Inside Wrike's Research On Shadow AI And The Future Of Work

[00:00:03] AI adoption is something that's accelerating inside every enterprise right now. But the big question I want to explore today is, are organisations really prepared for what they are unleashing? By the CEO of Wrike. And together, we will unpack new research that reveals a growing gap between AI ambition and AI readiness.

[00:00:30] And some big stats in that report too, with more than 80% of employees admitting they're already using AI tools at work. But fewer than half have received any formal training. And only a third of companies have very clear policies in place. With 42% of employees admitting to using AI tools that their company has not approved. And I suspect there's a few of you out there that are listening to this podcast too.

[00:00:56] But the question is, what happens when enthusiasm outpaces governance? So I want to explore the rise of shadow AI, the risks of fragmented adoption, and why integration and shared context might be the missing piece in realising that elusive ROI from those tech projects. And my guest will also share what connected intelligence actually means in real practical terms.

[00:01:23] And by that, I mean going far beyond the marketing language. And talk about why leaders need to move from scattered experimentation. And get back to focusing on problem solving with executive ownership and hands-on engagement. That's the secret sauce there. So if you're wondering why AI investments are stalling or struggling to scale beyond the pilot phase,

[00:01:46] I'm hoping today's conversation will offer a much better perspective on what needs to change this year and beyond. But enough from me. Let me introduce you to my guest right now. So a massive warm welcome to the show. Can you tell everyone listening a little about who you are and what you do? Well, thanks, Neil. Thanks for having me on the show. My name is Tom Scott. I'm the CEO of Wrike.

[00:02:12] Wrike is a collaborative work management software application working with enterprises across the globe. Well, it's a pleasure to have you join me today. And one of the reasons I was excited to get you on here was having come across your recent research that shows more than 80% of employees are already using AI at work. Yet fewer than half of them have received any formal training. In the past, we've heard about BYOD. Then there was Shadow IT.

[00:02:42] Now Shadow AI. So I've got to ask, so why is enablement lag so far behind adoption? Yeah, I think one of the things you often see in an emerging marketplace, whether it was BYOD or the current era that we're in right now, is you have people with that builder mindset just race very quickly ahead.

[00:03:10] They're the ones who want to tinker. They're the ones who want to get in there. And then that starts to expand. And what you find is that gap really grows in an organization. And the faster the new thing is emerging, the faster that gap that you're referring to starts to pop up. And there's always this natural tension for a leader as myself, both as a builder of product,

[00:03:38] as well as a runner of the organization of how do you strike the balance between moving really quickly and creating the creativity to explore and then bringing along the overall organization at the same time. And I think that's some of what the research is highlighting is this go, go, go, experiment, move really quickly.

[00:04:00] And then we'll follow up later with the rules of engagement, the process that's going to have to ultimately govern this. And I'd love to pull a few more stats out from that report. And for example, only 35% of companies have delivered company-wide AI training and just 34% have formal AI usage policies. As an ex-IT guy, that kind of thing makes me a bit nervous.

[00:04:26] But what risk does this create for organizations that ultimately believe they're already AI enabled, but the stats tell a very different story? Yeah, I think this does create an enormous amount of tension in an organization, especially when you're dealing with technology that is pervasive in the way this is.

[00:04:50] And pervasive, I mean like the ability to really make mass changes across an organization at the data and application layer. You know, one of the things you were talking about your time as an ex-IT, one of the things that used to protect was this concept of like a developer moat. You know, you've always been able to do extremely powerful things in an IT environment with,

[00:05:17] you know, changes on the command line where you can unleash massive destructive power if you weren't careful. But there was always this step that you had to go over. You're accessing it at the command line. You're accessing it in a PowerShell. What's different about the environment today is it's so accessible. You know, you can go in there and, you know, create some really large changes. And the more integrated it is into your system of record,

[00:05:46] the more integrated it is into, you know, your user access control, the more you can obviously do with it. And that's what raises the urgency to say, what are our policies going to be around this? You know, with some of our customers, there's obviously a relatively significant security review that you go through before, you know, they're willing to implement some of these tools. We see it emerging certainly on bigger customers.

[00:06:14] With what we're doing internally at Wrike, we run a risk committee where, you know, I'm a very big cheerleader in the organization saying, go, go. Like, I want to see you building. But then I do run a side risk committee where the head of security and I sit down with the head of product and the head of engineering and we're like, do we really want to connect this? And is this, and if so, who gets access to it?

