AI Transformation: The Rise of the Fractional Chief AI Officer | John Sukup | E1S10

Utsav: Hello everyone, welcome to Alt consulting conversations. I am Utsav, founder of StratOff We help companies drive AI adoption and innovation led growth. In this podcast, we focus on two things. First is the emergence of new consulting models and what the future of consulting looks like.

And the second is how companies are actually navigating AI adoption in practice. One role that's starting to show up more clearly at this intersection is the Chief AI Officer. Some companies are building this capability internally, others are experimenting with more flexible models. And that's where today's conversation gets really interesting.

I'm joined by John, who runs Expected X where he works as a fractional Chief AI ⁓ helping companies move from AI strategy to actual execution without the need of building a full-time internal function from day one. John, great to have you here. John Sukup: Right Utsav, thanks for having me. Utsav: Great so let me start by asking you, how did you land up with this idea of being a fractional Chief AI Officer?

John Sukup: So I've been in the field for pretty much my whole career, ⁓ or at the data field, we'll say. ⁓ I didn't out in ⁓ AI and machine learning, but ⁓ gradually my career kind of evolved into that. So, ⁓ Expected X actually formed, ⁓ it'll be 10 years ago, ⁓ come May. ⁓ The fact we managed to stick around this long is pretty nice compared to a lot of companies.

But Expected X started out as ⁓ an answer to how do we perform consumer market research in a method that ⁓ does not require ⁓ too much ⁓ prompting or ⁓ ⁓ interactions or focus groups for that matter. ⁓ over time, it's evolved into more of a pure play ⁓ implementation strategy ⁓ and consulting company, but still in the AI ML space. ⁓ The rationale for focusing on fractional Chief AI Officer is that from conversations with clients prospects and other professionals, ⁓ there seemed to be a lot of confusion ⁓ on how to a strategic plan for AI transformation within a company.

Most frequently, a company would say, well, that falls into the realm of our chief technology officer, our CTO but the CTO typically doesn't cover all of the bases. And if they've been around in the field for a long time, a lot of what's coming in AI is tangentially related to their everyday not exactly related. ⁓ the fractional part is because I think right now a lot of companies don't need a full time ⁓ AI Officer helping them create ⁓ direction and road mapping, ⁓ rather someone that kind of comes in at the very beginning of that AI transformation stage and just helps them answer questions like, you know, what should we do?

What makes the most sense? ⁓ How do we experiment without ⁓ through our annual budget ⁓ in a single month. ⁓ a lot of those decisions I think fall squarely on the shoulders of ⁓ roles ⁓ that's specifically where CAIO ⁓ in play. Utsav: You know, I have this question for some time.

I have spent around 10 years working with CIOs, Chief Information Officers. ⁓ this is much of last sort of 15 years ⁓ there's a whole wave of digitalization which came. And then you CIOs and then CDOs. And now we are seeing CAIOs.

⁓ Why do you think CIOs can't play this role? ⁓ Any ⁓ insights that? Because you have been talking to Chief Technology Officers. ⁓ And part of AI execution also means that you need to integrate AI into your existing systems workflow.

⁓ why CIOs are not say well positioned to play this role or vice versa? What do you feel? John Sukup: Well, let me preface it and say that I'm not by any means saying that ⁓ all CTOs CIOs or CDOs would be incapable of taking on ⁓ the role of a Chief AI Officer. ⁓ So just say that, you know, there's obviously a lot of talent out there that maybe they've been doing this for a long time and they just have a different title, right?

⁓ But I really think, you know, what you mentioned there CIO, Information Officer, Data Officer, Technology Officer. ⁓ Those are all parallel swim lanes, right? ⁓ But the Chief AI Officer really needs to have kind of a ⁓ holistic overview of each of those roles because each of those roles plays ⁓ a very significant part. ⁓ And I'll say that this is the same thing that we saw for machine learning, right?

It's a marriage of statistics, data, infrastructure. ⁓ So these are all kind of elements that belong in those three categories of leadership, but often ⁓ they're hyper-focused on, you know, that particular swim lane and having someone that can aggregate and think across those three roles. Maybe not at the same level of depth ⁓ but who can piece those parts together in order to create that ⁓ spark that leads to AI transformation. So again, not saying that the ⁓ chief CIO or CDO or CTO would not be suitable for the role of Chief AI Officer, but again, it becomes an aggregation of those three leadership tracks ⁓ I think sometimes is ⁓ difficult.

