Pete Donell Founder at Duet and Fractional CFO, shares his perspective on ROI in Technology and AI with Guy Hutchinson

Speaker A: Welcome to CFO Insights, the leading podcast for finance professionals in disruptive tech, brought to you by the startup CFO community. I'm Guy Hutchinson. I'm the host of the podcast as well as being a tech cfo. Uh, in this episode we're going to talk to Pete Donnell, fractional cfo, uh, and founder at Duet. In this discussion, we have a deep dive into why CFOs excel when focusing on data analytics and revops. We talk through the way AI spend is a positive trigger for CFOs exploring return on investment in R and D. And you'll be surprised to hear what steps leading finance leaders are expected to take in this important area. Pete, welcome to the podcast. Great to have you on.

Speaker B: Thank you. Thank you for having me.

Speaker A: It's okay. It's all right. Look, um, we see you at quite a few of our startup CFO events. Uh, you've spoken on panels and really contributed and I've always enjoyed our chats. Uh, and you and I were talking a number of weeks ago about things we're seeing in the tech ecosystem with, uh, how CFOs lead, how CTOs lead, how they can partner. Uh, and obviously we are recording this in June of 2026. So it feels like, um, somewhere in the middle of the AI explosion, you know, at least for finance people. Thanks for coming on. Uh, it'd be great just to get like a little view on your background, how you built your finance career, and then maybe we can dive into some AI centric topics as we go.

Speaker B: Yeah, of course, going way back, most M of my career happened by sort of luck rather than judgment really. I ended up doing accounting, uh, at university because I failed to get into my first choice university, uh, degree. And then, and actually that was of course run by one of the big four. So I ended up, I qualified in the big four, having partly worked there at university and then worked in a FTSE company which had been one of my clients. Kind of classic audit route into one of the clients. And it was actually there that a colleague of mine, it's a property company, uh, went and joined a prop tech and they needed a finance person. I'd spent seven years or so in corporate environments in big four, and then this footsie company and he said, would you like to come and join this 14 person, I think it was at the time, sort of pre seed, startup seed maybe. Um, which was a big culture shock when I joined. Yeah, I bet. Uh, but I went in and um, learned a lot and really I was the thing that helped Immensely back then, we're talking seven, eight years ago at least, um, was that we raised some money and that was uh, very new to me. But what I then have subsequently seen is that generally if a founder, uh, or an early stage business is bringing in a cfo, either cash is strapped or you're raising or both. And so I ended up joining a um, deep tech manufacturing software business after that where we um, post series A and then we did quite a lot of series B and then I joined a SaaS company um, uh, subsequent to that, um, which again went through a really good journey. And I did I think a lot of stuff that lots of our members have done like build up finance teams, implement processes, begin to get things like FPA firing, build financial models and fundraisers.

Speaker A: Yeah, fantastic. And that journey presumably you've had to very quickly develop like a kind of complete set of financial skills. Right. Because you've typically joined businesses that were growing quickly. You've been the person who was number one in finance and have had to have like ownership of everything from those aspects of the board pack to um, the financial planning, the budgeting, the financial analysis being, being the counterparty, the one that will um, happily sort of be a voice of reason in some of these arguments. Like how have you found that aspect of the journey? Like what key lessons were there for you as you adapted from starting off more of a corporate environment and then sort of ending up in these smaller but fast growing businesses?

Speaker B: Yeah, so I think on the sort of hard skills side I am particularly interested I guess, but I also think it's increasingly important in what I broadly term data and rev ops. Um, and I separate those slightly because I think there's data in respect of, of data engineering and infrastructure and building a data lake and so on, which I just think is becoming so increasingly important for CFOs and I think is something that increasingly has been given to CFOs.

Speaker A: Yeah.

Speaker B: And then rev ops, not just in the collection that data, but actually really understanding how excellent rev op tooling and processes really help your sales team perform but also help you forecast. The two are kind of intrinsically linked. So on the hard skill side I would say that very broadly being able to understand the environment for processes and software and so on that you would implement in a small startup is clearly just completely different to uh, corporate world. Um, but I think particularly now those two areas in terms of data and rev ops are important. I think on the soft skill side that's where the biggest M learning curve I found my first year In a startup incredibly tough. I would say it's probably the hardest year, year of my career. I found the interaction with founders and the team just completely different to what I'd experienced in a corporate world, which is clearly much more structured even in your human interactions. And I think even now I certainly don't think I've cracked it. I think the ability to resonate and influence and build a great relationship with a founder, but also do that, um, combination of optimism, pessimism that feels like something that you just always trying to get better at and varies depending on the personality.