[00:06:41] And so all the traditional questions you would think about, you know, user access controls, what's the system you're going to do that you're going to connect? Exactly what are you able to do? We're often testing ourselves to say, how can we deploy this in a safe way? What are the rules of engagement before we release it? And I expect that as the excitement phase, you know, kind of begins to level off and you start thinking about how do you solve these real problems,

[00:07:08] we will see that stat catch up more with a lot more formality of how you are actually going to release this technology into the wild inside a company. And a quick look out there, you could easily see that shadow AI is clearly on the rise. And 42% of employees are using unapproved tools out there.

[00:07:29] So from your perspective, is this a governance failure, a tooling gap, or just a signal that employees are moving faster than their leadership teams and their places of work? It's a signal that employees are moving faster than their workplace, where, you know, the media sort of exposure on this technology is just immense. I mean, going all the way back to the fourth quarter of 22,

[00:07:59] when, you know, the chat GPT first hit like the popular, you know, mindset, it's everywhere. And so if companies are not making these tools, you know, available to employees, they're taking it in their own hands to go do that because it does unlock, you know, an enormous amount of productivity. And that's why I think that the faster you're able to sort of come up with the rules of engagement for your employees,

[00:08:27] the faster that you're able to like make tools available to them, that ends up tamping down the desire to bring these tools into the marketplace because you're not going to stop it. So you're better off acknowledging that it's happening and putting a framework around so you can do it responsibly. And I think there are many workers that feel that better integration with core business systems is one of the things that is essential to realizing the true value in AI.

[00:08:56] But of course, there's a lot of politics around that. We've seen a lot of pros and cons around doing that. And also, we've seen a lot of arguments of the ROI in those expensive AI projects and struggling to get out of pilot phase. So do you think integration is becoming that hidden blocker to AI, ROI? I don't. So, yes, but I don't think it's the only aspect of it.

[00:09:21] So if you talk about what I think the blockers are to realizing value, I think, number one, there's an issue of what comes first. Is it the problem identification or is it adoption that you're trying to optimize for?

[00:09:43] Because if you're trying to optimize for adoption, you set off this scramble of looking for the problem for this tool that you have, as opposed to coming at it from the standpoint of I have a deep awareness of this specific problem. How can I architect a solution to be able to go after that?

[00:10:07] And I think that's some of the real tension that we see on some of these early implementations of just how specific is the identification of the problem. Once you identify a specific problem, then you can run classic transformation on it. And transformation is not just limited to the technology.

[00:10:29] And that was true in all prior technology transitions as well, whether it was mobile, whether it was the cloud, whether it was what we're talking about here. It's still you have to run through problem identification. How do you get top level ownership buy-in? How do you bring a motivated team that has that fluency to be the builder? And then how do you go run your communications program?

[00:10:53] But it tends to span across people, process, and tech like it always has. And that, I think, is one gate that's there. Now, going back to that question you were asking around pulling in other tools. Absolutely.

[00:11:10] And the reason for that is that the better the context you have available for a specific problem, the better the solution identification is going to be. And so, you know, we're, we do the same thing, you know, here at Wrike.

[00:11:29] And so, you know, when you build the agents that we launched, I think one of the power of the agents that we're able to deploy within the Wrike sort of infrastructure is we're building on context of your workspace. And so, that means similar tasks or projects that you've worked on were able to gather up that context. So, it's not just a simple prompt that you've asked. We're searching within that overall context to enrich it.

[00:11:57] And then that gives a better answer. If you were working with, like, multiple data sources, you know, the idea of being able to integrate those in advances, you know, it advances that cause as well. So, I think we're going to continue to see, you know, more and more integration on that basis. But that's also going to stress what you and I were talking about a few minutes ago, which is what are, you know, what are the access controls?

[00:12:25] You know, who's able to access that information? Because as you bring these things together, you increase the risk of your private data becoming public. And so, you have to be extremely thoughtful about how you bring this data together. But I do believe it is ultimately a limiting factor of how you develop that trust to get the ultimate ROI that you want. Well, I'm glad you mentioned trust that.

[00:12:48] Another massive stat in your research was your data showed that 96% of employees want AI tools to share context automatically. So, what does connected intelligence actually mean in those day-to-day workflows beyond just another marketing phrase or keynote speech that you might hear? But essentially, you know, customers, you know, myself included, want tools to remember, like, did we have this conversation before? Yeah. Yeah. Yeah.

