52past. But yeah, it will certainly help, but it
- 36:54won't fundamentally remove. some of the tasks
- 36:57we in finance, we will still work to set governance
- 37:01structures work to challenge the numbers being,
- 37:04you know, asking for, you know, what's the white
- 37:07elephant in the room, you know, being being provocative,
- 37:10and also being the one that's very transparent
- 37:14with numbers and owning that message. I don't
- 37:17think any of that is is is going going away,
- 37:20although we might use slightly different tools.
- 37:23Okay, so what would you say is one lesson that
- 37:26you learn working abroad that you still carry
- 37:29with you till this day as a leader? Yeah, so
- 37:32I think you have to be cognizant and I mentioned
- 37:36that their cultures are different. It's very
- 37:39presumptuous to think that you coming into that
- 37:42culture in that country that you know more than
- 37:46them how that culture works. So I think that
- 37:48the One of the core lessons I brought with me
- 37:52from that is if you want to make a change and
- 37:55that goes the same if you kind of go into a different
- 37:58country or you work maybe in a different industry
- 38:00that you haven't worked before. Be cognizant
- 38:04of people being there kind of knowing how it's
- 38:06been done before. Don't immediately change before
- 38:10kind of bouncing that idea and asking people
- 38:13I'm not saying that you're oh if I'm gonna reduce
- 38:16people I'm gonna ask people what they think about
- 38:18it that's not what I'm saying but that that you
- 38:20know how is it done in this in this particular
- 38:23environment be cognizant of that that's certainly
- 38:27a lesson that that I will bring with me for all
- 38:31times and be be humble be humble that you're
- 38:35you're not an expert and you don't pretend to
- 38:38be they know it all in that cultural context
- 38:42either. Nice. So some lessons. That's awesome.
- 38:47What would you say is an advice you usually give
- 38:50people, especially if there was a CFO, whether
- 38:53you're just starting with AI or just taking on
- 38:55a new leadership role? What would be your advice
- 38:58to them? So I think I love to say be humble because
- 39:02I think that's important in any context. The
- 39:07second thing is Be curious. I think with every
- 39:11role I have taken on, regardless of what's been,
- 39:14I have been so curious and excited about learning
- 39:17these new things. Whether that is kind of going
- 39:20into setting up a tax manual or whether that's
- 39:25setting up a new procurement process or a new
- 39:29forecasting process or how we partner with the
- 39:32business. So I think going into the mindset of
- 39:35being humble. You don't know it all, but also
- 39:39with the mindset of trying to learn it and be
- 39:41really curious about learning the environment,
- 39:44but fundamentally learning from the people that's
- 39:48there and being curious about the people. I mean,
- 39:52in the end, every organization, finance including,
- 39:55is about the people that's there. So it starts
- 39:58with being interested. interested in the people
- 40:01that are around you and that together with you
- 40:04will deliver the work. And when it comes to AI,
- 40:09I think both of those traits are the same. You
- 40:14kind of cannot expect things to change and everybody
- 40:17to embrace AI if you don't do it yourself. So
- 40:20you have to roll up, popular expression, roll
- 40:24up your sleeves and get stuck in there, but you
- 40:28do. And in the context of being humble, you also
- 40:32have to be humble about, you know, you're not
- 40:34an expert. And also if you want people to come
- 40:37with you on the journey, and I believe it is
- 40:39important people are on the journey, showcase
- 40:42that you yourself are learning and here are my
- 40:47flaws. And I think it's important to be inclusive.
- 40:51I don't like the expression of saying be on the
- 40:54bus or be off the bus. I think that's deeply
- 40:59putting people away. I think with AI, it's important
- 41:02to be inclusive. I do fundamentally believe that
- 41:05people can learn this skill as well. So make
- 41:08sure that you give people the opportunity and
- 41:11believe that they can do it. So being inclusive
- 41:13when it comes to AI is one of the messages I
- 41:16would really recommend people to think about.
- 41:18when they're stepping into a role of CFO or in
- 41:22this modern world of AI. Yes, thank you so much
- 41:26for sharing my last question for you. What is
- 41:29your favorite thing to do outside of work? You
- 41:31know, I have a great passion for wine and particularly
- 41:37Red wine so, you know, I live up north here in
- 41:41Sweden and would you would you think that we
- 41:44produce wines? Not so much, right? Thinking of
- 41:48Napa Valley and maybe France and Italy but but
- 41:51actually I do grow some vines and I've made wine
- 41:54and not so great quality, but I love, that's
- 41:59my big passion. And at least for now, and has
- 42:03been for a few years, but I'm also a little bit
- 42:05of similar, I get to know different things and
- 42:08I get passionate about that. But for the moment,
- 42:11and has been for a few years now, wine in all
- 42:15its different ways of making it, and of course,
- 42:19drinking it. So that's my passion outside of
- 42:22work. That is amazing. How do you keep the vine
- 42:25in the winters of Sweden? They do survive, would
- 42:29you believe, but there's different varieties
- 42:32that you need to use. You can't use your Pinot
- 42:34Noir and your Cabernet Sauvignon. You have to
- 42:37use a kind of local varieties of vine. But then
- 42:43you kind of work. So the quality, it's easier
- 42:46to make sparkling wine quality up here, but even
- 42:49red wine is possible. Thank you. Thank you so
- 42:53much for sharing. I guess for us in the US, we
- 42:56got to learn more about Sweden because we had
- 42:57Zlatan here for the World Cup. Oh, yes, indeed.
- 43:01Indeed. That's right. That's right. We love it.
- 43:04We love Zlatan. We love Zlatan. Yes. Well, thank
- 43:08you. Thank you so much for being on the show.
- 43:10It was super, super helpful to have your perspective,
- 43:13especially you being in the midst of what AI
- 43:17is bringing to finance. So thank you so much
- 43:19for being on the show. Thanks for inviting me.
- 43:22It was lovely to share some perspectives having
- 43:24a great dialogue with you. Thank you. That's
- 43:28it for today's episode of the Diary of a CFO
- 43:31Podcast. If you got something from this conversation,
- 43:34share it with someone in your world who will
- 43:36benefit from it. Don't forget to rate the show
- 43:38wherever you listen. We need those five stars
- 43:40so we can get in front of more finance leaders
- 43:42who will benefit from this type of conversations.
- 43:45Big thanks again to Rillion for sponsoring this
- 43:48episode. Don't forget to download the report,
- 43:50AI in Finance, at rillion .com slash AI report.
- 43:56Subscribe so you don't miss out on any episode
- 43:58and visit the diary of a CFO that come for more
- 44:00resources. This is what see a come on and I'll
- 44:02see you in the next one.