Speaker A: Yeah, yeah, yeah. It's a key thing. Right. Because most founders are uh, maybe like from a tech background or a sales background. That's been the trend and they're not necessarily used to working with a finance person that will be very structured, very process driven. And so the bit where you've got to establish this partnership where you can generate value by, for example, helping them to work on their targets and doing that in a very structured, logical manner that isn't necessarily how they see the world. Uh, and it can be incredibly hard to build those kind of founder relationships where they're really seeing you as a trusted partner.

Speaker B: Yeah, yeah. And I think you, I think that I've always found that you're normally better off showing the talent and building that trust via demonstrating value as quickly as you can. M. But also the thing, and I don't think I'm as instinctively good at this as probably others, but just really trying to figure out what makes them tick as an individual and how they think, um, it's just so necessary. Um, and I think that the better you can understand their kind of worldview and how they're seeing the business, the more you find the right ways to counterpart points. It.

Speaker A: Yeah, yeah, yeah. It's a good point actually. It's a good point. That piece where, uh, finding a way to connect with them, where you're not just having to work together, where you choose to have lunch together or you choose to spend some time otherwise socially together and you begin to unlock like how they think about the world, why they've taken the quite substantial risks that they've normally taken. Right. Most of these people had had reasonably successful careers before and they've chosen to embark on something that may be a huge win, may be fantastic, uh, for them, but it may well go nowhere as well. Right. It's quite hard to understand those things without really committing and engaging with them a little bit out of the normal context of work. Yeah.

Speaker B: And I think it's certainly, uh, true of founders who are generally somewhat extreme in some respect, they would do it. But I think that it's also true of your other colleagues across the, the C suite. I think you've got, you know, is it a technical leader who is built the thing from day one and has some key software engineers. They're terrified of losing and there's like, they're driven by a lot of fear and angst and responsibility. Um, is it a commercial leader who um, I don't know. This is their, their first, uh, step up into a role and they just really want to prove themselves. I think I've certainly learned that the better you can understand where someone's sitting and what is driving some of those, often the more difficult conversations, the better you can understand why in a budget conversation, this person's pay rise or this hire is particularly important and potentially then help nudge in a direction of okay, we want to achieve the same thing. Because you do almost by definition. But here's how we can do that together. And in a way where your fears I can understand better, but I can also help.

Speaker A: Yeah, yeah, that's a good point. Like you really want the founder to be transparent and open with you so that they're being their, their true self and then unpicking why they do certain things and how to support them on some of those things and to kind of magnify their powers by, by, by, by you being their business partner. Um, that, that could be an incredible thing. And I mean it's easy to talk about that. It's actually pretty hard to go and do. Right. Uh, but certainly it's something that we see CFOs and startup CFO talking about at our meetups and um, finding that they can be successful in doing that and really start to make a dent in that important relationship. Brilliant. And there are some themes in your career. Right. Some of the themes are around, um, roi. And you mentioned rev ops earlier. Right. So in most B2B businesses where you've got some high growth ambitions, um, it's really rev ops that's going to unlock the predictability of your pipeline.

Speaker B: Yeah.

Speaker A: And how well you understand the potential for the business to grow with for example, a certain amount of marketing spend, a certain lead flow, um, certain number of sales people to handle those. Right. It's a classic kind of B2B funnel. Yeah. And it sounds like from, from your standpoint you've really engaged on that topic and look to be the CFO that can understand those things as well as the sales people. Can and the marketing people can. And to be somebody who could challenge those things but also really understand those mechanics for financial planning, etc. Was that like the first, sort of your first perspective on how do you get ROI from sales, from marketing in a high growth business?

Speaker B: I first encountered Rev Ops and data, not necessarily by a judgment of my own. It was um, I was asked to look after it, um, I think for a couple of reasons. One, because it was a, you know, a challenge that needed help with which I think can often come the way of the cfo, particularly if the finance team is sort of beginning to fire. But also I think there is a natural harmony partly for the reasons that you just described, um, in that there is a clear relationship between rev Ops and what gets captured and forecasting an roi. Um, but equally I think that it's both of those areas are kind of areas where data stewardship and even reconciliation or having that single source of truth, it's an innate part of how we operate on the accounts. So I think that there are manners in which it lends itself. But certainly what I then uncovered, um, and again by sort of chance really is just how powerful effective revops all the way back to understanding the customer journey and collecting data in the way that the customer wants to give it to you to help you forecast as opposed to asking salespeople to, to add a new field in the CRM. All of these things, um, allow you to really go through a journey of a being able to forecast more and more accurately and therefore deploy capital more and more accurately, but also partner your commercial stakeholders in a manner that's considerably more collaborative than competitive.

