Single Player AI vs Multiplayer AI in the Workplace and Why It Matters

[00:00:00] Your agents aren't producing accurate answers because they don't have a complete semantic understanding of your data. And Denodo is solving this and solving it through semantic consistency. Through semantic consistency, your agents can start making accurate predictions in real time. So see what else Denodo can do by visiting denodo.com to learn more. But now let me introduce you to today's guest.

[00:00:34] Welcome back to the AI at Work podcast. Now, if you've ever sat at a meeting and someone announced, we need an AI strategy. Only for everyone in the room to nod while secretly wondering, what does that actually mean? Well, today's conversation is for you because, well, yes, every AI vendor will promise productivity gains, autonomous agents, digital co-workers.

[00:00:59] But despite all this, many businesses are still running critical operations on spreadsheets that were created years ago. And they seem to be passed down like family heirloom, some of them. But my guest today is Justin Watt. He's the CEO and co-founder at Switchboard. And Justin helps organizations untangle some of that operational chaos.

[00:01:24] And they do that by eliminating manual processes and building workflows that actually scale. But a quick look at his origin story. Before launching Switchboard, he worked with teams connected to companies like IBM, Uber and Amazon, giving him a front row seat to how high growth organizations design systems that work. So today I want to tackle a question that many leaders are currently avoiding.

[00:01:51] And that is, what if your biggest AI challenge isn't AI at all? What if the real problem is the process that sits underneath it? So today I want to explore why spreadsheets refuse to die. The difference between what Justin calls single player AI and multiplayer AI. And why some companies are seeing genuine business value, while others are simply accelerating existing problems.

[00:02:17] And why treating operations like a product could be one of the most overlooked leadership skills in modern businesses right now. I will unite all of that with a few back to the future references. And once you hear our comparison today, you will never look at a spreadsheet in the same way again. And we won't even need roads, because where we're going today, we won't need roads. Let me introduce you to my guest right now.

[00:02:47] So thanks for joining me on the show today, Justin. Can you tell everyone listening a little about who you are and what you do? Yeah, absolutely. Thanks for having me, Neil. I'm here in Vancouver and co-founded and co-run a serviceist company that helps companies approach AI in the right way and deploy it in the right way. And by that, I mean, there's a lot of snake oil and shall we say salacious headlines about AI right now.

[00:03:13] And our goal is to understand how a team actually works and then come into their business and reinvent their process around AI. We'll make sure that the humans are still in the loop and whatnot. And before that, I worked in kind of large scale IT and software consulting and IBM and then a product firm called Medlab, where we got to build some cool products. And I got to work with some cool teams like Amazon and Uber and whatnot. And you're right in what you're saying. There have been so many big headlines around AI and the technology always seems to get blamed.

[00:03:43] There's that huge exaggerated stat, I suspect, that 95% of AI projects are not getting ROI. But I mean, you've argued, though, that most organizations, they don't actually have an AI problem. They have a process problem. So what are the biggest workflow issues that you're constantly uncovering when you go into a company for the first time? And why are so many leaders missing, though? Because you're seeing a lot of trends here, right? Yeah.

[00:04:09] Maybe the last question first is, you know, what do leaders miss or what do they overlook? Is often they take for granted that their business is largely human duct tape and spreadsheets being passed around all day. We work with kind of that mid-market size company. So the thing that always kind of blows my mind is, you know, asking a leader who says, we want an AI strategy. We want to be AI first. And I'll say, what does that mean? They say, well, we want AI. I say, well, where or how? Well, everywhere.

[00:04:37] And then you chat with people, you know, ICs or folks in the work, so to speak. And you say to them, how do you currently work? And that's our goal is to understand if AI were to help, where would it help? How would it help? And most teams just have not documented how they work. It's often, you know, the best problem. If someone couldn't make it in for whatever reason, the next day, a lot of processes would fall apart.