[00:13:18] At its simplest level, like, if I learn something and I then come back to, like, the next, you know, instance or chat or agent that I'm running, I wanted to remember that we had seen that problem before. Or this is how we solved it. These are the individuals that it has to go through.

[00:13:42] A lot of friction gets introduced if you have to explain the problem and the prior solution every single time. Like, the amount of context just keeps building and building and building over that time period. So, you have to have something that has, you know, that has memory.

[00:14:01] One of the things that's always gotten me really excited about what Wrike does is Wrike enables people to do work at scale. And by doing work at scale is because we structure these decisions across our work graphs. So, it's all the different relationships that come together that are working on a particular workflow. So, for this particular one, it may be, you know, you, me, and this other person.

[00:14:28] For a different one, it's like a different, you know, a different triangle. But we capture that context of how work is flowing across different people. It's essentially like the human memory being captured of what's there. And so, when you go in to ask these questions or deliver this work, we're building up off of, you know, off of that context.

[00:14:52] And if you extend the example a little bit further on some of the other integrations that you would do, that further enriches the context. And so, I'll give you an example of a use case that we've been developing inside Wrike is we run Wrike on Wrike. So, as our head of revenue rolls up, you know, revenue forecast and puts commentary on that, we're often building that inside Wrike.

[00:15:22] We started integrating other data sources into that. So, we capture CRM records within Wrike that we're pulling in through a connector and we're putting intelligence against that. And that intelligence is essentially to analyze week over week trends, analyze other insights on the data, help drive that into the commentary. But we use Wrike to essentially assign that accountability. So, great, you found this data source.

[00:15:51] You found this data insight. What are you going to do with it? Like, who is now going to be responsible for taking the action to go remediate or, in the other case, extend the gain? So, to me, that's an example of you identify the insight or the intelligence. You expand it. And then you're using that system to be able to assign the specific accountability and follow-up. And then you can trace that throughout.

[00:16:18] And I'd love to give people listening somewhat of a valuable takeaway here. So, if we have a listener who might be feeling that their AI initiatives are kind of stuck in pilot mode or just struggling to get it out of pilot mode, what practical steps should they be taking this year to move from, I don't know, a series of scattered experimentations to structured, scalable adoption? Any advice you'd offer here? Yeah, so, I think there are several things.

[00:16:45] And I'm drawing both on, you know, my personal experience of what has worked inside Wrike and also, you know, what we've been seeing and talking to our customers about it. And so, I think the number one thing I can point to is specific problem identification. It almost doesn't matter what it is.

[00:17:09] And I've had this conversation with every functional leader inside Wrike where I'm like, look, get specific. There's often this tendency with the new technology to take the boil of the ocean approach. Man, we could do so much with this. Let's just go look everywhere. I'm like, no, no, get really specific. I'm like, make it personal. Take a look at like what you personally do every day and set a goal of like, I'm going to disrupt this.

[00:17:37] I drop so much time on this and take that really specific problem and start to look for the solution that you're going to apply. The second thing that I have found is like the most important is a handful of highly motivated, fluent employees. And by fluent, I mean they are enthusiasts. They are the builders. They are the tinkerers.

[00:18:04] And they often have that founder mindset. And by founder mindset, like they're interested in looking for a problem and then applying, you know, people tech and process against it. Every interesting thing that we do inside Wrike, I can go find someone that meets that criteria. Where they, and they aren't just in a product or tech org, by the way. They're in every single organization.

[00:18:31] I can go across every team that I have and I can find someone that checks those boxes where they're an enthusiast. And I almost think of them as like they're the point of the spear. They're the ones that are going to iterate, find the way to work. But what you have to do then is you have to pair them up with that top-down ownership of it. Because in this era, it has to be executive level led.

[00:19:00] And it has to be, and at least from my perspective, the more hands-on the leadership team can be, CEO included, the better. Because you're learning along with them and you're able to say, here's what I tried. What do you, you know, what do you think? You've got credibility going both ways. And then it also helps when you get to the last component, which is that communication.

[00:19:27] Because you will have a substantial number of failures that in the building phase. I mean, just think about the recent news around like OpenClaw. I think the founder of that, OpenClaw was like his 43rd or 45th attempt at building. And the others had largely been failures, but he was highly motivated. He was really excited. He kept going through.

[00:19:55] And that's the one that hit. That's pretty consistent with what I see in any innovation endeavor. And so you've got to have that mindset to keep experimenting. And then when you hit, communicate. Like share those findings across because that's how you make it repetitive and you continue to scale the overall organization behind it. And for any CIO or COO listening today, what ownership model, training framework, and onboarding approach?