Speaker A: And that's also some of the more fun work to do as a finance person. Right. You know that bit where you are accepted by those teams that are driving growth as somebody who's also knowledgeable in that field. And uh, it allows you to unlock certain things. Like my personal experience of startup and scale up world is that when you speak to founders who are clearly in the scale up stage and you talk to them about their product and what's driving revenue now, it very often isn't the first, first thing that they started off the business with, right? It might be that the business has had a micro pivot or it might be that the first three or four years they were selling into small and medium sized enterprises, uh, and then they unlocked the first enterprise clients and they pivoted as to who, who their target customer was. And I think one of the, the bits in RevOps where the CFO can very often be like a powerful um, counterparty. To sort of argue for thinking about the business growth differently is to unlock that data set and to be the independent person. Because obviously if you're the salespeople or you're the marketing people, you're not completely independent. You are living inside this and there's a little bit of status quo bias will come with that. And so you can be well positioned to say, hey look, you know, have we thought about what would happen if we took these three salespeople off the small and medium size market and got them onto enterprise? How would that work? What would that mean for our growth? And so um, it seems to me that it's not just about helping the business to tick over successfully, but you can also, you can also be a part of revolutionizing some businesses.

Speaker B: Yeah. And that one of the businesses I worked in the SaaS business, we definitely went through this journey from what would be called founder led sales and, and uh, a sort of early uh, stage sales approach through to professionalized pipeline management sales team, uh, sales team members with more structured, repeatable go to market motion. I think um, in some ways the job of a CFO is to create that leverage for a founder, um, as you go through that scaling journey. And I think that leverage is, comes from being able to manage a head of sales to their pipeline rather than, and perhaps jump in on the deals that are crucial, um, as opposed to trying to be in every single deal. But exactly as you just described. What you then might also see and begin or what might emerge is then here's the trend that we hadn't necessarily expected until we saw that of our five different um, flavors of ICP that hit our pipeline, this one seems be to be converting a pace at high value. Yes, absolutely. And that's where I think you ideally want to be a CFO for a higher percentage of the time is I'm not offering personal opinions that I've seen. I can do that sometimes and I need to do that sometimes. But actually I'm simply providing data that appears to tell us to go in a certain direction.

Speaker A: Yeah, absolutely. And that harks back to the comment you made earlier about, you know, really strong finance leaders. In the end it's the data and finance people tend to have a good mindset for um, how do you create your data sets? How, how clean is your data? Is it reliable? Are we all looking at the same reports? Do, do some people sort of cook up their own reports and like rely on those but actually it's not interpreting the data set that we have correctly, or it might even be querying data that we don't believe to be truly clean. Um, so it's that magnification of if your finance people really are like excellent practitioners in the world of data cleansing data analytics, then um, that's your platform to be this commercial operator?

Speaker B: Yeah, I think so. I have a bit of a kind of personal crusade against this, uh, stuff that you see a lot on LinkedIn, which is about this idea of this strategic CFO that I think encourages people to operate great in this. I'm in the boardroom saying impressive things and this may be my own limitation in what I understand, but I think most of them, the most strategically impactful things that I've done have been implementing processes that get you really high quality data so that your sales team can see it and therefore drive forward. You know, they're not, um, it's what you just described. It's high quality data being given to the right person, probably with a, you know, helping them understand the story that that data might be telling. Um, but this kind of, and frankly some of the most strategic stuff that I've done that I think helped drive the business forward is making sure month end is closed within a day and being able to monitor revenue daily through a dashboard that you know is going to close within a material accuracy at the end of the month. And I think that that seems to get devalued a little bit now in terms of, oh, it's not strategic to be in month end close. Uh, I personally don't see it that way and I think it risks people kind of focusing on these conversations that without the data is at some point will become a bit hard to then really see the value out of.