And just to add an extra note to that, if you're a brand new organization, for example, like a funded startup that might not have any ⁓ maybe you're focused solely on AI products, then ⁓ a Chief AI Officer could be, ⁓ maybe should be one the first leadership roles ⁓ you in outside of, you know, the founder and ⁓ CEO. Utsav: So it really depends upon the context of the organization and what capability do they have in terms of their current IT demands, how much data is sorted, how much ⁓ of capability and I think readiness is with the IT organization to take on this role.

⁓ it's not, I think it's a very well put point that it's not mandatory for every company to have a CAIO ⁓ they have someone playing this role and can sort of take on this additional capability. One thing I am pretty curious about is timing. So when you speak to companies, when do they actually need a Chief AI Officer? What signals tell you this is the moment to bring in this new capability or ⁓ John Sukup: So in terms of timing, I think it's as early as possible ⁓ when the conversation starts ⁓ to just around ⁓ AI ⁓ any of use case within an organization that has no experience.

⁓ I think that is the ideal time to in ⁓ CAIO ⁓ role. ⁓ And the reason for that is obviously, you know, if you can start with the expertise early on rather than trying to cobble together, you know, from your your tech department and your business department and ⁓ you know all points in between I think when you start with kind of that fundamental ⁓ foundational role at the very beginning ⁓ that is the ideal time not after. We've started experimenting or now we want to put something into production.

⁓ I think that that time you obviously can still bring in a CAIO, but oftentimes it might be a little bit late to the party, so to speak. ⁓ And some of the design decisions that maybe you've made up to that point might be suboptimal and something that you could have avoided if you had brought them in at an earlier stage. The second aspect I would say to that for when to bring in a ⁓ is not just maybe in the stages of we want to do something with AI, right? Pretty much every company is saying that right now in some capacity, ⁓ also when you actually have devoted ⁓ some of your budget ⁓ to ⁓ enabling AI transformation within your organization ⁓ you know, this goes with anything, even in, you know, something you buy as a consumer, right?

You know, money talks, right? You can talk all day about doing this and that, or I want to buy this or I want to buy that. But once you actually have skin in the game, so to speak, ⁓ then it becomes more of a reality where you want to see ROI on that ⁓ investment. So ⁓ early stages, budget ⁓ devoted, that is typically when you want to bring in a Chief AI Officer.

Utsav: There's a book which I read by ⁓ Hal, which in the title is fascinating. Questions are the answer. So ⁓ lot of times when clients have a very clear articulation of the problem they are solving, you can help them quite early in the game. Otherwise you might ⁓ work with them for a few months to actually unearth the real problem they are trying to solve.

So when clients sort of come to you and ⁓ what questions do they ask in terms of, hey, we need a CAIO or what challenges do they share with you that they are facing for which they are looking for an answer and it seems like a CAIO is a right answer for that problem. So it would be great to sort of know what kind of conversations you have at an early stage with the client. John Sukup: So again, you know, depending on what type of organization you're dealing with, ⁓ if you're dealing with say a funded startup, for example, it's going to be a little bit different situation than if you're dealing with ⁓ a small to mid-sized business or you know, an enterprise level business, depending on how you measure those by headcount or revenue or what have you.

But ⁓ from perspective, at least, because I typically try not to work ⁓ at the enterprise level ⁓ and work at that SMB and founder level. ⁓ If you're dealing with a to mid-sized business, ⁓ sometimes it comes down to ⁓ board decisions, right? They want a strategy, they want something in place, ⁓ want the leadership that's in existence to start answering questions about AI. ⁓ And again, these are some of the things that I brought up ⁓ response to your last question.

you ⁓ for startups, it's probably typically the same thing. It's just maybe different people asking the questions this could be ⁓ ⁓ So, know, VCs or ⁓ angel investors or what have ⁓ But in terms of like the typical questions, and I'm going to formulate ⁓ specifically to align ⁓ with some of the different elements included in my Chief AI Officer toolkit, which I know you and I have talked about, but we haven't talked about on the podcast yet are really four kind of high level questions that clients typically ⁓ ask.

