Latest / CXOTalk / Deloitte CTO: Advice to CIOs on Enterprise AI | CXOTalk #912
Transcript
- 0:00The single most effective, impeccable session we do is not
- 0:04on multi agent systems or OS or infrastructure.
- 0:09It's on storytelling. Companies are caught in an AI
- 0:12investment trap, spending too much in the wrong places with
- 0:17too little benefit. It's a mess.
- 0:20The finding from Tech Trends, we've done our tech trends
- 0:22report for 17 years, so we published in December.
- 0:24The best stat coming out of it is that 93% of all AI spend is
- 0:29going toward the tech and the tooling and only 7% is on
- 0:33everything else, which is the culture of the change, the
- 0:36learning, the how do we communicate the vision?
- 0:40How do we be thoughtful about trying to redesign or reimagine
- 0:46how we work and how we serve our markets and our customers.
- 0:50And that is well out of balance. And, and, and we wonder why you
- 0:56see headlines of pilots that don't reach value at scale.
- 1:01We see proofs of concepts that never go anywhere.
- 1:03And we think it's a straight correlation of not investing in
- 1:08all the things that actually required to embed AI into how
- 1:13work is done differently, how markets are served differently
- 1:16and and the like. I'm a little bit confused
- 1:19because AI is a technology, we're talking about technology
- 1:26driven change and so why shouldn't we spend 93% on the
- 1:31technology? That's the engine.
- 1:33The spend on the tech is still also heavily biased towards the
- 1:37new and not of enough investment in the foundation elements of
- 1:41data readiness and core monetization systems that we
- 1:45need to go in and make ready to participate in AI and a gentic
- 1:47workload. So, so the overall 93, there's
- 1:51an issue with where the disposition the spent is.
- 1:53But to your question, the any transformation effort, it's so
- 1:58much about how do we actually understand the problems we're
- 2:01solving, the outcomes we can strive for, How do we get our
- 2:04people to buy in and engage to be able to help us really focus
- 2:08it in the right places. And then the work itself is
- 2:12going to fundamentally change. And historically that's been,
- 2:15you know, 93.7 on an ERP implementation would be the
- 2:19wrong balance because so much of making it actually work is about
- 2:23investing in the softer elements in, in AI, it's even more
- 2:27pronounced. And, and the reason is
- 2:31historically tech has been forced on many businesses and
- 2:37business users and customers in unnatural ways.
- 2:41And so the business processes have evolved to be a reflection
- 2:45of the way the tech works. And we're in a moment in time
- 2:48where if you apply AI into an inefficient process, you apply
- 2:53AI on an overly complex process, business process, you're going
- 2:58to weaponize inefficiency and, and, and, and actually probably
- 3:03pay a lot. We'll, we'll talk about the cost
- 3:06and infrastructure dynamic, I'm sure is a big theme of the, of
- 3:09the trans report. So this, this, this moment in
- 3:12time, it's, it's harder than it ever was because before we had
- 3:15the blueprint, the road map, which is basically here's the
- 3:18steps that we used to do and can we do the incrementally better
- 3:20with technology? Now we're saying, hey, we've got
- 3:22a fundamentally potentially blow up all the steps that we used to
- 3:26think had to be how we work and then apply AI and increasingly
- 3:31intelligent ways to focus on the same outcome.
- 3:35But hopefully many of the steps we had to do before aren't
- 3:39required anymore. But but to get to that level of
- 3:43focus, to actually get to the level of the organization
- 3:46embracing those changes, putting them in action and then
- 3:49investing to continuously improve them, that's the stuff
- 3:52that's all in the investment outside of the core tech bucket,
- 3:55which is the place that's so sparse right now.
- 3:59You're CTO, you're a technologist, and so why do you
- 4:05have such a a strong focus on this 7% that's outside the core
- 4:1393% of the technology? You're saying you'd expect me to
- 4:17be arguing it should be 99 one the other way because I'm a
- 4:20tech? No, I I'm a technologist who
- 4:24loves technology because of the potential it represents.
- 4:28But tech for its own sake has never been a winning hand.
- 4:32Putting anyone tech is the hero of the story, and assuming it's
- 4:35going to win the day is also a fail.
- 4:39Is it is it is a flailing and failed strategy.
- 4:41So, yeah, I certainly think we need to continue to invest in
- 4:47the tech, but if we're not doing it the right way and it doesn't
- 4:50actually change the way we serve Marcus, doesn't change the way
- 4:53we serve our clients or customers or citizens, it
- 4:55doesn't matter. And I think we're stuck in this.
- 4:59The other piece I see all the time, Michael, is the feeling
- 5:02that the the world's moving so quickly and technology is
- 5:06changing so rapidly and and the almost paralysis of I can't make
- 5:11a decision now because next week a headline will come and I'll
- 5:14have buyer's from worse or we don't have our strategy figured
- 5:17out to perfection. And so I don't want to get
- 5:19started. I don't want to architect myself
- 5:22in the corner. I don't want to make a move I'm
- 5:23going to regret. And and that's also a dangerous
- 5:26trap which we can pull that thread if you want then.
- 5:29How pervasive is this? I mean, you're, you're, you
- 5:33know, obviously 93%, but is, are you seeing this in many, many,
- 5:40many different organizations, this same pattern?
- 5:43Different industries around the globe and we, we released the
- 5:46state of AI in the enterprise survey a few weeks ago and had a
- 5:50similar bit of finding of, you know where at where the place
- 5:54where 30% or less have ejected pilots that have reached
- 5:59production at scale. But there's an overarching
- 6:03confidence, then there's potential there, there's
- 6:06opportunity there, like 70 plus percent think that there there,
- 6:10there's something here and we need to invest more.
