Context, Control, Collaboration: Why Capability Was Never the Bottleneck

0:00If you are not in the details around what is the existing process, then you could be extremely successful automating your workflow. And what you may find is that all you really did was automate mediocrity and you didn't get the full outcome of what's there.

0:20Matt Paige:Welcome to the Talking AI Podcast, where we talk AI with both experts in the field and early adopters. I'm your host, Matt Page, and we're here to demystify AI for you so you can get some value from it. Let's talk some AI.

0:35Matt Paige:Here's an idea most enterprises still haven't accepted. Your AI isn't underperforming just because the models aren't good enough. It's underperforming because of how work is structured around it. Messy workflows, scattered data, no clear governance, drop even the best tools on top of that, and it struggles. And piling on more tools can make things worse, not better. And Tom Scott has a rare view of this playing out as the CEO of Reich. He sees how AI is actually landing inside 20 ,000 plus organizations from NVIDIA to Jaguar to Land Rover, all while in reinventing his own 20-year-old company in real time.

1:10Matt Paige:And his case is simple. Capability was never the bottleneck. It's context, control, and collaboration. But Tom, welcome to Talking AI. Hey, thank you for having me. Looking forward to the conversation. Yeah, and I want to jump right into this. The models are better than ever to the point where you have the U.S. government pulling back Fable and then we got a taste of it and it goes away in a day by the time it doesn't go away, but you'll be metered for using it. But the models are better than ever. So why are there still companies struggling to get value from it? And I don't know if it's value, but I feel like it's value outside of like individual silos of people is what I've begun to notice.

1:50I think it's because when you really step back and you think about what is happening here, this is transformation writ large. And transformation is never just about the capability of the technology. It always comes down to the people, the process, and the tech. And the thing that is really unique about the moment that we're all having right now is being experienced by pretty much everyone at the same time. Prior technology waves played out in more of a wave-like fashion where you had the technology adoption curve, where you had the early adopters, you had Main Street, you had the late adopters coming out.

2:38So people have time to be able to absorb the change. Instead, what's happening is we're all collectively having this conversation about the technology, all while our people and processes are trying to catch up at the same time. And that's creating a lot of the friction that you're referring to in terms of how do you create that sort of recurring value that happens at scale and not just within an individual or a team silo.

3:06Matt Paige:Yeah, and we have the early adopters, their heads are spinning with everything going on. The late majority is what the heck, I'm not ready yet. And the laggards are just freaking out, I feel, in a big way. And it's the seminal book, Innovator's Dilemma by Clayton Christensen. It's like, just go back to that. I think so many organizations are facing this right now. But you talk about three different terms around this, context, control, and collaboration. And besides the fact they all start with Cs, why do you think these are such important elements when it comes to enterprises and companies actually starting to get value from AI?

3:43So context control and collaboration is the way we at Rake talk about the value of our platform in the era that we're in right now. And if you step back and you quickly define what those terms mean, I think context is pretty uniformly understood that the more specificity to whatever the topic at hand is, in a pre-AI era or in the current era, the better grounded you are in terms of the decision making around that. And the platform we built here at Wrike was always about gathering context to improve decision making. But I'd say everyone's got a pretty good understanding of context. Control, I think people are beginning to have a similar understanding of.

4:38And control can mean a few things. It's what's the audit trail? What's the governance model for users? whether they are human users or whether they are agent users or whether there's some combination

4:52Matt Paige:of the two. Quick break in the pod. I keep hearing the same pattern with companies I talk to. Pod's helping employees move faster, but in many companies, the business itself hasn't changed. The value is still trapped in isolated chats and experiments. And that execution gap is why forward deployed engineers have become one of AI's most talked about deployment models. Stay in bed with your team instead of advising from the outside. It's also why the FDE model is now central to every client engagement we lead at Hatchworks AI. As an official Anthropic partner, we embed Anthropic certified FDEs to identify high value business problems, build and deploy the solution and put governance and security around it, then transfer the capability back to your team.

5:27Matt Paige:If your cloud rollout is still mostly individual usage, check out how Hatchworks AI FDEs work at hatchworks.com slash cloud dash FDE. You can also find it in the show notes. Now back to the show. And control also goes to the heart of the cost question, which is how do you control the amount of context that you are supplying it to the models in a way that you're not blowing the economics around it? So I think there's a lot of conversation today that's coming into control and some of it's termed governance, but to me, it's all part of this control framework of how do you build infrastructure to be able to do this in a repeatable way.