[00:05:01] And then on the flip side, a lot of what people want to do with AI, they're just not ready for because, you know, the business runs on the way I attribute it. So to make it more succinct, the year that the movie Back to the Future came out in theaters is the same year that Excel was released. That's how old Excel is, as much as I love that movie. But just to paint that picture of most teams operate on Excel and they have a smattering of SaaS applications and they say, we want AI.

[00:05:28] But what is AI going to look at when you say, you know, how the business runs and is on 12 people C drive of a spreadsheet that they're passing around all day? And so that's what I often see leaders miss is not taking that step back to say, are we just applying AI or are we going to reinvent how we work? And sometimes that means that, you know, digital transformation stuff that they might have never done. But a lot of it is just not taking a step back to say, where is it actually going to apply?

[00:05:52] And when you say Excel is as old as the first Back to the Future movie, I feel compelled to say, great scotch. 121 gigawatts. Yeah. But of course, fast forward to present day, every company seems to be experimenting with AI right now. So from what you've seen, what separates the organizations are seeing that real business value from those that are just adding another tech to its existing problems and becoming part of that stat I mentioned from the MIT study?

[00:06:21] Yeah, I'm finding the folks who are successful is a they have that leadership mindset that I just shared of, you know, we're not going to boil the ocean here. There is no magic bullet AI system or technology like most technology. It's a, shall we say, gathering of different component pieces that make it successful. And so the teams that are successful, I have found kind of working firsthand with many of them is those who treat it as let's land and expand, as in let's pick one area of the business, one department,

[00:06:51] one main core workflow that we obviously see a lot of, you know, manual work or wasted time on administrative busy work happening, or just a lot of people involved shuffling stuff around and starting there and just picking, say, one workflow and trying to layer AI into that. And then seeing what works there, both on the technology side, but a lot of this is cultural, too. That's the other part of what I see working is getting people rallied around doing this together.

[00:07:15] It's not, you know, the unsuccessful teams I've seen is where management says we will dictate where AI goes, how it works, who uses it, without any real knowledge of that on the ground work that the teams are doing and how they would use it.

[00:07:29] And so that combo of the technology being used in small places to prove it out and start to get value out of and then growing from there, paired with getting the culture and the team on board with that, both the learnings and the ideation and the rollout is what leads to successes from what I've been seeing. And as a fellow gaming geek, I especially appreciate how you make an interesting distinction between what you call single player AI and multiplayer AI.

[00:07:57] So tell me about the difference there and why so many AI deployments struggle once they move beyond an individual user's productivity gains. Yeah, I think as much as we all have probably enjoyed using ChatGPT or CloudChat, a lot of people think that that is AI. And so they view it as, well, if I can throw a PDF and an Excel file in and say, analyze this and look at my emails, great. That's AI. But that's single player AI. That is just like a video game.

[00:08:27] To your point, you know, there is one person playing a game on guardrails and it takes only their input and gives only them their outputs. It's very different than, to use the gaming analogy, going online and playing Call of Duty or Fortnite or whatever it might be, where everyone is making decisions and everyone is interacting with each other. And that's how businesses run. So a lot of people get stuck in this mindset of, well, we just give everyone copilot or cloud licenses and we've got AI.

[00:08:53] But the way that a business really gets efficiency and use out of AI is plugging into your workflows. If you think of a lot of businesses as a line of, an assembly line of steps of different processes and different departments working together, you want to inject AI into the parts where it makes sense. But it has to know what others are doing or what information they provided or, you know, using a CRM as an example that gets fed off to a project management system, which gets fed off to an invoicing system.

[00:09:19] You know, that's multiplayer environment where you probably have many people across many departments and their AI should function as such, not be stuck in a chat that no one else can see what went in, what came out. And just as I was listening to you there, I felt compelled to go down the Back to the Future rabbit hole. And you're exactly right. I think Microsoft Excel was released September 30th, 1985. And Back to the Future, in the movie itself, they were working on October 26th, 1985. It's just phenomenal what you've uncovered there.

