AI Isn’t the Problem Your Strategy Is

Join us as we talk with AI and technology leader Dr. Pawan Anand to unpack why so many organizations are struggling to see results with AI—and why the issue isn’t the technology itself.

From staggering failure rates in AI pilots to the pressure leaders face to prove ROI, this conversation cuts through the hype and gets to the real challenge: strategy, culture, and readiness.

Pawan shares insights from over 20 years in the tech industry and his doctoral research on integrating generative AI into agile environments. We explore what it actually takes to become an AI-enabled (or AI-native) organization, why data readiness is critical, and how leaders should begin thinking about AI as a “digital colleague” rather than a replacement.

We also dive into:

• Why up to 91% of AI initiatives fail
• How AI is compressing timelines from months to weeks
• The role of culture in determining AI success
• What leaders get wrong when rushing into AI adoption
• The future of “AI agents” and the rise of human + AI collaboration

If you’re a leader trying to make sense of AI, this episode will challenge your assumptions and give you a clearer path forward.

Join us weekly as we dig into the real stories behind work culture transformation.

Not theory. Not fluff. Just honest conversations with leaders and innovators who've been in the trenches.

Transcript

Legacy Tech Versus AI Native

SPEAKER_02 0:00

A lot of them, there is no legacy environment that there is a baggage of, right? So you can be AI native, you can start using AI right from the word go. But whereas some of the large enterprises, there are a lot of legacy technology, legacy infrastructure that needs to be AI ready.

Welcome And Guest Introduction

SPEAKER_01 0:19

At Elements of Culture, we sit down with experts in leadership and team building to explore the DNA that drives a thriving organization. Hello everyone, welcome to Elements of Culture. My name's Taryn, and I'm joined, of course, with my co-host Julie. And today I'm so excited to introduce our guest with you. This is Pavan Anan. He is an amazing thought leader in the space of technology NAI. He's been in the technology industry space for over two decades. He also is on the Forbes Technology Council. You published many articles, including one that dropped today, which I'm excited to hear a little bit about. You also recently finished your doctorate degree around the topic of an ideal health.

Pavan’s Background In Enterprise Tech

SPEAKER_02 1:11

Thank you, Dun, and it's a pleasure to be with you and Julie.

SPEAKER_01 1:16

Yes, thank you so much, Pavan. So I know I touched on a little bit of your background, but tell us a little bit more. What do you want to share with us outside those those kind of topics that I touch on? Tell us about yourself personally. Where are you based out of?

SPEAKER_02 1:31

I'm based out of the Philadelphia area and I've uh lived in uh uh Philly for about 12 years now. Okay. And uh it's been amazing and uh worked with uh a lot of large enterprise customers within telecom, media, entertainment, gaming industry uh in this time frame. And uh last few years have been very interesting with uh the whole advent of Gen AI and now with agent AI in terms of how the technology is changing the world that we live in. And on the personal front, uh getting a doctorate uh while working uh and solving customer problems was uh very, very interesting. And uh as you touched upon the topic of my dissertation, uh it was uh integration of generative AI into agile software engineering, which we can uh double-click more and talk more about it as we go along.

What Makes AI Adoption Hard

SPEAKER_00 2:29

Yeah, Pavan, I would love to kind of pick your brain a little bit because after being in the industry for 20 years um and just seeing the change and the rapid change, talk to me a little bit about what you think a lot of industries, enterprises, big, small, the United States, and and all over the world are probably facing with the implementation of AI. Like what are some of the actual difficulties um businesses are having great question, uh Julie.