6:04I think the last one is probably the least understood and that's collaboration. And it's interesting because Reich was founded 20 years ago on the idea of collaboration that you had to build to really unlock the human potential. You have to be able to collaborate. like massive exchange ideas, align on them, track progress. And when you come into a world where you are introducing all other types of users into this, and you start with the premise that there is a role for a human to it, collaboration becomes more important than ever. But I don't think we're yet at the point where that is broadly understood in the market of just how important that collaboration aspect is going to be.

7:00And it's the human in the loop collaboration. It's the insight. It's the creative process that is part of that collaboration. But to me, it's hard to really have a holistic discussion on value without framing it up under those three Cs. Yeah, it's several rabbit holes I want to get down off of this.

7:21Matt Paige:I think you hit it right. Context, that element is very well understood. But I still feel like there's this element where, yes, the models are great at taking in lots of context, synthesizing it, pattern matching, doing all of these amazing things. And it's frankly mind blowing blowing in terms of what it can do. But there is this element of shared context, having existing processes and procedures and the models being able to do those things in a repeatable way. And we have things happening like skills. We have agents that do things in a repeatable way. Magentic loops is the new hot thing that everybody's talking about.

7:59Matt Paige:But how do you think about, and whether this is you specifically at Wrike or within the clients you serve, how do you think of going from that element of individual, I'm getting lots of value from this, to entire teams are not only getting value, but it's in a consistent nature, if that makes sense. It does. And I'll give a, I'll give a simple example for a use case that, that I work with. I think it actually illustrates it. It illustrates it well. Like one of the, one of the use cases, I think practically everyone has used a model for is some form of research, right? So whether you are researching to prep for a podcast or you're a family researching for your summer vacation or you're me and you're doing competitive research on what others are doing in the marketplace, it is an extremely powerful use case.

9:04but it is often an individual endeavor where you are delving into this and you can create skills, you can create a specific project around it, but ultimately you are creating some form of artifact that is not easy to collaborate on or share with others other than the form of a final document or tax threat. So that is well understood by anyone who works in a model. And you can understand how you could create a great deal of inefficiency by multiple people spinning up the same versions of a similar research project. What I will often do is when I am doing research of that nature, I will often connect it up via our NCP connection.

9:58So take NCP is just a form of being able to connect multiple types of systems into a model, but using your APIs. And in our case, I will connect up the models that I use into my instance of write, where as I create research, I will extract the key areas. And I'll use that to track both context and collaboration with my senior team. So let's say I'm digging into a thread on competitive features in my space or a go-to-market move that I'm seeing in the marketplace. I can use this to set up a structure within Rake where I've got a project, I've got the shared research, which is that context. And then I'm actually assigning actions and accountability on that basis where we're able to look at it.

10:51When I then come and pick up a future use case as a follow on, I can instruct the model to go back in there and look at that prior project or task that I've set up, use that to inject into the context window rather than spinning up all this all over again as a research thread. That's a really good example of how I see a way to tie some of these pieces together. And that's, it's a very different conversations with customers on the types of workflows that they are utilizing. But that is a common example that I use when talking to them as a way to take something they understand both at the work management level, as well as at the model level and say, look, here's a way to make this more powerful for you to be able to get value.

11:40Yeah.

11:40Matt Paige:It's like things that you do more than once or more than one person is doing. Those are great opportunities for figuring out, okay, how can we automate this? and not do the same thing over and over again, like you mentioned the research task. That's an artifact that should exist and should be referenceable in a way. But I'm curious, you mentioned MCP. Curious your take here, because I've experienced this personally, where I'm a consumer using different products, and every product now has their version of AI or agent or co-pilot or something like this, right? And from my standpoint, it's, man, do I really want to use 30 different agents across 30 different products?

12:21Matt Paige:Or do I want to essentially bring my own agent, kind of like BYOD back in the day, where I can bring my own agent to the party? And the way of doing that is via MCP. But I'm curious, how do you think about this strategically? Because there's value on both fronts, where there's agents that are native in the platform itself and users are interacting with those. There's obvious value of being able to connect via MCP so users, agents can reference the system of record. I feel like there's this third pillar of almost agents that are geared towards interacting with other people's agents versus humans necessarily.