Speaker A: I, I do agree with you completely there. I think the discussion about strategic finance is very often it's this imaginary world where uh, the CFO can be like, kind of like a McKinsey's person and like, oh yeah, I've been like strategizing and here's my slides and this is going to change everything. That's just not really how most businesses work. Right. That, that is in the main a fantasy. And actually some of the things that lead to a strategic outcome are just kind of grimy things about tweaking processes, improving things here, having a better understanding as to what the value of your clients are, how efficient your sales and marketing are. Um, do you have a product which is truly sticky? Is the headroom on pricing like those things are not. Well, in the end the outcomes are strategic Right. But you wouldn't walk into the boardroom with a big strategy vision on those things. Um, it's just something that you should be always thinking about, always talking about. It just, it just emerges from like good business practice.

Speaker B: Yeah. There's a slight caveat to that which I've seen, which is I think there are some more on the kind of capital markets, whether it's investment banking background or, or um, the PE type mindset around value creation and particularly kind of the financial engineering where I think there are some people who are incredibly strategic about how they think about capital markets and how they extract value out of a company. Um, so I think there is an element of that, but I think this idea that, um, generally in the operating motion of the business itself, I think a lot of the work to do is to get high quality, quality data in front of people and help contextualize that sometimes actually get less data in front of them than more that, you know, is actually going to be impactful to the business. Um, but yeah, I do struggle with this um, concept of uh, I like the way you put it, the kind of McKinsey consultant type, because I think there's a place for that, There's a place for putting together a really strong PowerPoint internally off to the board. But, um, pretty quickly the data isn't there to back it up. Then it's just a deck.

Speaker A: Yeah, yeah, absolutely, absolutely. We should talk a little bit about one of the key sort of like lessons in your career around, um, what you wanted to measure and how you wanted to help some of your businesses to double down on things that were most impactful. And we chatted in the past around your viewers, like, hey, look, if you look at RevOps, there's a refined methodology, a shared language between people in different departments about how that works and how to measure, ah, optimal outcomes. But the minute that you step into say the tech and product silo and you begin to have the same conversations about roi, how do we know that um, doing this product release is going to give us a return? Uh, historically tech businesses have struggled with that. It hasn't been easy. Uh, but somehow everybody knows that that's the lifeblood of these businesses, even though it's hard to measure. And you've embarked on, uh, quite an interesting project last year to start to feel out like how this could be more measurable. It'd be really interesting just to hear about that and how you've taken those steps.

Speaker B: Yeah, so I mean this goes back in particular to the um, deep tech startup that I worked in where my job was really to try and understand the technology and uh, translate it into a financial outcome that was not evident in revenue or anything else because this was still in the scientific validation phase really. And we, we did that successfully. We did a $45 million Series B and got the outcome that we hoped from that. Um, and then what I observed in more in SaaS and in other areas is exactly what you've just described, which is certainly the hardest part of my job as a cfo, I found, was how do I help allocate capital into R and D or tech? Um, I think you can get a really strong rhythm with your commercial leadership, particularly as rev ops, as we've talked about, begins to get embedded. I think the sort of support functions like finance, legal, hr, it's a lot more, um, intuitive that really you're just trying to be efficient in those areas and deliver good quality service to the rest of the business. Um, but there is this huge chunk of the P and L that we have historically largely seen as just a single line of opex might capitalize it, but most earlier stage companies certainly don't, um, which actually have hugely varied impacts on your business, uh, and where you're deploying that capital. And I think that tends to be a bit of a black box and it tends to be a somewhat, um, challenging back and forth, particularly with a technical leader to try and unpick what we should be spending there and why. Um, I think historically that's been reasonably manageable because software engineering has been super hard and it's taken a long time. And so we put together a roadmap and we ship that roadmap and the conversation has been framed much more in terms of what can we do for the number of engineers that we've got. But suddenly the feedback loop on everything is dramatically shortened. Um, not least because AI can generate code very quickly, doesn't necessarily mean it's valuable, um, but it can generate it very quickly. So that dynamic is changing hugely even in the time where, you know, I've started looking at this in depth. Um, but my fundamental belief, I suppose, is that by empowering CTOs and by empowering technical leaders with a better use the word shared language, you know, with a, with a better shared language for how they're creating value, we're going to help them make better decisions. And pretty quickly I think that's going to be a necessity for businesses to succeed as opposed to an afterthought.

Speaker A: I can completely see that. Um, like, I think it's just going to be, it's going to be a discipline that businesses will have to have. And I, I feel that when you think about the rate of advancement in technologies at the moment, right. You know, whether it's AI or other things, to imagine that you could have a, uh, startup or scale up whose expenditure could be in the millions, if not tens of millions in a year. Uh, and the best, like maybe half that or more is going into tech or product for that, for that not to be measured or deemed not measurable just does not make any sense really. Uh, it's just a case as to how that could be unlocked.