SPEAKER_02 2:58

So uh let us uh you know decouple a couple of parts of the you know question in terms of how the technology has evolved and how is it affecting uh the efficiency and productivity gains uh that you know large enterprise and SMBs are uh also trying to leverage AI for. So there are two ways to look at it. Is one is of course the advancement of technology and uh it is changing by the week. Uh, there are new uh models that get released uh every other week by large organizations, which these days are called as frontier AI firms, uh, and uh how it's being used and leveraged to increase productivity. And uh I'll I'll take a couple of examples so that uh we all understand uh whether whether it's to release a new service, new offering, or a new product to your consumers, uh, which could have taken anywhere between say four to six months from conceptualization to you know getting some of the MVP release for your uh alpha and beta teams uh to test uh get the product feedback and get the get it released for your wider customer base. That cycle has shrunk from months to weeks now. So you can get an MVP done say in a matter of three to four weeks from conceptualization to an MVP, and then the rollout can be you know become shorter by say 30 to 40 percent. So essentially what it does is you're able to make money off of your new offering sooner. So that's one way to look at it, and the second way to look at it is if you're having some sustainance product uh where you are keeping the lights on, it's business as usual. Uh, there you can support and maintain those offering services and products with a much smaller team because the teams can do much more with AI. So it's not about what AI can do alone, but it's more so what humans can do with AI. That's where the real productivity is because uh you know humans have a lot of tribal knowledge of the domain of uh the industry or the organization that they have been working for many, many years. That is something that AI cannot replace.

SPEAKER_01 5:22

Yeah, Pavan, that's that's so true. And we do talk about that with um lots of thought leaders like yourself in the space of AI, that while it can be so helpful and productive, there's still a human element that is so valuable. You mentioned something that's really interesting, the quick advancement is for you know the product and sexualization tooling and increasing the tournament time by 30 to 40 percent. That that is you, that that's a huge opportunity for businesses to be able to uh increase the timeline to get to that compatibility, right? From IT and interacting with collective leaders like yourself. Do you feel if you are attended businesses right now, do you feel like a lot of businesses are into place to be able to take advantage of that opportunity with a quick turnaround time? Because I I think there are some that are a bit more advanced for sure, maybe some of the larger organizations, or maybe there are smaller startups where they can pivot faster. But what's your thought on that?

AI Readiness Starts With Data

SPEAKER_02 6:30

Oh, great question. Uh so uh if you if you look at uh that couple of terminologies here, right? So one is AI-enabled, AI powered, and then the third one is uh AI native. So the startups that you talked about, and a lot of the uh data quality, you know, underlying data foundation, data layer uh has to be prepped so that it becomes AI ready so that the AI models can be accurate and able to produce the results that the organizations are hoping for. Because there's a famous saying in AI, garbage in and garbage out. So if the data underlying data is not ready, then the output will not be accurate. There could be bias in response, it could the models could hallucinate just because the underlying data is not AI ready. So a lot of uh in I would say last two to three years, a lot of investments in terms of time, engineering effort, uh has and and strategy have gone in to make sure that the underlying data is AI ready for the models to consume and then uh get your workflows uh AI

From Months To Weeks With AI

SPEAKER_02 7:43

powered. I'll I'll take an example, uh Taryn, right? So last year at IBC in uh Amsterdam, uh there was a very interesting case study presented, and this is from the advertisement industry where right from conceptualization to getting a product out to the market is like hey, you know, let's we are launching a new shoe. Let's take an example of a shoe launch wherein it has X, Y, and Z features which we want to highlight in in that in the ad. And from the time you conceptualize the whole thing, you write the script, then you hire the actors, you get the camera, light, sound crew, you book the uh the stadium or you know, indoor place where you want to shoot, and then once the shooting is done, then you have your post-production, then you do editing, you know, sound mixing, all that. So typically, uh on an average, something like this for a 30-second ad would have taken anywhere, say 18 weeks from conceptualization to delivery. The the case study that was presented, it took them seven days to do it. So that's the power that we're talking about. But again, that is just one off example, it's not a generalized outcome that we are seeing all across, but the number that I gave you is true, like of 30 to 40 percent increase is is in a lot of different pockets, right? So where it's more uh text enabled because there are different uh forms of output that AI can produce, whether it's text, whether it's audio, image, video, the first form that uh we all witnessed was text. So where it's reading documents on for legal or for document summarization or for code inputs, coding, those are areas where we see a lot of uh I would say productivity improvements, but when it comes to some of the other areas, uh there is still a lot of catch-up to do. But again, the whole point of AI is intelligence, right? The more you use, the more intelligent it becomes, the better data sets we have to train the models so that the future models are bigger, better, more accurate, and uh consume, say, less power to run.