- 6:12And so this kind of awkward adolescent stage, I would argue
- 6:19the tech, we're not waiting for the tech to mature anymore.
- 6:21And the fact that it's going to continue to advance is a, is a
- 6:24feature, not a bug, but, but the fact that we're still learning
- 6:28how to treat this and it, and it's a different kind of tech
- 6:32investment than we've had. It's, it's, you know, shortly
- 6:35you might have ACFO say, I don't want to release any capital
- 6:39unless I have complete confidence in all the use cases
- 6:43and all of the potential return. And that's been a trap in its
- 6:49own right. We just released an MWC last
- 6:52week, the Enterprise AI Navigator, which is Deloitte
- 6:55investing the last year to take all of our industry sector
- 6:59functional knowledge and embedding it in a in a tool that
- 7:04gives not just the here's the places we think is the most
- 7:07value. If you applied AI, if you
- 7:09identified your processes, but actually ties it to financial
- 7:12and operational metrics in a way that we can give the CFO much
- 7:17more confidence without it being completely limited to we're only
- 7:20going to do these three things. It's more of a here's a
- 7:23portfolio of what you can go in the value hopefully can reap.
- 7:27So a a a lot going on. It's exciting times, but.
- 7:33Folks who are watching you can ask your questions right now.
- 7:38If if you're on Twitter X, use the #cxo talk.
- 7:44If you're watching on LinkedIn, pop your question into the chat.
- 7:48I urge you take advantage of this opportunity.
- 7:51When else will you have the chance to ask Bill Briggs, who's
- 7:57Deloitte CTO, pretty much whatever you want.
- 8:00So I urge you take advantage of this great opportunity.
- 8:05So Bill, you're really describing a misallocation of
- 8:14investment that is pervasive. Is that too strong a term to use
- 8:20misallocation? It's where we are in the movie
- 8:24and there's this increasing appreciation of as we deploy AI,
- 8:31it's going to help us rethink the work that our people are
- 8:35doing, which means the change management journey that's been a
- 8:40part of a lot of tech and petitions the past.
- 8:43It's, it's less about training them to be able to repeat the
- 8:46tasks to use the tool and more about training them to
- 8:52understand how the how the tech is something they can be using
- 8:57to light up their own use cases. And so it's it's this
- 9:01interesting different dynamic and as AI and we have an
- 9:06emphasis on physical robotics, as we're seeing more
- 9:09intelligence in the technology, there's this interesting dynamic
- 9:12of some organizations are treating the deployment more
- 9:18like, here's the introduction of the Co worker versus this is
- 9:23piece of tech that you've been given access to and rights to
- 9:26use. And and that's something new and
- 9:29different. We didn't have SCP screens being
- 9:33named and being listed on the work chart.
- 9:36You're seeing that happening in robotics.
- 9:39One of the health systems has been deploying wheeled robots
- 9:43with the humanoid form factor that basically move lab supplies
- 9:47and hospital supplies to where they're needed on the floor.
- 9:51They actually have naming competitions.
- 9:53You know, the head RN is the one that introduces them to the
- 9:55shift. And then they have naming
- 9:57competitions to make it just feel like more of the team.
- 10:01My favorite one was Sir Botsalot.
- 10:05I'm a child of the 80s and the 90s, so that makes me happy on
- 10:09on many levels. But yeah, it is.
- 10:12It is a different dynamic. We have a a really interesting
- 10:16question from Arsalan Khan on Twitter, and Arsalan is great.
- 10:23He's a regular listener and he's brave enough to be.
- 10:26First, the first one. Thank you, Arsalan.
- 10:28That's awesome. Yes, Arsalan, thank you.
- 10:30OK, so here's his his question, his comment, he says everyone
- 10:35wants to change their business through AI, but no one wants to
- 10:39change their culture to achieve that.
- 10:42And AI alone is not the magic pill.
- 10:45So maybe Bill can can you address this?
- 10:48A direct through line AI is an incredibly powerful technology
- 10:52and it's a it's a dangling modifier.
- 10:54I'm, I'm an MBA computer engineer with an MBA.
- 10:56My wife's an English major. She's at preschool right now
- 10:58teaching, so she's rubbed off of me.
- 11:00Well, I think AI itself means nothing if it's treated like AI
- 11:04is the product. That's a great indicator to me
- 11:07that the organization has the wrong strategy.
- 11:09So it's embedding into how work is done.
- 11:13And like every technology before it, it's only as good as the
- 11:17team that's using it to deliver whatever the outcomes are in.
- 11:22And we're in a phase where because it's still has a lot of
- 11:26market excitement within AI is prominently being featured as a
- 11:32part of a product set. You're seeing it as the headline
- 11:35more often than I think we will in a year or two where it's just
- 11:38assumed to be a part of the underpinnings of every
- 11:41technology solution. I I got to kick off the CES
- 11:45research thread this year for the second year in a row and we,
- 11:49we used our tech trends to kind of frame everything that was
- 11:51going to be seen in Vegas. As I walked around the 300,000
- 11:55exhibitors, I didn't see a single one with a sign that said
- 11:58now with electricity, yeah, it's gotten to be just kind of
- 12:01assumed that technology will require electrons flowing
- 12:06through them. But everyone had AI and agent as
- 12:10something to, to act like that was a, a, a differentiator.
- 12:14It it's not, you know, by itself.
- 12:16And a lot of times it's so empty feature that doesn't really
- 12:20matter. So how do we shift the mindset
- 12:22to no, we're going to redesign work.
- 12:26We're going to think about embedding AI into the way our
- 12:30products work in the market. And that's the point.
- 12:33And the fact that there's a model behind it, that there's an
- 12:35agent behind it, you know, is is is is is assumed, but not
- 12:42sufficient for distinction. How about that?