Speaker B: One thing that I discovered fairly early on is I think there's a misconception around tech and R and D. The other things can be projected perfectly. So, oh, we can't do this. I know that you can see sales quotas and measure roi, um, of your commercial functions, but we just can't do that in tech. And that's both CFOs and tech leaders, by the way. But actually anyone who's done a lot of rev ops or worked on a marketing budget or, or even just calculated unit economics knows that there's a whole heap of assumptions that go into it. It's a combination of hard facts that we have in the data and a series of assumptions that we're going to try and learn and improve over time. And so I think the reason I say all that is that I think one of the challenges with R and D is you need to have a model for how it's creating value in your business, even though that's going to be wrong. And like with almost any other model, we understand this from a financial modeling perspective. And I mean model in quite a broad sense. We understand this innately within our kind of commercial or we've, we certainly learned it maybe not innately, but in R and D, we haven't done it. And the earlier you start envisaging that model and having that hypothesis, the earlier you can test it, the earlier you can actually make that model better and better and better, which as I said earlier, I think is going to become an increasing necessity for technical leaders. Because that feedback loop is shortening.

Speaker A: Yeah. And how would somebody. So how would that conversation start about we need to create a model for understanding, uh, if we're succeeding in our tech efforts. How do you start the conversation? What's an early framework look like?

Speaker B: Yeah, so, um, a classic, uh, the exact same problems that exist within sales and marketing exist in engineering. So why is it hard to understand engineering value? Because When I shipped something, I might have shipped it, but the sales team also sold it. And so is the revenue attributable to me or is it attributable to something else? So there's an attribution problem. There's also a timing problem. I might build something nine months ago that suddenly starts generating value. And that value might not be a customer paying for it. It might be that next time my engineers working, it's much easier for them to work in. So these problems that get described in engineering actually really analogous to sales and marketing because again, if you've tried to measure, uh, ROI of different marketing channels or different sales channels, you have sales and attribution problem, that you have attribution and timing problems almost always. So the reason I say that is that beginning to build the model and the framework in which your engineering team is creating value is the starting point has to be how do you understand how your organization creates value? Uh, and I think CFOs play a really important role in this because by definition you should be the one who has that formula in your mind most clearly. And it probably should be in the financial model or certainly key aspects of the drivers are in the financial model. And then over time you probably improve your financial model. I think that's the exact same process in software engineering is how do I model what customers value and what I'm building that customers value? How do I split that between new product and existing? Very similar to new customers and existing customers. Um, how do I understand the internal infrastructure that continues to drive that? You know, what security requirements do I need to meet? What, um, database infrastructure do I need to have to process things? Which third parties am I relying on? Um, you know, there is a series of ways in which it's kind of like financial modeling. You can, you can begin to model and at least have hypothesis for how you're driving value. Now what you might find, which is very common in engineering, is we build a feature because we think a customer wants it and it generates no revenue because it doesn't work. But that actually is directly analogous to sales where you might have four customers that you're trying to pitch to in your pipeline and only one of them converts. We don't lose our minds over that. In sales, what we try and do is make sure we have enough pipeline, whereas in engineering we have this tendency to, to go, why has this feature failed? Everything's wrong.

Speaker A: Mhm. So is the idea that the technology teams can learn a lot from m, the way that sales and marketing think about growth And ROI in that, uh, the salespeople do not lose sleep over the leads that don't close. They're in the business of selling and winning some business and along that journey, because it's not just one or two leads that you're approaching, it's a whole bunch of them. Some of them just won't move down the funnel. Right. And some will be genuinely closed lost. And if they obsessed about the closed lost leads, um, that be, you know, dispirited from, you know, closing business. Right. Whilst essentially part of what they do is to go about your work with some zeal, with some optimism. Uh, and that allows you to get through that process of like, look, you might be three that you don't close for one that you do, but actually if you close enough of those then you're good and you've certainly got ROI on the time from sales and marketing. Is that how you think that tech and product should be thinking about things?