SPEAKER_00 9:54

So we usually um have discussions um with other enterprise leaders and startup companies what do you think that some companies are doing wrong with AI? Because we've heard some horror stories and some bad um bad experiences with AI, and it's not everybody, but we've definitely heard that. And so in your expertise, like what do you think, like what are the things that the companies are doing better with AI? You know, what are some assumptions possibly they're making in regards to that? Because it seems like some of them really want to get into AI, but um when implementing or trying to implement, that is not working, it's not happening, they're losing tons of money. Where do you think the breakdown is? Like, what are you seeing within the field just as a person um who's an expert in that field?

Why Most AI Pilots Fail

SPEAKER_02 10:44

Sure. Great question, Julie. Uh uh almost a quarter back, there was a uh an article that was published uh, and I can send you the link offline, which uh said 91% of POCs or pilots or experimentation uh fail for AI. And the simple reason for that is uh a lot of organizations uh initially started doing AI because there was a lot of FOMO that hey, a lot of my competition and others uh in the industry are using AI. Let's also use it, right? So it's it's it's okay to uh think that okay, I'm losing out on a great opportunity to use uh uh you know the latest and the greatest technology, but if not done the right way, it will not re yield results, right? And that's true for any technology or any project or any you know large digital transformation that we're talking about, whether it was mobile adoption, whether it was cloud adoption, there are a lot of organizations that are still you know finishing up their cloud transformation journey. There are a lot of organizations that still run 20, 30-year-old legacy code this on legacy environments. So, in order for AI to be successful, it's very important that your foundation is laid correct in terms of what data your AI models or systems will consume, what's the expected output, and what's in it for the stakeholders, you know, we humans who are you know using the the system, designing the system, upgrading the system, maintaining the system, right? So, as to what the role of AI there is, right? And the interestingly enough, there are two aspects uh that I want to touch upon.

Onboarding AI And Proving ROI

SPEAKER_02 12:30

Number one being uh, and and I wrote about this, is uh AI should be treated as a digital colleague, right? So there has to be proper onboarding, the way you onboard a human uh colleague, you know, when you're uh somebody he or she is joining your team. Uh similarly, there has to be expectations of the role what AI would do, what's the input, what's the output, what's the uh what is what is what what will others do around AI. So that uh when when there is in future an org structure, you might see that there are a few humans in the team, and then there are a few AI agents in the team who have a defined role and a defined outcome that is expected out of the team, and then that's broken down between humans and agents. So that's the whole terminology of digital colleague that I had talked about. And the second thing that is very, very important is that you need to understand that running AI models and infrastructure it's not cheap. So a lot of uh conversations that I've had with you know C-suite and technology leaders is CSO CFO's office is asking them, hey, what's the ROI? I'm okay to invest, say, you know, let's hypothetically take an example of 10 to 12 or to 15, 20 million dollars a year, but what is it gonna take for us to have the ROI? How and when do I make 50 million on top of the 20 that I'm investing, right? So the whole concept about ROI is also playing a big role in making the right investments, right? Whether you are a small organization or large enterprise, this holds true because you know a lot of them could be privately owned or publicly owned. You need to uh be answerable to your stakeholders, right? And and shareholders for that matter.

SPEAKER_01 14:20

Well, Pavan, um, there you've given us so much to think about in what 10 minutes. And I have so many questions that are floating around my head, and I'm sure other listeners that are um joining us for this conversation are probably thinking about questions that they would want to follow me too. But a couple of things. You were talking about the digital calling, and I know that you know, some leaders that we've talked to recently, the culture that they've created in the organization is you know, no AI idea is a bad idea, let's just kind of follow in. And then you have some companies that are a little more reserved and are like, well, what's the RLA gonna look like? And so we're thinking kind of bold. Um, but I like the fact that you've talked about that you really have to look at the the um AI agent as a digital analytic. There has to be proper onboarding because if you don't, like what what extent is it really going to lead to, right? I mean, we that's the same thing for people as human beings when you onboard an employee. If you don't have proper onboarding, it's not going to be a great success or a great fit for the team. So many good things you touched on there. I want to ask you before we pressed record today, you said you actually had an article published today with Forbes. So tell us a little bit about that.