[00:09:49] I'm never going to see the world the same again. But, I mean, why do spreadsheets survive for so long? Because many grown businesses still rely on those spreadsheets to run their critical businesses and the processes there. So, why are they still here? What's the smartest path away from them without disrupting a business? I know I've given you about eight questions there, but you've blown my mind today. I think because they're easy to wrap, like for, call it your average employee, they're easy for people to wrap their head around.

[00:10:17] Everyone has to learn it in some shape or form at some point in their education or growing up or their first job or two. And so, it's structured data in an unstructured system as in it's not, you know, full of APIs and connects to everything that everyone else does. But you can at least provide some structure and you can learn the low-hanging fruit of, you know, it usually takes an hour or two for someone to learn the basics of a formula. But that's why they get so stuck there.

[00:10:44] The challenges, most people are, call it stuck in 1996, where, you know, Excel has not had groundbreaking features since, call it the late 90s, early 2000s. And so, that's a lot of people's technical understanding is what Excel can do. And they view that as kind of their system of record in their business. Because we often will ask a business to, when we're starting to work with them, to kind of rate different departments' technical abilities. And then we'll chat with those departments.

[00:11:13] And I cannot count the number of times where a business leader says, oh, so-and-so in such-and-such department is very technical. And you should chat with them. They help run this place. And then we chat with that person. And we're all excited, you know, to chat about modern technology and how we can integrate it with their business. And say, you know, we heard you're the best. Show us. And they know how to make a pivot table in Excel. Like, most businesses bar for technical abilities is macros or pivot tables in Excel. But I get it because it's easier to do than software development.

[00:11:42] That's why things are changing so rapidly as it's now becoming easier and, therefore, cheaper to create the actual software business needs. And I was also reading before you came on the podcast with me today, you've said companies should treat operations like a product rather than a collection of fixes. But what does that mindset look like in practice inside any organization?

[00:12:06] And how could maybe a leader listening to our conversation today begin applying product thinking to their internal workflows? Yeah, I think if you I mean, at the high level, if you read the research and studies, but to be more real about it, I hear it every day from teams and companies of process is messy, process is undocumented. The way that we work is, you know, you talk to 10 different people, you get 10 different answers.

[00:12:30] And so I'm a bit biased coming from working on so many products, kind of technical and software products for a while. But companies that kind of treat it like a product and to answer your question, as in, you know, the version of Airbnb in the app store today is not the same as three years ago. It's not the same as six years ago. But they don't just release things here and there and don't tell anyone. They do product launches and they do software updates in the app store.

[00:12:54] And so business kind of treating it like, hey, if we have a bunch of process changes, why don't we group these in a logical way and roll them out at the same time, communicate them at the same time, educate and train people on changes at the same time? It tends to be a lot more successful because that's a lot of people's complaints in their businesses. Everything changes so rapidly and different leaders make different decisions that conflict with each other. And now we're running our processes out of Excel again because no one actually knows how we do our work or things change so rapidly that we can't keep track.

[00:13:22] And so treating it a bit more like a product, you know, staging it in release cycles and communicating it in a way that everyone learns at once and has changes communicated at once to them tends to be just way more effective. When you lay it out, it sounds logical, but the practice of most businesses is not to do it logically with process change. And looking back at your career there, I mean, you've worked with teams connected to companies like Uber and Amazon before launching Switchboard.

[00:13:50] So were there any lessons from high growth organizations that you learned that maybe mid-market companies can adopt today to help reduce some of that operational friction without just adding in more complexity, which often seems to happen? Yeah, I think the thing that I've been spoiled by working with some of those companies is there's a higher bar for technical understanding. Certainly when you're working with a software development team, of course, they're going to understand it.

[00:14:15] But even the concepts of software development, a lot of non-technical teams on the business side have to learn at companies like these. And I find that that comes from leadership. You know, these are, of course, founded as companies that live online and have apps and whatnot. But a lot of the leaders aren't technical, but they've instilled this culture of having this bar that has to be met for understanding technical concepts and how things work technically. And you don't have to have a computer science degree or rocket scientist to figure out a lot of this stuff.