Speaker B: I think I'm certainly inspired a lot by, by RevOps in terms of how we think about certain aspects of software engineering. I actually think the problem you've just described is more of a leadership problem than it is a problem at the engineering level. And so we talked a little bit earlier about kind of creating that leverage RevOps allows the founder to move out of. I'm going to jump into every single deal and into. I'm going to observe pipeline materializing. Actually I think what engineers suffer from the most is that because we don't have that ability to create the leverage for them in the shared language. Often whether it's founders or a salesperson who's stroppy because they've lost the deal or something like that, we look at every single feature release and begin to scrutinize what went wrong and often change the product operating model or do something differently and flip back and forth. Now I see it through the lens of capital allocation. What I want to know and I think is going to become an increasing challenge for CFOs and CTOs need to be in lockstep with them, is are we deploying enough capital? Is it going into the right places? That's whether it's going to the right customer type or market, whether it's being the capital is being executed by a human or an agent and what level of experience or model is that agent? I think that. So the question that I think engineers and CTOs and leaders can help begin to get founders and CFOs into is not why did that feature fail or have we shipped this roadmap item on time, the question should be, am I allocating enough capital into these bets to give myself a good chance of the revenue growth that I'm projecting? There's a really good podcast with the Revolut founder talking about how he has these 20 experimental product squads and seven of them were successful last year, which he sees as a great success. And, uh, this is probably the single most successful company in Europe, certainly in the software space that has this exact methodology. Um, now you can't treat every bit of engineering like that. It's a little bit like a sales pipeline and then account managing your existing customers. There's bits of software engineering that needs to be understood by, okay, how do we keep maintaining that bit of the product, um, but without helping a CTO create that model? I think we're always leaving them trapped and unable to be as strategic as they could be.

Speaker A: That's a really interesting viewpoint and I think you've explained that incredibly well. And so this has led you to actually build a product. Right. So we should talk a little bit about Duet and what you've been working on for the last sort of eight or nine months. It'd be great just to distill that down to what, uh, could, what could a technology leader, uh, do now with a product like Duet?

Speaker B: Yeah, so we are a services and platform hybrid in practice. So what we do is really two things. We come in and help build the model that I've just described. We work with your CTO and cfo, uh, or team and help you begin to picture that value model for your engineering function. And then the platform helps you begin to track and learn and improve that model, which is the nature of how a model gets better. You can't kind of one and done it. So we come in and it might be we've seen companies that are wanting to go into a large transaction transaction in a year's time and therefore want to really understand how software engineering is creating value and almost be able to create that case for engineering hiring. We've seen companies where, um, they have an engineering platform team, which I think lots of, um, existing methodologies do this really focus on just the revenue side, the customer side, whereas actually engineering. There's all sorts of different investment types going on in there. So we've seen ones where they've got a platform team and they need to support it. Or we just, you know, seen a lot of companies, particularly more recently, who just want to understand what on earth, uh, this large bill that comes in at the end of the month from one of the foundational model companies is you know, is worth, are we actually getting good value from it? So that's what we do. We come in and help you understand the model for value. Ideally get ah, get you on the path to your CTO and CFO being much more strategic.

Speaker A: So that's an a framework, a model that you set out that should allow that business to start off the process on the understanding that this is a sort of continuous learning environment and that's um, being reshaped over time.

Speaker B: Yeah, that's correct. I think like um, any forecasting area, it's directional, not absolute and it can't be perfect. The aim is to get better and better. But what we are pretty committed to, we've done a lot of research to support is that we understand um, um the financial implications of software engineering better than most. Um, and whether it's something as simple as can you benchmark the level of activity that you're doing based on your growth rate, based on perhaps because you're a um, cyber security company, so you're a software engineering team, should probably be investing more in cyber um, or I know you're an AI native company where you should be doing a hell of a lot of research and development right now. You know these are use even at that very basic level of budgeting. Those are useful reference points. Then the further and further you dig into building that model, the more and more you can begin to deploy and allocate capital in the right way. But it will be directional, um, it can't be perfect. And it also M requires that sort of. Mindset of okay, I may not, you know, I probably still need to allocate 25, 30, 40% of my time to things that I can't see a direct ROI out of. What we actually find is that engineering teams tend to spend too little time on genuine R and D and too little time on paying our technical debt and therefore are not able to really influence the business as strongly. Because right now if you want to be a hyper growth business, I think your engineers are going to have to

Speaker A: be doing R and D. Yeah, yeah, yeah. And this, this, this problem that you're solving through Duet, you're building this right in this moment where you know, progress with AI tooling which is accessible to all companies now. It's not like you don't need to be a big company to access this kind of technology. Everybody has it. Some of the you know, Claude releases in the last six months have been astonishing. Uh, like how does, how does that Change how you think about measuring roi, right? Is it, is it making the problem bigger, uh, harder? Like, like have you had to think about the change in scope for Duet around AI?