The Internet Of Agents Idea

SPEAKER_01 15:33

What was the topic coverage?

SPEAKER_02 15:36

Yeah, uh, so uh there um interestingly, I talked about Internet of Agents, right? Uh, because when we talk of AI agents, uh the first construct is that the AI agents work within the companies, right? That you you might have an AI agent uh in your customer service team, they might be part of your engineering team doing you know new product development, so on and so forth. Similarly, now there is going to be an ecosystem wherein agents from one organization talk to agents from other organizations, right? Uh, because we humans do. Because if somebody who is who has sold a product or is selling a platform to an organization, what and how do you make sure that there's the best possible implementation and integration within your customer ecosystem? Now, if there's an issue, you typically raise a ticket, you talk to your supplier, and then they have some engineers assigned, and then engineers from your organization come together and then you fix an issue so that you know it's resolved and you're able to use the product for the best possible outcome. Similarly, if there are an agents on your AI agents on your team and there are AI agents in other organizations, how do they interact in in the coming future? So that is something I talked about, the whole internet of agents uh philosophy. That is something that we are gonna see pretty soon because there is uh thousands and thousands of agents that are gonna work with the humans as digital colleagues in order for organizations to achieve outcomes.

SPEAKER_00 17:16

Yeah, I mean, I don't know that we've ever had that perspective of digital colleagues. I think that's one that I think people need to take into account of how to work alongside AI. Because originally I think there was a lot of fear about is this gonna replace my job? Right? Like that was a fear that was going on. So I think digital colleague is a good um way to for us to kind of look at like how to work alongside AI. I'm gonna change gears a little bit because you finished your doctorate degree and I we had brushed a little bit about your thesis. I would love for you to talk a little bit

Doctoral Research On AI In Agile

SPEAKER_00 17:50

about the topic of your uh thesis and your results, if you don't mind, because I think it's very intriguing. Um, and congratulations, by the way, on that. But yeah, can you tell me a little bit more about um the why behind why you were going after this topic as well? But also what did you find?

SPEAKER_02 18:10

Sure. And uh thank you. So it was uh a great experience uh you know doing the whole doctorate and uh uh during the research, and by by the way, the topic is integration of uh generative AI into agile software engineering, a very, very timely topic, and uh a lot of it was fueled from the conversations that I've had with the customers across industries that resulted in finding out that a lot of people are thinking about it. So let's get a little bit more empirical and do some more digging around and find what's what are the things that organizations are doing. So uh it was a you know two-method, multi-method approach of both uh qualitative as well as quantitative research, wherein uh I did uh over over two dozen uh interviews of technology leaders across the globe from uh different industries, uh, software, uh services, banking, telecom, uh software platform, uh some small organizations and some very large enterprises as well to get different perspectives of small versus large, uh, and also some you know AI chief AI officer role, some you know, CIO, CTO kind of role. So it gave a perspective of you know how organizations are thinking about AI top-down in terms of strategy, how to execute, what is the outcome, what are the what is the ROI that they are expecting. Now, the second part of the research where we did uh where I did the survey of uh engineers across the globe and again at different levels, coming from different education backgrounds, wherein they have used AI in some shape or form. So we did a before and after study where we are able to do a like a time in motion comparison of what was the time required, effort required to do a certain task without AI and with AI. And then you are able to slice and dice in terms of uh organization culture because some of the organizations are early adopters of technology, some of the organizations want the technology to mature, they are able to observe what some of the other organizations have done, what's the learning so that they are they do not spend larger cycles as compared to some of their competitors and peers. So there's different strategies of organizations out there, and also when you triangulate that with the amount of experience that you know an engineer has using AI, uh, it was very interesting to see the results, right? So,