[00:14:45] A lot of it is frameworks and understanding. But when you all have that common understanding, then you're rowing in the same direction instead of everyone bringing their own kind of interpretation of how things work or how they should work. A lot of times, too, that's the other benefit is you get better ideas when people understand how the sausage is made. Not just, oh, well, so-and-so in accounting made a pivot table in a spreadsheet six years ago that no one knows how it works. So we just go to them to make all these updates all the time.

[00:15:12] If everyone has a more shared understanding of how things work, then you can have better ownership and discussions about improving it. And when researching you, I also read that you'd said that AI doesn't fix broken systems. It accelerates them. So are you able to share an example? You don't have to name any names, but where a company thought it had an AI problem, but what you actually discovered was just a workflow or a process issue that needed solving first. Is that something you get to see a lot? Yeah, we've had quite a few of those.

[00:15:41] There's maybe a recent example that comes to mind is we had a legal team who it was mayhem for them. They were constantly chasing all the information that they needed to confirm, you know, is this agreement we're signing too accurate? Is there any further updates? Do both sides agree? What terms are they okay with? What are we okay with? Et cetera. And so they said, we want AI to come in and help the legal team and figure all of this out. And very long story short, we got in there pretty quickly. We realized the issue is not the legal team.

[00:16:11] It's all the inputs that they get. The challenge is they have, you know, a 30-person sales team all making their own decisions on what information to gather and when. And then they have a leadership team that focuses on just revenue driving. And it becomes legal's problem to deal with all of the fallout of sales teams not gathering the information. And so we said, yeah, sure, we can apply AI and point it at legal.

[00:16:34] But if we can get some better process around the sales process itself, then half of these issues that we would solve with AI are just solved inherently with better process. And so playing that back for the leadership team, they said, oh, so what does that mean? We said, well, we have to build about half the AI features or kind of functionality that you thought. And so I think they were a bit taken aback. We're like, we're not going to charge you as much to do a bunch of silly work.

[00:16:57] We're telling you, you should probably spend half as much as with us and half the time and get better outcomes. And they agreed inherently, of course. You also often talk about operational workflows that span intake, approvals, onboarding, handoffs and multiple departments. So where do you see the multiplayer AI having the greatest impact over the next few years?

[00:17:21] And are there any big mistakes organizations and leaders listening could avoid making when trying to introduce AI into these cross-functional processes? It can feel overwhelming. There's not too much room for error there. But any advice that you would offer around that too? Yeah, I think it goes back to that start small mantra that I shared earlier of not trying to just say there's AI in every department. We're an AI first or AI native company.

[00:17:48] It's what is a department that would be successful both culturally and technically to bring it into. But more to that point, I think the multiplayer side of it, you have to have a source of truth for where this data is looking at. So a lot of times people will start whatever their relationship is with their customer, client, user, whatever it might be in a CRM. But that CRM is not connected to anything else.

[00:18:10] So you'll have other teams go and export information from there and bring it into usually spreadsheets, but project management systems, finance systems, ERPs, whatever it might be. And so not having these systems talk to each other is usually kind of shooting a company shooting themselves in the foot because then you have to have AI pointed at multiple things, reconcile data and information.

[00:18:30] When really, if you had either a single source of truth as kind of a hub and spoke model where all of your data comes into as a system or you have the systems that you do have integrated with each other tends to save a lot of headache and a lot of unneeded technical work to stitch it all together. And we've all seen so many massive changes over the last three to five years, so much so it's become impossible to predict the future.

[00:18:54] But if I did pull out a virtual crystal ball and look ahead to the next 12, maybe even 18, 24 months, where do you see AI having the biggest impact on business operations, improving business outcomes? And again, any practical steps that leaders listening should be taking now to better prepare their teams, their systems and processes for this future that we're heading towards?