And I know some of them might sound pretty ⁓ generic ⁓ or like, why would you ask a question like that? But the surprising thing is, is you do hear it over and over and over again. ⁓ Because I think a lot of people in the AI space want to pretend like, you know, I got AI covered, I don't need any help. But when you get behind doors, ⁓ It's like, ⁓ I really don't know what to do or what's going on.

⁓ So the first question is really just what can AI even do for us? ⁓ And this is typically where we answer that question by evaluating what the organization is currently doing, what industry are they ⁓ in, what ⁓ policies do they currently have in place, data governance? ⁓ What is their tech stack, is just basically how is their organization formed ⁓ to allow us to take, know, kind of the next step, ⁓ is how do we, you know, how do we pursue this or are we even ready to pursue AI?

I ⁓ think this is one of the most challenging questions is because a lot of companies, you know, they want to, they want to skip ahead, right? ⁓ They want to go right to the action. ⁓ Right to building something large language models and get it out in front of their customers yesterday. ⁓ But the fact of the matter is that ⁓ organization that I've ever worked with ⁓ has ever been in a state, ⁓ they already have some existing AI capability that they've been working on.

But when I come in fresh, I would say that almost no organization ⁓ is in a state that I would say is ready to pursue ⁓ The third question is, you know, how do we implement ⁓ these changes? How do we implement ⁓ AI transformation? And that's a very broad question because it doesn't just mean the implementation of the technology side, right? That part is actually probably relatively straightforward.

⁓ The more challenging implementation piece is the people change ⁓ aspect, which I know that, you know, you're very familiar with you know in your line of work, so ⁓ that's probably the more challenging piece is you know how do we get people to? ⁓ Utilize ⁓ AI within our organization and then finally you know specific processes ⁓ our organization can we tackle with AI first ⁓ people want to see quick wins. ⁓ doesn't necessarily mean that that quick win has to be ⁓ you know a massive ⁓ you know, windfall in savings or ⁓ on behalf of AI.

It just means we have to show ⁓ here's where we were ⁓ here's where we were after, here's the delta, it's positive. This typically brings up more, you know, opportunities for ⁓ projects moving forward. Utsav: ⁓ You've built a fairly structured way of approaching through your toolkit that you just mentioned and we'll put the link to the toolkit in the comments here. Could you share the three-phased approach which you broadly covered right now and identifying the right set of apportionate closing the gaps and finally doing the execution, but in context of a real-life example so that people can see and visualize what it means to run through the entire process.

John Sukup: So the three-phase approach ⁓ is, I'll outline that first ⁓ and then briefly touch on ⁓ a case ⁓ for its So the three-phase approach is, ⁓ phase one identify opportunities, phase two is close the gaps, and phase three is implementation. So just from the names of the three phases alone, you should be able to see how that ties into ⁓ the CAIO toolkit that I just mentioned. So identifying opportunities, phase one is what are the highest ⁓ ROI opportunities I can implement with AI.

⁓ And high ROI, again, doesn't necessarily mean high ⁓ income or it's not measured in dollars, right? It's measured in what is the difference I can make from going from here to here. So we want to think of ROI in terms of percentage change, not necessarily in dollars earned. ⁓ So in that we typically identify and interview stakeholders within the organization.

⁓ We prioritize ⁓ use cases based off that information, ⁓ align our executives in, ⁓ or the organization's executives in workshops, create an action plan. ⁓ And then one of the more important elements that I think kind of circling back to what we talked about earlier was ⁓ the Chief AI Officer and why would you need something like that? And one of the components of phase one is creating an AI charter, which is really kind of like a organizational constitution on what we're gonna do with AI, how we plan to use it, ⁓ ethics considerations, what we will and won't do.

⁓ And basically kind of like the atomic level document that outlines everything we plan to do with AI today and in the future. Phase two, close the gaps. So again, this is really saying here's where you are today, here's where you need to be in order to achieve some successful implementation of what we prioritized in one. ⁓ lot of this is handled by some of the ⁓ frameworks and methodologies I mentioned earlier in the Chief AI Officer toolkit.