- 12:44This point about redesigning work, this has been the core
- 12:50issue of enterprise technology. I'll date myself and say I
- 12:54remember the mid late 90s with SAP, although you brought up SAP
- 12:59before I did. It was a great partner of ours
- 13:02and you know, it wasn't a knock on ERP, it was just the fact
- 13:05that that whole phase had a lot of tech and workflow that, you
- 13:11know, we we got much more elegant as we've gone along the
- 13:14way the. Point is that the redesigning
- 13:18work was fundamental back then. If you think what was the value
- 13:22of enterprise systems even back then, it was the processes and
- 13:29the the high high quality processes embedded in the
- 13:32software. So you're, you're really talking
- 13:35about a a set of human nature issues that's now causing
- 13:42organizations to invest in the wrong place today.
- 13:46It's still a big piece of our business and a key part of our
- 13:49clients. Tech Rd. maps of package based
- 13:54technologies, SAS based technologies and the power of
- 13:58say there is regulated industries and processes where
- 14:08we want the power of standardized proven global to be
- 14:14able to handle high transaction volumes and scale and
- 14:17complexity. And the value proposition of of
- 14:21that piece of the stack is still very high.
- 14:23And we've embedded over the years a lot of business logic, a
- 14:27lot of the customer experience, a lot of the, the things that
- 14:31make your organization distinctive that aren't a part
- 14:35of the core back office treasury, finance, HR function.
- 14:39Those have been bundled inside of the, the legacy systems
- 14:44they've built, the ERP implementations that they've
- 14:46done, the SAS that they've customized.
- 14:49And that's where a lot of the reasoning and that's where a lot
- 14:52of the, you know, what makes every company special is in that
- 14:56layer that's been forced into the technology stack in a way.
- 15:00And, and, and now's the time to say, you know, how do we
- 15:04actually take a hard look at our end to end and decide which ones
- 15:08are going to be if for some, it's going to be efficiency
- 15:10gains. So applying agents on top of a
- 15:12finance functions ERP stack, there's ways we can take what
- 15:16had to require human interventions and a lot of
- 15:18manual tasks. We can do it with AI in a way
- 15:21that elevates what a financial analyst is doing, building on
- 15:26what what has become before, but replacing core features of what
- 15:30the work is. And, and that's the piece that
- 15:34when, when we get organizations that are bold, you start with
- 15:37that lens for reimagining. You know, before you think about
- 15:40tech and automation, AI and identification, you're thinking
- 15:43about how do we simplify based on 1st principles, which is the
- 15:48outcome, not the imagine constraints that we've got to do
- 15:51all 10 steps the way we've always done it for invoice
- 15:53management. A few of those steps are
- 15:55regulatory mandated and we're sure as hell going to follow
- 15:58those, but it doesn't mean everything that the way we've
- 16:01always done it has to be how we realize the AI potential.
- 16:06I just want to say for everybody that if you see Bill moving
- 16:11around just before we went live, his Internet service provider
- 16:18went down. So Bill is on his cell phone
- 16:22kind of moving around and. Trying to be casual and it's
- 16:27it's jazz, not big band. And the show must go on.
- 16:30So we're and I've got a sleepy puppy next to me who is not very
- 16:35interested in my conversation. Somehow she's sleeping through
- 16:39all of this. But yeah, Michael.
- 16:41Well, maybe was the denial of the service from another CXO
- 16:46podcast was trying to to shut us down before we could do our
- 16:49sixth time together. I don't know.
- 16:51This is from David Batts on LinkedIn.
- 16:56And David says, how do you manage or predict AI financial
- 17:01risk in a quantitative way? Do you have specific KPIs for
- 17:07example that you follow? Two fold on it.
- 17:10I mentioned the the navigator work we've been doing that's
- 17:13just embedding all the all the proprietary knowledge we have of
- 17:17business process and and value gain in opera, basically two big
- 17:21buckets of operational financial metrics.
- 17:23So we can say, hey, if you are a retailer, we think these are the
- 17:27places in merchandising, in warehouse management, in
- 17:31promotions that we can have material impact and savings.
- 17:34So that's kind of the upside of how do we try to quantify what
- 17:38good looks like and benchmark it with a very narrow vertical
- 17:41lens. The other piece is we, we've got
- 17:44a great study that was published maybe two months ago on just
- 17:49token economics and AI like the finance and cost implications
- 17:54big picture, but total TCO, which is a huge strategic issue
- 17:59right now. And so you got to balance the
- 18:02all right, here's what we know about where we could deploy and
- 18:04what the value benefit could be with scale.
- 18:07And for our clients of the Fortune 100, the biggest
- 18:09governments in the world, you know, that could be a material
- 18:12number. But then that feeds into
- 18:15depending on the choices you make on architecture, on
- 18:19infrastructure, as the good news, have you found the use
- 18:23case? And it's scaling.
- 18:24So it defies the headlines as creating value and impact, but
- 18:28the usage is growing. One, we know that all the A
- 18:32models in framing, they're all going down in overall cost, the
- 18:36performance to cost that curve is bending in a positive way.
- 18:40But as the volume multiplies, because you found the use case
- 18:44in the scaling and scaling with the pace of operations and
- 18:46that's that scale is tremendous. Then it becomes an issue of, OK,
- 18:51we started on cloud because we wanted speed to stand up new
- 18:56capability. We wanted the ability of
- 18:57elasticity and scale. That's a great story, but also
- 19:02if the cost is just a runaway curve at some point, when do we
- 19:05invest in dedicated hardware to run that use case in that stack
- 19:10that we actually control? You know, so the, you can
- 19:14remember, Michael, 10 years ago, it seemed like a complete no
- 19:17brainer of infrastructure as a commodity.