[00:19:17] Yeah, I think the first thing that we do when it's not done, which is 90% of the time it's not done, is just map out the processes because it becomes very obvious to everyone involved, including those that we work with and the leaders at a lot of these companies, when it's staring them in the face, the visuals of, you know, this process that's 30 steps. We don't even need to add AI right away. You should just cut out 10 of these steps because you're repeating processes, you're creating busy work and meetings about meetings, about emails that were sent three weeks ago.

[00:19:46] So a lot of it is just mapping it out. And you can simplify a lot of your business just doing that and taking that step back. But that step back, more importantly, allows you to see where in that metaphorical assembly line of your business, AI makes sense, because it's not going to be in every single step. There's AI has gotten incredibly good technically, and it will continue to, but there's still going to need to be humans in the loop. And so you want to figure out what steps throughout processes is it useful to keep humans in the loop versus where I can handle it.

[00:20:14] So that's the technical and kind of business operation side. The other thing I don't see a lot of folks talk about is the cultural change that's going to come. We already see this in the headlines, of course, with the Pope is writing novel-length encyclicals about AI, and we've got a lot of talk about politics now getting involved. But all of that, to me, points to it's inevitable that this is going to change businesses and how people work. And I think a lot of people are scared of that.

[00:20:40] And leaders are often ignoring that discussion, partly because they don't have the answers. No one really does, but the fact that folks are feeling this is, I think, going to start to show up in retention and job satisfaction numbers more and more for those who aren't talking about it with their teams to say, even just at a minimum, we're going to figure this out together. It created as more of a team effort instead of a proclamation from on high saying, this is our AI commandments moving forward when no one really had any heads up or say or involvement or even awareness that it was happening.

[00:21:10] Well, at Switchboard, it is a software studio that builds internal tools and automated workflows for growing companies and also help them move past spreadsheets that are not ready for full enterprise solutions just yet. For anybody interested there in continuing this conversation that we started today, where should they go to find out more information about you and everything we discussed today? Yeah, our website is withswitchboard.com.

[00:21:36] Unfortunately, just switchboard.com is, as you can imagine, not available, but withswitchboard.com is our site. And then feel free to look me up on LinkedIn, Justin Watt, W-A-T-T, sharing more of these kind of thoughts around there and happy to answer any questions if anyone has that they want to reach out on there. Well, I, for one, have learned so much today around the real reason that companies still run on spreadsheets, how to replace them without having to rebuild everything.

[00:22:03] And also that great concept you raised there of single player AI versus multiplayer AI and why AI fails inside businesses right now. But there's so many big solutions out there. I'd love how you brought it all to life. And I'll add links to everything you mentioned. I would encourage everyone listening. Go to techtalksnetwork.com. There'll be a blog post with all the links and everything we've talked about today, as well as a picture of Justin so you can see what he looks like.

[00:22:29] But more than anything, thanks again for putting all this in the language everyone can understand today. Really appreciate your time. Yeah, absolutely. Thanks again, Neil. Take care. I loved how Justin was able to cut through the noise surrounding AI and bring the discussion back to something refreshingly practical. Yeah, it's easy to become distracted by the latest model, the latest platform, the latest headline. But as Justin reminded us today, AI rarely fixes a broken processes.

[00:22:57] Simply helps broken processes fail faster on many occasions. And I also love that distinction between single player AI and multiplayer AI. Because most organisations have experimented with AI at an individual level, helping employees write emails, summarize documents or generate content. It's time to think bigger than that. The bigger opportunity lies in connecting people, systems and workflows across the business.

[00:23:27] Where real operational friction exists. And there was also an important reminder here that successful AI adoption is as much about culture as technology. And the organisations making meaningful progress aren't trying to transform everything overnight. They're not trying to boil the ocean. They're starting small, learning quickly and bringing their people along for the journey. So thank you for listening to this episode of AI at Work.

[00:23:54] And I look forward to speaking with you again next time. Bye for now. Bye.