12:57Matt Paige:They're basically making the experience stupid simple for somebody else's AI to interact with that system of record. This is all very nebulous and moving very fast, but just curious how you think this is playing out strategically and maybe where you think it's going to land in the future. Yeah, this is a really good example of something where I had a going in point of view that has been challenged in conversations that I've had with our customers. My going in point of view was that there would be a cleaner segmentation of how customers thought about that question that you just posed, which is, is it a bring your own model discussion?

13:47Is it a use the agent inside the existing application? Is it the, we just want to keep using the application as it is today? And I find that sometimes, in fact, frequently I'm having all three of those conversations with the same customer and often the same person where the, and I'll tell you the common, the commonality is that everyone wants to see a more open architecture in terms of what you are able to do with a platform. And some companies are referring to this as a headless model, but whatever you want to define it as, it's how do you expose more and more of the attributes of your platform through API is a way to expose tools for that MCP context.

14:46Correct? Yeah. That is a common thread on practically every customer I talk to is that there is a great deal of interest there. And that is something that we are investing in our platform to continue extending that capability. The next part is I do think there are a lot of customers that are going to want to bring their own model and bring their own agents. And they require that in terms of working with that architecture that I was just describing. And that's a function of the fact that a, a lot of the bigger customers have relationships with model providers where they've got tokens that they've already purchased or broader, broader tool sets that they are trying to build out.

15:44And we're going to work with them and enable that. The other point though is you can still create a better experience inside the application for some use cases. And the reason for that, think of it as creating a more custom harness for what you are doing inside RIC or whatever your SaaS application is. And on some use cases that are just very core to the workflows or the use cases that you are supporting, I expect that you are going to see those agents be better suited for some of those use cases. And that is where we're spending our time as well. And by working, we're still going to be leveraging those same tools that I was talking about in terms of opening up from an API standpoint, because we'll be calling on that same thing.

16:44It's just, we will be customizing it more within our platform to be able to be able to deliver that value. And so I see this as something that we've talked about for a while publicly and it's that you got to meet the customers where they are. And I've got what I want to build. And then I've got to match this up with some of the constraints that are out there. The way we test this is my senior leadership team and I really prioritize getting in the room with customers whenever we can, telling that three C's story, because that makes a lot of sense. And then testing with them of where are you in your journey?

17:25What are you trying to maximize as an organization so we can understand where do we need to be as an organization to remain valuable to our customers?

17:36Matt Paige:Yeah, I just saw a post you did on LinkedIn a few days ago. So you were just traveling for 12 days, literally doing what you were just talking about, visiting customers across Europe. But I thought it was interesting. The hard part isn't the strategy, it's the execution. So in talking to those customers, I think this is valuable for the audience because I think a lot of people are feeling this. What do you think is, I think, causing that gap, but I think more importantly, how are those that are successful traversing that gap from strategy to execution, like the people that are doing it well, is there anything, any attributes about them, any things that they're doing to go from strategy to actually making something real?

18:19I think one of the key success factors that I've seen and I have experienced is the notion of being hands-on. And I've written about this. I've talked a lot about this, both internally and externally and externally to write. It's that because so much of this is new and because so much of successful process change is in the details. I feel that where I see success in translating strategy to like execution, it's when you have leaders that are right in it with their teams and are able to demonstrate that sort of credibility and authenticity of what they are trying to do. And it's because Because it's really easy to say, hey, I read on X yesterday that this particular company is getting 100x for what they're doing.

19:24Can you make this happen for me tomorrow? Versus having the credibility to say, I built this thing over the weekend. It's really interesting. Can you explain to me why we can't do this on a broader scale? Those are two very, very different conversations. And the answer may still be, can't do it. Let me explain to you why. But getting in there and really feeling where those tensions are comes out. And I'd say the other area where this really comes out is when you get into the details around process. Because if you go back to what we were talking about at the very beginning, transformation is always accurate around three things, which is people, process, and tech.

20:07And if you are not in the details around what is the existing process, then you could be extremely successful automating your workflow. And what you may find is that all you really did was automate mediocrity. And you didn't get the full outcome of what's there. And so again, that really requires you to get in and understand what are we doing? What does it look like for simple purposes? What does it look like on a piece of paper? And where does it flow? Who decides where is the value added step? And then at that level, you can say, we can take the big ideas and turn them into a repeatable play.