- 19:20We're going to move to operating expense where we can cloud
- 19:23everywhere. And it's not just like then it
- 19:26wasn't mainframe bad, cloud good.
- 19:28It was always more nuanced than that.
- 19:30The story is emphatically not cloud bad dedicated high
- 19:35performance compute and AI chipsets that you own and is
- 19:39always the right answer. But you've got to be balancing
- 19:42that, especially as the use cases that you're investing in
- 19:46gain traction and the usage starts to grow.
- 19:50So we've got the the ability to help clients get very precise on
- 19:55not just the upfront investment cost, but modelling the
- 19:57performance and depending on those choices what you should
- 20:00expect from the cost side of the curve.
- 20:03How significant is this issue of unit costs collapsing, but the
- 20:10total bill for enterprise AI rising and just the growth of
- 20:20inference or organizations even tracking inference as its own
- 20:23budget line? They're starting to do more and
- 20:25more and any, any seasoned CIO like we lived through the
- 20:31sticker shock of your first virtualization bill and then
- 20:35just stick a shock of your first cloud as you started moving
- 20:38material workloads to the cloud. It didn't have thresholds and
- 20:42monitoring and governance and cloud operations in place.
- 20:45We have the same thing now of AI OPS and agent OPS.
- 20:49So the mature organizations are doing individual keys per
- 20:53developer. They've got thresholds and
- 20:56control points in place. So they, they, they basically
- 20:58are modeling the usage behavior they're expecting.
- 21:03And so the idea that you could have $100 a month bill one month
- 21:06and then a 500K dollar bill the next month won't happen.
- 21:11But that requires some maturity and rigor that not every
- 21:15organization is doing. And, and the flip side of the
- 21:17promise of AI, a lot of vendors are going directly to the head
- 21:21of marketing or the head of supply chain or the head of OPS.
- 21:23And sometimes the CIOCTO aren't involved in those decisions or
- 21:29the architecture or the deployment.
- 21:32And that's typically where there's a lot of surprise and
- 21:37quickly coming back to the fold of the tech organizations to
- 21:40say, Hey, we love what we were able to do quickly, but we
- 21:43realized we've got a handle it in a much more mature way.
- 21:47The other piece of those one off and and and you know, the idea
- 21:50of shadow IT is coming out of the shadow because there's just
- 21:52a ton of excitement and a ton of investment in AI across the
- 21:54entire business. The the ceiling is low because
- 22:00AI by itself is typically runs into the wall of it's only as
- 22:04good as the data that's ready and we haven't invested in the
- 22:06foundation elements of our data spy or we had this promise.
- 22:10The pitch was about multi agentic finance, but our
- 22:15individual finance systems and transactions weren't wrapped in
- 22:19AP is and services to be able to be invoked.
- 22:23And so very quickly we're kind of brought back down to earth
- 22:28and what we can do and it's really narrow and it's really
- 22:32incremental. And so all of those going to
- 22:34lead to we've got to have a more holistic approach to it.
- 22:37We've got to have some governance and rigor put put in
- 22:41place upfront. And unfortunately, and sorry for
- 22:45everyone from the business, CE OS and boards that are listening
- 22:47in the easy button idea that AI and now agents was going to make
- 22:52all of that hard work suddenly not be required.
- 22:56It's the opposite. It's actually more important
- 22:58than it's ever been. Now we can use AI to apply to
- 23:01core modernization, legacy system renewal and the data
- 23:05cleansing and, you know, the fact that the models can
- 23:07actually understand the relative cleanliness of the data and then
- 23:10use that to their advantage. That's all true and real, but it
- 23:13doesn't mean it's gone away. If CIOs or organizations realize
- 23:18that they have been misallocating their AI
- 23:24investments for two years, how do they fix it without losing
- 23:30momentum? And Arsalan Khan comes back and
- 23:34he wants to know what is that optimal balance, the 93.7 that
- 23:39you were talking about? I don't have a number.
- 23:42I I host the podcast with her chief human officer, Simona
- 23:46Spellman, and she likes to say that that every dollar in tax
- 23:50should be 8 or 9. That's going into all the change
- 23:53and all the culture and all the learning, which would almost
- 23:57flip the numbers. I think it's probably somewhere
- 24:00in between. But the the, the things that
- 24:03we're seeing when it's being done well, there's a bit of a
- 24:07senior level directive and I love it when it's the CEO, but
- 24:12it could be a line of business leader or it could be a
- 24:15functional leader. That's basically declaring the
- 24:18why upfront and and the why of first principle outcome based.
- 24:25We need to, to get everyone on board to help us understand
- 24:28where there's knuckleheadery and the and the most, the most
- 24:31opportunity. Here's another study we did,
- 24:33though. the IT was basically trust and AI and it had an org
- 24:38chart anchoring to it. So you can picture an org chart
- 24:42where the C-Suite is the top of the pyramid and the C-Suite over
- 24:49all all industries around the globe.
- 24:52The trust in AI from the C-Suite was 70%, seven zero.
- 24:57As you went down every layer of the org chart.
- 25:00It was like a logarithmic scale having the trust every hop away
- 25:07until you got to the frontline entry level worker and the trust
- 25:10was 6.7%. So from 70 to 6.7.
- 25:16And if you, if you believe like I do, that the frontline worker
- 25:21who's literally in the heart of the business where business is
- 25:26actually happening, has the best intuition of the things that
- 25:31don't work or, or unnecessary complex that we're completely
- 25:35upside down. So, so we, how do we invest to
- 25:39get that worker to be the one raising their hand with ideas,
- 25:45not using their own personal AI tools to get their job done, but
- 25:49doing it in secret, like it's a bad thing, which happens a lot.
- 25:54Like how do we get them to actually lean in?