Speaker B: I think that the, the real, the huge positive for us is that it's accelerated a conversation. I feel like two years ago when I talked about to CFOs and CTOs about software engineering ROI, it was a bit like a. Okay, yeah, that's interesting. I've never really understood it, um, but really I uh, tend to get to the end of the year and just kind of asked for a number of software engineers.

Speaker A: It's not just you doing that by the way.

Speaker B: No, I think it's massively accelerated the need to understand and orchestrate and allocate that capital effectively. And again the shortening of the feedback loop I think contributes to this because it used to be that you determine your budget, you might ship three or four really meaningful things in a year that you think are going to drive the P and L that's changing now. So the feedback loop being shorter means you need to understand it more quickly. Um, I think the other thing that we see a lot of and we have to grapple with is that most software engineering teams are doing this in an almost completely different way like how they're deploying AI and there's a huge spectrum. The vast majority are, if not all now are using GitHub, Copilot or Codex or claw code or something, something within their day to day practice. But are they harness uh, engineering and do they have orchestration layers of different agents doing different things? There are some right at the far end of that spectrum but actually it's a smaller proportion than you might read about. But what that means for us, it's one of several reasons why we almost maintain this quite services LED starting point. I mean if companies now call it forward deployment engineering, but it's basically trying to go in and understand exactly how do you operate and exactly how does your financial model work. We need to do that before we can actually architect something that's going to be meaningful for you. I'm not, I'm skeptical there's an off the shelf platform that can do that, certainly not in the short term. So that's really helped us almost narrow in on how we need to operate to help companies like this.

Speaker A: Yeah, yeah, yeah. And when you are ah, um, when you're thinking about what other finance leaders might be going through right now in that uh, they, they might be in a business where they're beginning to get quite deep into like their product using AI, the teams using AI token costs are ramping. Like, what are the foundational stakes for being on top of that? Like, like, like pre taking a structured approach like you might have through duet. Like, what's the first foundational steps that any finance leader should do if they think some of this AI spend is getting out of control?

Speaker B: I think, I personally think the biggest gap that I see is you actually just have to understand your business drivers and your business levers. It's actually just good that hasn't changed. I think it's good quality CFO stuff and I think that there's lots of tools out there that are doing some flavor, uh, of understanding token spend. Where's it going, what's it doing? Um, and most of it is what I would describe as developer forward, I. E. Software engineers have started building it and then they get to this point where somebody's asking them how much, you know, what value we were getting from this. But actually a kind of key part of our hypothesis is you've got to reverse engineer that, uh, at some point you have to say, how does our business create value? And work backwards from that to determine where is AI being deployed, to understand how it then can create value. And this is super important a role I think CFO is to play. Because if you do not have a clear hypothesis for how your business is creating value, if you don't have that formula, it's almost impossible to measure AI, measure the roi, certainly kind of understand things like should it be a human or not that does this particular bit of activity. And the classic is, okay, well we've saved half an hour, but we're not going to lose this person. So it's never going to hit the P L. You're the steward of how activity within the business in the long term creates financial value and hits the P and L. So I think the overarching thing that I believe most finance, um, people can contribute to the, to AI deployment and making sure AI is successfully deployed is have a really good hypothesis for how the business creates value and then really try and understand what are the constraints within the system. So there is no point rolling out a AI SDR tool and doing loads of AI LED outbound. If you already don't have a system for onboarding customers as well as you can, you may well have a load of SDRs who are really excited about doing it. Or perhaps AES and the uh, you know, commercial leader wants to get rid of the SDRs. But, and look, you could take the cost cutting approach But I think the vast majority of companies that are going to create a lot of value are going to be able to almost knock off those constraints one by one. How do I AI enable the onboarding process and then work backwards from that to go? Cool. Now that we've done that, we can ramp up, up the pipeline. So I think that that's an underappreciated um, role that the CFO needs to play. From the conversations that I've witnessed, yeah, interesting, interesting.

Speaker A: So you would recommend that CFOs look to understand what the fundamental business constraints are and work back from there.