20:49Matt Paige:Yeah, I think to that point you mentioned, and folks need to be wary of this because, yes, you can map out the existing process. And yes, you can figure out, OK, let's automate these pieces with AI. But should you even still be doing it that way in this new era? That's the biggest question that a lot of people are missing because you have this, you have intelligence. It's like literally intelligence that you have at your disposal. And people should really think, do we just rethink this, whatever process you're looking at holistically? I think your going in position has to be, we're not going to do that, that we are going to completely blow up the way we were doing things before.

21:29And that doesn't necessarily have to be your first step. I think your first step is to understand what are we doing today? What are we trying to achieve? are we being bold enough in our thinking? And then I think you have to look back at what you're doing and literally start subtracting at that point and say, how much of this can I take away? Not because these people were not doing a good job or at one point it didn't make, it made sense. But I do think you have to start by subtracting as much as possible as part of that. And then you can start to holistically think about what's the system design of what you were trying to build What's the minimum number of pieces that's unique there?

22:10Matt Paige:Yeah. And you're at a rare vantage point because you're not just advising clients on this. You're going through your own reinvention story, I would imagine, as a company because you've been around for 20 years. I'm curious, what does this look like for you internally as a company going through this transformation? Because there's the people side of it. There's the product side of it as well. And you're you talk about process earlier. In a lot of ways, Wrike is a process focused product and company in a sense, right? And it was very much deterministic. And then you have this probabilistic tool that comes into the mix.

22:46Matt Paige:And I'm probably throwing way too many questions out of you, but what's that been like for you leading through this transformation? Because again, like a lot of our listeners are going through this exact thing right now. I'd say like transformation is messy. no matter how you want to frame it. And transformation has to be owned by the CEO. I don't know that there is an exception to that. So when you go back and you ask about transformation here, I've been with Rike for four and a half years and I've been CEO for three years tomorrow. So four and a half years, you came in at the beginning of 2020.

23:35Matt Paige:So you were right when everything just completely changed. So I would say I have led multiple transformations here. I led a transformation when I came in the door with the founder, because I didn't come in as CEO. And I led a transformation with the founder right when I came in the door to reestablish right as a standalone business. I led a transformation when I took over the CEO spot from our founder three years ago. And I'm in the middle of leading a transformation right now as we adopt into a new era. And I think there are a few attributes that I always look to on this. And one is personal accountability.

24:24So I was making the point that the CEO has to lead it. And I've always tried to lead this business with that accountability in mind, that you have to make the call and you often have to make the call alone. But you still owe a lot of answers on why and how you made it. Because you need people to have confidence in the thought process and the willingness to follow you. My comment on hands-on plays into that as well. I am not the technical founder of Reich, but that is not an excuse of where I should be spending my time. It's just a fact. And that means that I need to demonstrate the use of technology and how it is advancing, because that is a way for me to help lead the company through this change.

25:23It's also a way for me to recruit the next generation of changemakers. So if you think about, you've got to be able to state the why change, and then you've got to demonstrate the fortitude as a leader that you're going to take them through it. And you've got to expand that capability. But as you start going through this transformation, you find out that there are all sorts of friction points. You have people that rise to the occasion and really surprise you in terms of just how much that they take on responsibility and could drive experimentation. And you get surprised the other way as well, where you have people that had been really strong contributors or even leaders for the business that just aren't ready for the moment that's in front of you.

26:14You get all of that through those transformations that I was talking about. I'd say we are in the middle of it right now. We, last year, we kicked off what I referred to as a series of sprints to help advance some of the transformation in this area. And the reason I called them that was both to simplify the number of things that we were trying to do and also put a flashlight on it. We've created employee groups across all of our teams where we're pushing experimentation. And as we see progress, we look to scale it and communicate it and roll it out further. But that's that hard part that I was talking about of taking the strategy and then turning it into execution down at the function and team level to actually generate that change to the organization.

27:09Matt Paige:Yeah. And I think what you said in the beginning, folks need to listen to you. Transformation is always messy. It's just inherently always messy. And the end of the story is never clear until you're at the end of the story. Right. So I think anybody that is feeling these pains in their organization, like maybe they're having to pull people along or maybe maybe you're getting pulled and you don't like it. I mean, it's just it's the natural way of change. I think what is it the saying goes that the two two things people hate the most is change in the way things are like it's just like there's this natural tension that always exists.