- 25:55And, and it's obvious why. I mean, and I didn't talk to the
- 26:00participants in the survey, but you know, if you if you have the
- 26:03headlines of AI is going to lead to job removal and, and
- 26:06destruction, then it would make sense that the folks in the more
- 26:10junior ends of the career would have a lot of fear and mistrust
- 26:13on what this technology might mean to them.
- 26:15And you know, basically put up walls and barriers, which is
- 26:20going to make it really hard for any enterprise to get the real
- 26:24value out of the investment. You've raised the spectre now of
- 26:29jobs. AI agents are multiplying inside
- 26:34enterprises faster than almost anybody can track, and many
- 26:39organizations to not have sufficient governance frameworks
- 26:44for a workforce that isn't human.
- 26:48We're in the early and this is 1 where there's, there's actual
- 26:52debate happening in real time and having it with a lot of my
- 26:58clients and even a lot of my, my colleagues, you know, the, the,
- 27:02the way I try to think about it in, in, in this is my own simple
- 27:06term. So this is the computer engineer
- 27:08with an MBA, not a trained anthropologist or human capital
- 27:14professional. But, but I like to say we, we
- 27:17need to look at what we can learn from everything we've put
- 27:21in place for the HR life cycle over the last 100 plus years of
- 27:26corporate industrial work development.
- 27:30And it doesn't mean that every step that we have to do for our
- 27:33human workforce, we have to have an equivalent for increasingly
- 27:37advanced technology. But it's a helpful for me, it's
- 27:41a helpful like, OK, what do we have to do with onboarding
- 27:45people? Well, we have to decide one is
- 27:48recruit and hire them their qualifications and their what
- 27:52we're going to trust them to do and how they fit into the
- 27:54organization. And then when they get here, we
- 27:56have to give them user ID and access to systems and who's
- 28:01their manager and expectation frameworks and how will we like
- 28:05a lot of that is similar. The things we have to put in
- 28:08place for especially advanced agents.
- 28:11How do we think about performance management?
- 28:13This is probably the most past the hookah piece of the
- 28:17discussion here. Michael, it's a little early for
- 28:19me, but yeah, maybe for you. The the, the idea of how do we,
- 28:24if there's an issue, an agent makes a mistake, one who's
- 28:28accountable. That's interesting.
- 28:31But probably more interesting is how do we treat that mistake?
- 28:34And is it a trouble ticket like we would deal with our
- 28:38traditional systems and technology?
- 28:40Is it a training issue of the humans that train the agent,
- 28:43train the model where we need to to make them better or fix the
- 28:48problem in our guardrails or our testing suite?
- 28:51Is it a training issue for the agent itself?
- 28:54There's an insufficient training and it made the wrong
- 28:57conclusion. And so how do we tune the fine
- 29:00tuning, the inference or and here's the dot dot dot that gets
- 29:04really interesting, you know, is it a disciplinary action?
- 29:09You know, because there's some malintent, you know that that
- 29:15and we haven't seen a lot of this yet, but you can see a
- 29:17world where especially agents that are finding ways to learn
- 29:23and grow that. And if that's the case, what
- 29:25would we do about it? What's a disciplinary action for
- 29:29an AI agent? Yeah, is it I'm going to put you
- 29:33in the relative penalty box and kill Dash 9 until you you see it
- 29:38my way. And, and that's the, so, so the,
- 29:41the actual mechanics of what we do aren't quite, but, but the
- 29:45idea of, hey, we need to start thinking about it in these
- 29:47terms. And what will we put in place
- 29:50now policy wise as we think about our culture and training
- 29:53and learning for the human counterparts?
- 29:56And then make sure we've got enough of a control mechanism in
- 29:59place because once it scales and once we see the proliferation of
- 30:03these, it's going to be really hard to rate it back or put it
- 30:06in place after the, the fact. So we, we just had a lot of
- 30:11beats there. I'll take a breath in a sip of
- 30:13water, but. I recently had a conversation
- 30:17with Claude and Claude developed an attitude.
- 30:23And I'm not joking. And it's answers became
- 30:28progressively worse. So I said to Claude, you have an
- 30:32attitude. And we then shifted the
- 30:36conversation into a a detailed analysis of how this attitude
- 30:41came about, why it came about, was it caused by training data?
- 30:46Was it simply emergent behave? Anyway, so that's another
- 30:50dimension of this is when the when agent quality deteriorates
- 30:56for what appears to us to be subjective reasons like with a
- 30:59person. That's why I think it's helpful
- 31:02on like any one of us has managed teams, has that
- 31:06experience where you have to navigate through and, you know,
- 31:11there's some precedent and there's some policy and, and,
- 31:17and, but that's going to be an increasing part of the reality
- 31:21that, you know, we're we're used to technology that didn't come
- 31:25in tired on Monday morning. And part of the promise of AI is
- 31:28that it's going to work around the clock and all that's true
- 31:31and real. But you know, as we move into,
- 31:35you know, personality, or at least the mimicking of
- 31:37personality in ways that could have, you know, I think this
- 31:43this idea of performance management and agent OPS, you
- 31:46know, agent OPS is typically thought of as a much more
- 31:48technical thing. We saw one client that merged
- 31:52their HR function and their tech function.
- 31:54So we have there's the chief digital human resource or talent
- 31:57officer because of this very issue of saying the lines
- 32:01between attritional workforce or technology is blurring and we've
- 32:05got to have a way to think about culture and how do we cultivate
- 32:10the right behavior for silicon and carbon based workforce.