[00:20:24] Anything there that you'd advise or recommend that they do to ensure that AI will strengthen their workflows rather than just create more friction? You've probably seen a lot of mistakes in your career around this as well. But any advice here to leave people on? You know, we obviously, you know, interview pretty heavily for, you know, for that mindset. And it's not just skill set that I'm talking about.

[00:20:51] It's also transformation mindset in this era and that hands-on piece. Second thing I would mention again on the CIO front is what I was saying a minute ago about hands-on. I can't emphasize enough how important it is for technical leaders to be hands-on in this moment.

[00:21:15] And one of the things that I've tried to reinforce with my own executive team is they need to be builders. And I don't expect them all to be equal to each other because some have, you know, 30 years of experience in an area. But I expect them all to be hands-on in the context of their own background.

[00:21:36] And from that standpoint, I believe that is super critical for a CIO, COO trying to lead an organization through transformation like this. It's you cannot be aware of what's possible if you are not out there banging away on it yourself to at least test the boundaries there. It'll help you in recruiting.

[00:22:00] It'll help you in driving transformation in the team, but it will just help your overall understanding of like what is possible and go back to that problem identification of like, how can we get really specific? What is this best applied to? And then we can extend it from there. Fantastic advice. And looking further ahead, anything in particular that excites you about where we're heading, what everyone's working on, what you're doing at Wrike, et cetera? Anything that you're particularly excited about right now?

[00:22:28] You know, what I'm really excited about right now is our roadmap tends to cut across three really big buckets. So we invest in our core platform, and that's a lot of what we do. It's like keep investing in why people bought Wrike in the first place, continue to make it more usable. A lot of that is like customer-driven feedback of how we extend the platform.

[00:22:54] What is next is we're doing a lot of work, what I refer to as like AI inside Wrike. And what I get really excited about here is we're not building this as an add-on. We're building this deep into the product as a way to extend why customers bought from us in the first place. So the specific application, the trust and governance, the traceability.

[00:23:24] And what I love so much about this track is we did a lot of co-creation work with customers last year that led to our launch at the beginning of this year. And just seeing these use cases start to come to fruition where people are like, I've always wanted to be able to do X. And now I'm deploying it with your agents, and we're getting almost immediate ROI from them.

[00:23:50] That is what brings people back to look at more as we continue to extend the number of actions, as we continue to extend the context, as we begin to extend this to more of the organization. That's really what gets me excited of like that customer feedback loop and then tying it into that longer-term vision. Exciting times ahead. And anybody listening that want to continue discussing some of the things that we've talked about today or learning more on the research that we've discussed as well.

[00:24:19] Anywhere in particular you'd like to point everyone listening? So you can always get information on Wrike on our website, so wrike.com. We're extremely active on LinkedIn as well, both with the company as well as me personally. So feel free to reach out across any of those mediums. Perfect. Well, I will add links to everything you mentioned, including the research that we've referenced today as well. And so many big takeaways.

[00:24:46] I think workers are aware of the gaps and the missed opportunities of fragmented AI adoption and poorly integrated solutions. They're going out there doing things on their own. You watch a few YouTube videos, you can do almost anything, you know. And I think that solution you mentioned of connected, intelligent, structured data, integrated workflows, so much good stuff coming along here now as this space matures. But just thank you for joining me and sharing your story today. Really appreciate your time. Neil, thanks so much. Greatly enjoyed it.

[00:25:15] I think if there's one takeaway from today's conversation, it's that AI success is not about how many tools you deploy, but it's about how you clearly define the problem, how intentionally you train your teams and how thoughtfully you govern access and integration. Because Thomas made it clear that employees are not waiting for permission. They are already out there experimenting, building and solving problems in real time.

[00:25:44] So that means the real question for leadership teams is whether they will create that structure, training and ownership models to better channel that energy safely and productively. And as AI tools will inevitably continue to grow and tool overload could quickly become the norm, I think there's a real opportunity here in connected intelligence, bringing people, systems and workflows,

[00:26:09] bringing all of them into alignment rather than just layering on top of more disconnected technology. If you look at your own organization, are you optimizing for adoption or are you optimizing for outcomes? Let me know. TechTalksNetwork.com. Send me an audio message, DM or just connect with me on socials. Love to hear from you. But that's it for today. So thank you for listening as always. And I'll speak to you all very soon. Bye for now.