And then these three is the implementation part, which again is pretty much aligned with the CAIO toolkit, which is ⁓ determining does it make more sense to ⁓ build something from scratch ⁓ or should we buy an existing AI solution? ⁓ When we make those considerations, what's the total cost of ownership over time? ⁓ And again, once we figure that out, how can we align that with what our intended ROI is? ⁓ So vendor evaluations, long-term road mapping, these all fall under the phase three implementation.

So again, my side, the Expected X organization typically handles that ⁓ implementation more on the technical side rather than maybe the people-focused side, which again, I think, ⁓ you're probably better attuned to and in your experience have handled more than I have, ⁓ which is fine. In terms of ⁓ case studies, ⁓ so briefly, ⁓ one case study that we highlight is working with a regional ⁓ information technology partner. ⁓ Their managed service provider, I think they had about 100 SMB clients.

12 million in revenue ⁓ staff ⁓ of a little over a hundred individuals. So pretty small and half of those individuals were engineering staff. ⁓ the organization was a, you know, like I said, ⁓ MSP ⁓ and they were ⁓ a service provider for ⁓ customer ⁓ calls for ⁓ calls. So A lot of the organization up to the point where we had started working with them, they were handling ⁓ several thousands of incoming ⁓ support calls ⁓ a day and resolution times were ⁓ relatively low, maybe falling within ⁓ industry standards, but they really wanted to shrink the gap on that.

So the first thing that we did was again, I'll put this in the context of the ⁓ Chief AI Officer toolkit, because that's what we used as part of that three-phase approach. ⁓ So we did the opportunity scan, which was kind of the first phase to say, where can AI really help your organization? ⁓ And we obviously found, based on what the organization is, that ticket triage and auto resolution were two of the main elements that we could address right away, ⁓ as well as helping them with their knowledge base.

And that knowledge base had two different elements because it could be attached to the AI systems as a grounded reference source for handling some of these pretty frequent IT support calls. We moved on to the readiness scorecard. ⁓ So the readiness scorecard, and this happens almost all the time, and this is why I'm not going to call out the name of any organization. ⁓ One of the elements that we look at is the data foundation, right?

⁓ And the data foundation was pretty weak. The ⁓ ticket data that they were getting for support tickets was ⁓ inconsistent, didn't really follow a standardized pattern. ⁓ notes, again, were all just manually created. ⁓ So no standardized plan there.

⁓ And then even their monitoring was spread across three different systems. So you can see how this even if you're not talking about AI implementation could create several different ⁓ issues. So we identified the application, so ticket triage and auto resolution, determined that they weren't really in the right stage ⁓ for building that out. So we addressed those problems that I mentioned with mostly standardization and process ⁓ optimization.

And then we moved on to determining, you know, how should we build out your ticket triage system? We actually did vendor interviews for some existing ⁓ providers that actually have those systems ready to go. ⁓ We actually landed on a provider that we were going to use. Again, I won't use names, ⁓ but ⁓ after some thorough vetting of that provider, found out that the initial costs with them, the pilot program would have ballooned to four times the original amount when taking in regular everyday workloads.

⁓ So long story short or long story long, because it's probably been going on for a while, is we reduced ⁓ their tier one tickets to a 62 % auto resolution rate ⁓ using the implementation of our ticket triage auto resolution AI system. ⁓ Average drop time or average resolution time went from six hours to roughly 50, I want to say about 50 minutes. And then ultimately the ROI and the way we're measuring it here is in terms of the ⁓ cost of working with Expected X as well as the infrastructure and ⁓ everything else that we had to put in place in order to implement.

⁓ The ticket triage auto resolution was roughly about 400 % ROI just on that alone, which like a lot, but again, know, ⁓ not dealing with dollar amounts, but ⁓ it did have a significant impact on their organization. Utsav: Thanks for that in depth and I think that sort of example brings to life ⁓ kind of opportunities exist and what kind of benefits ⁓ can get ⁓ using such a service. So if you look at this role fractional Chief Officer, ⁓ do you for companies this is ⁓ more of a transitionary role that I need the support for a span of time till I develop internal capabilities to manage it within my organization or do you think of it as a more continuous service that somebody needs?