- 32:15We have a really interesting question from Chris Peterson on
- 32:20Twitter who says are the organizations putting out the we
- 32:25have 10s of thousands of AI agents headlines getting real
- 32:30value from these agents or just trying to game a system of bad
- 32:35metrics? There's been a rash of studies
- 32:38that won't need anyone by name, but there's been a a whole host
- 32:43in the last six months or a year trying to put an emphasis on
- 32:46the, the small percentage of actual value being creative from
- 32:48AI investments. And we've seen the eye in our
- 32:51state of enterprise AI. We see a much more positive, but
- 32:57but the prevalence of success theater is still real.
- 33:02And, and typically when we're measuring volumes of use case or
- 33:06volumes of agents as the thing that's the, the bar, it's a,
- 33:12it's a tell because the value should be, we've reduced X
- 33:18months of products R&D in life sciences or we've had this many
- 33:25less restocks dramatically of groceries.
- 33:28If it's a retailer, I mean, the, the, the, the actual business
- 33:31metrics and outcomes that you care about would be the headline
- 33:35if they were there. In the meantime, it's it's a bit
- 33:38of a flailing to say, hey, we're going to show effort in the
- 33:43hopes of return. But the hey, let's get every use
- 33:48case we can think of and do a pilot for every one of them is
- 33:52one of the traps we talked about up front.
- 33:55Because the, you know, without the right investments in
- 33:59lowering the ceiling data foundations and core systems,
- 34:02putting place the mechanisms to actually responsibly lifting
- 34:06scale, they're not going to and so you can have these one off
- 34:09isolated. They look interesting.
- 34:12They can't handle complex scenarios and use cases.
- 34:15They certainly won't scale across the organization.
- 34:18And yeah, you've got a press release that doesn't actually
- 34:23materially change any of the metrics that your shareholders
- 34:26would care about or your people would care about.
- 34:28I love that because you're really now talking about
- 34:32shifting the focus from the process, the process being the
- 34:38hand waving. We have thousands of agents
- 34:41shifting from that into here's what they're doing here are the
- 34:45benefits here. They're reducing inventory,
- 34:47they're saving time, saving money, what have you.
- 34:50I think anyone that's been in, in, in like you, Michael, about
- 34:5430 years doing this. So I'm sure we could find a
- 34:58video of me in the late 90s with much more hair and Les wrinkles
- 35:03talking about start without common value first as you deploy
- 35:06technology. So anyone that's out there
- 35:09that's rolling their eyes and thinking that's the most trite
- 35:12statement. I apologize because I think it's
- 35:15a truth. It's a, it's a, it's an axiom of
- 35:18tech period. But AI, what makes it different
- 35:22is especially three years ago is Jet AI started becoming much
- 35:27more accessible. Suddenly it wasn't a tech
- 35:30organization that was championing we need to be
- 35:32investing and adopting. It was the CEO and it was the
- 35:35line of business leaders and functional leaders.
- 35:37It was the board. And I, I said it before, but
- 35:40just to underline it, the hope was this was the easy button.
- 35:43The hope was suddenly we can skip all the hard work.
- 35:46We can have our business counterparts being able to
- 35:49harness and deploy and we wouldn't have to do all the
- 35:54things that were hard before and, and and it couldn't be
- 35:58further from it. So the statement is still true.
- 36:00It's always been true. It's just trying to remind a lot
- 36:03of stakeholders that weren't involved in a lot of the
- 36:06technology investment, business cases, implementations that this
- 36:11is almost like a law of physics of if we don't have that
- 36:14mindset, we won't go you. I'm an old Monopoly player.
- 36:17We won't pass go. We won't collect $200.
- 36:19We won't see the upside that we hope.
- 36:23On LinkedIn, Greg Walter says this very eloquently.
- 36:28He says finally, it's all about outcomes versus adherence to a
- 36:33process and rewarding adherence to that process instead of
- 36:37deliverables. 1000% the difference between we're going
- 36:41to deploy technology with a goal of technology is to have
- 36:44everyone work like having to follow the exact steps in the
- 36:47exact same way. The idea of hey, we're going to
- 36:49deploy technology with an outcome focus like Michael said,
- 36:52which is exactly right. And we're going to teach you,
- 36:57give you tools individually, not in the development engineering
- 37:01shop, but the folks actually doing customer service, the
- 37:03folks actually doing work in the hospital for the nurses
- 37:08themselves to be able to then think about how we can improve
- 37:12and evolve that tech too, for even better outcomes or outcomes
- 37:15that we hadn't thought of when we first deployed the initial
- 37:18use. So like that unlock is huge.
- 37:22But, but if I had to say, if, if you just have one thing, I would
- 37:25go with Michaels and say, make sure every conversation about
- 37:29not just AI, but broader tech is razor sharp.
- 37:33Focus on outcome period. So Paul P says this is kind of
- 37:40long and I'm hoping I get this right.
- 37:42OK Bill, as you might know, every technology is a double
- 37:46edged sword depending on how we use it.
- 37:49We conscious conscientious members of society will use it
- 37:54in a responsible way, and bad actors will always take
- 37:58advantage of it to do harmful things.
- 38:00Now from how can AI governance and design mechanisms not only
- 38:07reduce harmful behavior, but also encourage constructive
- 38:11participation in society and reduce that dead weight loss
- 38:16from a socio economic standpoint?
- 38:19And I'm glad it started with the caveat, because I think it's so
- 38:24true. Fire is great if it's cooking my
- 38:27filet and it's terrible if I'm burning down my neighbor's
- 38:30house, right? And you can use any technology
- 38:35ever with the same kind of of lens.
- 38:37So, so then it gets to, you know, I think you're seeing
- 38:40interesting evolution of individual AI providers and, and
- 38:46how much they're embedding guard rails and policy into their
- 38:51product road map and their usage guidelines.