Culture And Experience Drive Outcomes

SPEAKER_02 20:59

one thing that was very surprising, and I'll come to the uh uh the expected ones, but let me talk something about the surprising uh outcome that uh culture was organization culture was one of the key drivers in defining the outcome of AI. When I went into the research, you know, including my mentor and my entire dissertation committee, everybody was surprised to see that how can a technology such as AI, the outcome of it, uh one of the pivotal angles there is organization culture. So that was negatively impacting. So, and this was uh uh last year, March, when I presented my defended my dissertation. So the research uh and the survey was done uh towards end of uh November, December, and of 2024 and beginning of January of 2025. So back then, as you said, Julie, the the fear was there that am I gonna lose my job right so uh now a lot of it is getting more and more uh streamlined so that uh uh so that now there is expected outcome of using technology and the whole concept of uh digital colleagues is something that is uh making sense and now we are seeing more adoption uh more and more organizations using uh AI as compared to say a year back and if you compare you know a year back from then now coming to the expected part we did see that there was uh productivity improvement and efficiency gains while using AI and uh one interesting find there was that people who had more years of experience their productivity gains were much higher than some you know the engineers who had less years of experience so which kind of showed that if you know how to use the technology you are able to gain much more out of it somebody who is you know relatively less experienced in the field of your work.

Five-Year Outlook For Agentic AI

SPEAKER_02 23:11

Yeah that that makes sense Pavan I want to ask you you know you just finished your doctorate degree you are on the Forbes Technology Council you know you are what I would consider above pro level when it comes to AI technology integration and we're so uh honored that you would be a guest here on the Elemental Culture podcast show you mentioned early on in the conversation that in the space of AI and technology and as someone who's been in the industry for more than two decades that AI you see things changing so quickly even weekly right so what does five years down the road look like uh for someone who has really immersed themselves in the space of technology and AI what what do are you what kinds of things are you anticipating us seeing as a culture as organizations a great question Tarin and uh uh I I call myself an avid learner so I I just try to uh learn given any opportunity so as a as more and more advancements are made in in the field of AI we are gonna see much more uh you know and by by the weeks by the months we are gonna see new nomenclature new terminology getting coined so uh now we're we are all talking about agent tech you know there could be autonomous system or autonomous AI that could uh come uh in near future and uh again there is human in the loop and then there are some decisions that AI could take with you know guardrails defined on what should be uh the decisions so those are things that we will see in near future and all of it is getting towards AGI which is your artificial general intelligence which is a few years or maybe a few decades still away from we don't know yet but one thing is for sure that we will see more and more organizations enabling their workflows their processes uh and their product and engineering teams with AI so we'll see much more wider adoption and consumption and uh we all are seeing massive rush for constructing data centers right so those will also come to fruition and we'll see more adoption uh so my thought there is that uh there will be much more collaboration between humans and AI so human colleagues as well as digital colleagues there will be much more collaboration we will uh you know become experienced at uh using AI so uh the faster we do the better we become sooner yeah that that's so good I think you mentioned too you you're an avid learner you're constantly learning and I think in a time where there are so many changes going on we we kind of have to be that way um and if we're not we're falling behind you mentioned before we wrap up Paulan you also mentioned that you're a father um you have two children um you recently uh finished your doctorate degree you're constantly learning you're publishing articles um how do

Balance, Mentors, And Pivoting

SPEAKER_02 26:18

you what kind of advice do you have for leaders that are out there how do you keep everything balanced in a world that seems to be moving so fast all I can do uh is try and uh then uh things fall into place and uh we just have to plan better in terms of uh what we want from our uh you know personal life professional life and uh and and if something doesn't work pivot right we have to be nimble agile in terms of what we are trying to see if there are results if those are things that we like right continue to evolve and if there are things things that don't work no there's always uh ways and means to pivot talk to your mentors talk to your colleagues talk to your friends and and of course friends and family are your backbone of who we all are as humans uh to you know help us mart forward.

SPEAKER_01 27:15

Cool that's so good pivot I mean keep going keep learning we have to pivot

Final Thanks And Wrap-Up

SPEAKER_01 27:19

that's so true. Havan thank you so much for joining us for this conversation it's really uh been so enjoyable learning from you and to look forward to reading some of the articles that you've uh recently published as well.

SPEAKER_02 27:31

It's uh my pleasure to be here Taryn and Julie thank you for having me thank you