John Sukup: Think it's on a client to client basis or company to company basis. ⁓ The goal that I've set up with Expected X as fractional Chief AI Officer ⁓ is not to come in and be the long term Chief AI Officer, because I mean, that would defeat the whole purpose of the fractional part. But I think it's really, you know, for clients to get kind of a of what it's like to work with the Chief AI Officer for a given amount of time. And then based off of their own particular situation, people that they have working at their organization, ⁓ at the end of an engagement, they'll either ⁓ decide that they need an ongoing support in that leadership capacity.

⁓ So my is to transition ⁓ and have kind of a succession plan. depending on how long the engagement went for, I typically don't do anything over two years, ⁓ but depending on how long the engagement went for, ⁓ succession plan for someone to take over as a Chief AI Officer. Or ⁓ if during that time ⁓ it has, you know, we've been able to work with CTO, CIO, CDO, or some combination of that, sometimes an organization might just divvy up the responsibilities of the Chief AI Officer and align them closer with those rather than bring on yet another, ⁓ you know, C level, ⁓ someone in the leadership team.

Utsav: So let me close with this. If you had to give one piece of advice to a CEO thinking about AI today, and just for everyone's context, I think you work more with small and medium enterprises and well-funded startups who have this critical gap of getting the right kind of support when it comes to AI. What would that advice be to the CEO? What they should absolutely do?

And one thing that they should absolutely avoid. John Sukup: Well, absolutely avoid, I'll start with that. So absolutely avoid hiring ⁓ excessive engineering headcount ⁓ at the get-go and I see this a lot see this ⁓ with organizations posting, you know, jobs or talking about it on LinkedIn is the tendency seems to be to throw a lot of money at the implementation piece ⁓ and the tech talent to build out AI capabilities. ⁓ And I think that that is something you should not do for pretty much everything that I've talked about up to this point, right?

You know, just making sure that your organization is even in the right state to best maximize the gains that you can get from an AI transformation. So don't spend a lot of money on engineers. ⁓ Bring in one medium level, low medium ⁓ experienced engineer, senior engineer, ⁓ and someone a leadership standpoint. Nowadays you can accomplish quite a bit in terms of the engineering capability without hiring ⁓ hundreds of engineers.

⁓ think there's a of organizations that say we dive code everything now and we haven't written a single line of code and that's a conversation for another time. But don't invest heavily in I would say right away. ⁓ Absolutely do ⁓ other part to your question ⁓ is a component that. about strategy ⁓ in terms of where you be with AI.

Not tomorrow, not next week, but by the end of the year, by the end of the next two years, by the next five years. Don't just create a bunch of prototype projects, throw them against the wall and see what sticks, right? You have to actually have a purpose and a focus and a plan in place. I will caveat that by saying, you know, most people have seen that the AI world is changing on a daily basis.

⁓ So making three to five year plans with AI can be challenging and you should always be open to revising those. ⁓ But absolutely start from the strategic perspective rather than going from the implementation perspective. And if the strategic perspective says we're not ready yet, we need to optimize the way our company works. I mean, you'll gain a benefit from saving you know, a lot of extra time trying to implement an AI ⁓ transformation.

But I mean, your underlying business, regardless of what you end up doing with AI in the next five years, your underlying business will still become much healthier when you've achieved optimal processes and workflows. Utsav: John, this has been a fascinating conversation. Think what stands out for me is AI is not just a technology shift. It's something more bigger.

It's something more structural. It's forcing companies to rethink how they should build capabilities. Like you just said, don't get all the engineers at get-go. Just be very conscious in terms of what capabilities you want to bring inside versus how you want to access certain capabilities outside your firm.

And the model that you are building with fractional Chief AI Officer is a clear signal of where things are headed and also a point around these capabilities were earlier accessible only to large incumbent organizations. With AI you are not able to offer it to a larger chunk of mid-sized businesses which were absolutely not able to access top-end consulting talent just because it was out of reach and with the kind of model you're building it sort of helps serve them and help them strengthen their business model.

So thank you so much for joining. It was lovely to have you.