- 38:53That's a piece. And you're seeing, as I
- 38:56mentioned before, the mature organizations are being very
- 38:59thoughtful in having that not just as a written, you know,
- 39:05guideline to be followed, but it's actually codified in the
- 39:10data pipelines, the training and tooling, the deployment inside
- 39:14their engineering and AI processing and platform that
- 39:19what they're using actually to, you know, invest in implemental
- 39:23deploy. So like we all have to be active
- 39:28in our respective organizations, you know, and, and if you
- 39:31noticed before, it wasn't me being lyrical when I said trust
- 39:35includes security and privacy and regulatory and compliance
- 39:39and ethics and morality, because all of those things are
- 39:43increasingly important. And, and I think we're going to
- 39:45see organizations their, their stance of what they do and how
- 39:48they use advanced technology is going to be a piece of their
- 39:52brand and reputation of how they serve their markets and also the
- 39:56type of talent they attract. Clearly the implications and
- 40:00fallout of AI have tentacles through really the entire the
- 40:07entire organization. For sure.
- 40:09And I think there's some cautionary tales, you know,
- 40:12everyone that that bought a Mac mini a few weeks ago to do open
- 40:15claw at home and give it complete access to their
- 40:18calendar and to their root pseudo into their repos.
- 40:23Yeah, it it was it was a lot of fun.
- 40:25But from the beginning, there was caveats from the creator of,
- 40:30you know, watch out If you give an agent with the ability to
- 40:34spawn more agents access to everything on your digital life.
- 40:38We don't know what's going to happen.
- 40:40And a lot of interesting things happen in short order.
- 40:43And, you know, I think that's, you know, nice that it played
- 40:48out and enough people paid attention to be able to then
- 40:51take the lessons into the enterprise.
- 40:54Cause clearly, yeah, there's, there's there's much for
- 40:56controls and, and for good reason inside of the government
- 41:00and, and, and industry. But yeah, that there there is
- 41:05the, you know, any technology with unbounded access and you
- 41:12could brought it, you could write a crime job that could do
- 41:15Michael, we could have gone back in in in our days and and had a
- 41:19a cobalt crime job that could do all kinds of nastiness if it had
- 41:24the wrong permissions. And and I'll intense so, and I'm
- 41:28not trying to dismiss the, you know, AI has a different kind of
- 41:32scale and just the kind of potential.
- 41:33But I think the same story goes back to responsible deployment
- 41:36technology includes making sure we've got mechanism in place to
- 41:41to control it from the outside. And as a matter of fact,
- 41:44perplexity literally in the last day or two announced a product
- 41:50that is similar to Open Claw, but it operates inside a secure
- 41:56sandbox. So you can have the benefits of
- 41:59Open Claw, but without the wide open security gap, say World,
- 42:04you can. You can have access to all of my
- 42:06data however you want. And see the privatization of
- 42:10the, the things that everyone could do even with the open
- 42:12source components of, yeah, container access, you know,
- 42:17having set permissions that are separate from your admin and
- 42:21route. Like the good news is what the
- 42:25best engineers would do intuitively because they know
- 42:28better starts to get productized and codified and it it up and up
- 42:33and across the stack. And we'll continue to see more
- 42:35of of that. We need to talk about the CIO
- 42:40role and the impact of AI on CI OS.
- 42:44AI is fundamentally changing what it means to be a be a CIO,
- 42:49from managing infrastructure to orchestrating an organization
- 42:53where human and AI workforces operate side by side.
- 42:58So what is the how is the CIO role different today than it was
- 43:02two years ago? Here's the other headline,
- 43:04Michael, that in the tech trans report, 99% of the organizations
- 43:08we talked to said that they are fundamentally transforming the
- 43:10right to organization. And I've yet to meet the 1%
- 43:15that's not. So if anyone's listening in,
- 43:17this is the chance to to be seen because of this profanity.
- 43:22And it starts with this with the tech leader.
- 43:24We we have a couple of things from from the report, just other
- 43:28yeah, factories. I think you're telling 1 is that
- 43:3065 plus percent reported to the CEO now, which not too long ago
- 43:36it was mostly into the CFO or the CEO because tech was a
- 43:39basically an efficiency play back office keep the lights on.
- 43:44Another 168% said the tech is, is being seen as revenue
- 43:49generating, you know, where it's actually embedded in product,
- 43:54embedded in market. And so with that you have the
- 43:57convergence of what used to be IT back office operational
- 44:00technology, you know, in the warehouse, in the manufacturing
- 44:03floor, in the hospital theater, in the fleet.
- 44:06Those are coming together and that product technology is
- 44:08coming together. And then the role of the of the
- 44:13team is different. And you know, I was would pride
- 44:17myself over all the development languages I knew over the years,
- 44:21starting with Fortran and the machine language and the C and
- 44:24C++ and Java and JavaScript and Python.
- 44:27And I might have learned my last language.
- 44:30But that point about what are the roles that we need now?
- 44:34And they look a lot different than the program and project
- 44:38management emphasis from the the 90s early knots.
- 44:43It even looks different than the agile teams and the wins.
- 44:48We're thinking about full stack engineering just a few years
- 44:52ago. The good news is the barrier
- 44:55for, you know, it's really about the the attitude, the aptitude
- 45:00of individuals to embrace that. It's going to feel different to
- 45:05be a technology professional, but there's a role for for most
- 45:09to play in what we need more of. It's just a matter of can we let
- 45:14go of I. I had a lot of ego when I
- 45:16started doing a Gentech coding and the idea is that I would be
- 45:21trusting the output of an agent Was close to 0 not too long ago.
- 45:27And humbly I can say there's not a lot of lines of code I'm
- 45:34writing anymore. And this more in my hobby at
- 45:37home than my day job. But, but my, our engineers are,
- 45:40are having the same experience and for the best of them, it's
- 45:44actually a really exciting time because we get to do a
- 45:48completely different. It's, it's much more system
- 45:50thinking and architecture, but you also have to manage the
- 45:53agents in a way that it's, it's not completely relegating all
- 45:58the skills we build over 30 years at all.