Speaker B: What I tend to see is somebody in a business has decided has got this AI type role where they're going around and they have a list of things that people think are interesting and pain points they'd like to do in the business. And then in the inevitable prioritization process you get to some form of cost benefit analysis which tends to be, oh, we can save this much time of this person or we only need to hit, you know, sell one more customer using this AI tool and it pays for itself whatever it is. Like, we've all probably seen some flavor of this and that is, I think it's not a bad process to do the bottom up bit, particularly to get buy in. But I think that uh, unless you have the clear top down view of it's kind of really good strategic financial planning. What is the thing that takes us From, I know 10 million to 20 million of revenue because if you know that that's going to be 50% growing your existing customer base and 50% completely new business and you can identify where you're weakest in those, then suddenly, um, you have your place where you should probably start thinking about deploying AI the most. And everything else is actually noise. This is a prioritization and capital allocation exercise. Um, now obviously we're very focused on how you do that within software engineering because historically that's been through roadmap processes and all this kind of stuff. But I think as a principal for the CFO in general, you ought to be the one who's building that map for the business and helping the stakeholders understand that.

Speaker A: Yeah, absolutely. It does sound like a fascinating project. Uh, I will have to get you on in a year or two's time to catch up on uh, how it's all going, but wish you well. It's a fantastic thing to do. And uh, I think a lot of finance leaders are going to find it fascinating and really want to get to grips with um, how they can have the kind of transparency that Rev Ops has given them for sales and growth into that black box, which is engineering, and to really unlock that kind of partnership with the cto. It's been a really good conversation, but before I let you go, um, the one thing I'd say is in this moment where it just feels like

Speaker B: more

Speaker A: than half the conversations about finance are really about AI being deployed in finance, um, in this moment, finance careers are changing substantially. Um, if you had one lesson to share from your career with people that might be on the CFO path, um, have got that job title in their sights and they're on track for doing that. What, what should they be doing now to make sure that they're going to be a fantastic CFO in a few years time?

Speaker B: Uh, I feel like this is a, it's going to sound glib because of what I'm doing, but I think you should be talking to your CTO a lot. I think that when I look back on my career, I had one CTO I interacted with a lot and I learned so much. And I've had other CTOs where, where I didn't interact with them so well and I think I failed them. And I think that, you know, we probably could have achieved a lot more if we had a better line of interaction. I think that that was true five, seven, eight years ago, but it's going to become even more true now. So I think that, you know, all the usual stuff applies if start playing these tools. But I think that human interaction is going to be increasingly critical. And I think, particularly on the technical side, I think if you, you're carving out as much time with your technical leader, uh, as you are with your commercial leader, uh, that's going to stand you in pretty good stead relative to other finance professionals.

Speaker A: Interesting. So that's really broaden out your concept of, you know, what it takes to be a fantastic business partner. Make sure that the tech leadership is getting time with you. You're overtly finding reasons to have lunch with them, to understand what's important for them and to really unlock the sort of understanding as to what they're doing and why they're doing it is a precursor, uh, to the kind of transparency, you know, which you're clearly passionate about.

Speaker B: Yeah, I, I think, Iris, I, I have no idea, and I don't believe anyone has any idea how humans and interact, agents are going to interact over the next few decades. And irrespective of what we're trying to do with Drout, I think that there is an. I actually think there's been a problem since software engineering really became a big value creator, which is that it's been really poorly understood by finance. I think that problem is about to get extremely acute. Um, whether that's because we've got lots of agents that we're needing to manage that are consuming value, or whether it's because our product is agentic or it's. Our product is at least in. I think every product will at least have some level of AI capability within it. I think that in order for businesses to maximize their value, somebody is going to have to be both financially and technically literate to a degree that enables that. It's no longer going to be enough that we manage the humans and the sales and hopefully tech can keep up with the roadmap. I just don't believe that future exists. So I think if you're. Yeah, if you're a finance person who is understanding technology as well as you can and listening to those people who are experts on the ground in it, I think that's going to be a powerful thing to do.

Speaker A: Yeah, yeah, that makes a lot of sense. Pete, this has been a fantastic conversation. Thank you very much for joining me and uh, talking through these really important topics. Really appreciate you being on the pod.

Speaker B: Yeah, my pleasure. Thank you.

Speaker A: I hope you enjoyed our discussion today. We're really proud of the podcast following we've built up as we run our podcast in conjunction with the startup CFO community. We're able to uh, access many of the experts who are changing the face of the modern finance function, allowing us to hold these discussions, playing our role, shaping the modern finance leader and informing career journeys in the age of 8 AI. We're seeing the most dramatic changes in how CFOs apply themselves to supporting high growth businesses. It's a time where peer support will add more value than ever before. And lastly, if you're not in our group already and want to join, just go to Startup CFO Tech and click to apply to be part of our exclusive community offering.