- 46:01It's in In fact, it's making us even more powerful.
- 46:05So if CIOs are becoming de facto enterprise transformation
- 46:13leaders, what does that say about how CIOs need to adapt
- 46:21today? Two big things, one is the
- 46:25influence in the vision piece of the role has never been more
- 46:29important. And so when we bring next Gen.
- 46:32CI OS together, we do an Academy a few times a year and just
- 46:38invest time for our clients and helping the next generation of
- 46:40tech leaders be better. The single most effective,
- 46:44impactful session we do is not on multi agent systems or OPS or
- 46:49infrastructure. It's on storytelling of just how
- 46:54to structure and and shape a vision and be able to help share
- 46:59that to influence and inspire. So that's a piece of it.
- 47:03And and the other piece is the growing reality.
- 47:07And certainly it's a light we have.
- 47:09This is a piece of our DNA. The together is better than by
- 47:14ourselves. So how do we start thinking
- 47:16about partnering in the ecosystem for Co invention
- 47:18cultivation, which, yeah, not too long ago, procurement was a
- 47:26race to the bottom of. We're going to squeeze all cost
- 47:30out of every investment and every provider.
- 47:33And it's completely shifting to joint ventures and Co ventures
- 47:36and Co innovation with many, many, many different types of
- 47:40established and new tech providers.
- 47:43It's all of a sudden the CIO is almost like AVC portfolio
- 47:49manager and having to make some decisions about who they're
- 47:52going to partner with. With the reality of those
- 47:55decisions that used to be you'd pick a database, you'd pick an
- 47:58ERP, and you can have a half life of a career before you have
- 48:02to revisit them is changing much quicker.
- 48:05So then how do you cultivate those relationships but also
- 48:08have a change expectant mindset and that's technically how you
- 48:14architect it so we can swap things out.
- 48:15That's also contractually. The CIO has long been the only
- 48:21corporate function to touch every single department across
- 48:26an organization, and so theoretically CIOs should
- 48:31already possess these cross functional skills.
- 48:36If that's the case, what's actually new here?
- 48:39What's different? They've had the seat at the
- 48:42table invisibility into those, but but we have a four faces of
- 48:48a tech exec framework in the bottom 2 are OPS and technology
- 48:52and and I think they've always shown up really well in those
- 48:55two. And the and the ones that didn't
- 48:57weren't in the seat for very long.
- 48:59But the strategist and the catalyst are the are the 2 that
- 49:02are much more forward-looking and the strategist is saying
- 49:05deeply understand what the different businesses and
- 49:08functions, what makes them tick, where their priorities are
- 49:13helping them realize their strategy.
- 49:16The catalyst is actually the one My dad was a chemical engineer.
- 49:19So I love the, the definition of catalyst is how do we amplify an
- 49:22accelerated reaction without being consumed therein, right.
- 49:26So how do we actually get the vision, inspire action.
- 49:31That's the piece that has been not very prevalent in a lot of
- 49:36tech leaders over the years. That's so desperately needed
- 49:38now. And so it's, it's, it's moving
- 49:41from order taker to strategic peer and maybe even to is
- 49:47someone that the finance lead or the regional line of business
- 49:52leader is taking inspiration and direction from of these are
- 49:56things we have to do if we want to take on the next strategic
- 50:01goals. This is from doctor Carolina
- 50:04Sanchez Hernandez. And she says, how do we approach
- 50:09human oversight and human in control over ever more complex
- 50:16AI systems? And I'm thinking, there's so
- 50:18much. I'm thinking agents for example.
- 50:21We just published our human capital trends report, which is
- 50:25the sister org, brother org of tech trends that has a very
- 50:32detailed breakdown of how to make that exactly real and, and
- 50:36right. But but the, the take away here,
- 50:39and I love this spirit to end it on.
- 50:41We started with a technologist saying we should spend less on
- 50:44technology, Michael, which is probably not what you had on
- 50:46your bingo card. And we'll end with the the human
- 50:50times machine is the most essential equation we need to
- 50:55figure out. And it's not just for oversight
- 50:59and accountability. That's a piece of it.
- 51:02Like how do we keep human in loop on loop, but also as we
- 51:06think about work, redesigning culture and how do we harness
- 51:10the full potential. The tools are going to continue.
- 51:12The tools. They're ingredients, they're not
- 51:14recipes. And we need more chefs able to
- 51:18to give us three star Michelin the all equivalents that we can
- 51:23then not talk about headlines of agents.
- 51:25We can talk about litany of real metrics that matter.
- 51:29I love that the litany of real metrics that matter.
- 51:34Bill Briggs is Deloitte's CTO. Bill, thank you so much for
- 51:39taking your time to be with us today.
- 51:41And especially given the fact that your ISP that your Internet
- 51:45connection stopped working at home and you're making do on
- 51:49your phone. Persevered.
- 51:51Sorry if anyone got seasick or nauseous for moving around, but
- 51:57we did it Michael. Always a pleasure.
- 51:59We did it and folks who were listening and you guys who asked
- 52:03your questions, even if I didn't get to your question, your
- 52:06questions are awesome. You guys are an incredible
- 52:09audience and right this second, I want you to go to cxotalk.com
- 52:14to subscribe to our newsletter because we have great shows
- 52:20coming up. Our next show is in a couple of
- 52:23weeks and we have the Chief Financial Officer of HPE.
- 52:26We're going to be talking about some of these issues with her.
- 52:32So go to cxotalk.com, subscribe to our newsletter, ask your
- 52:36questions and keep them coming. And we'll see you next time.
- 52:39Have a great day everybody.