Winning With Applied AI: What Business Leaders Need to Know

Daniel Newman:
Hey everyone. Welcome to The Six Five Summit. It’s day one and we’ve got our day opening keynote now, and it’s one that I’m really excited about. If you’ve been following what’s going on in the enterprise software and SaaS market, if you’ve been following my commentary on it, you’ll see that I’m extremely bullish on its prospects. You’ve got sticky companies that have sticky customers, that have really important technologies that are going to drive productivity, growth and gains and companies. And they have widely distributed customer bases, thousands and thousands, sometimes tens and even hundreds of thousands of customers. And I’m really excited about how this is going to impact the future of business. You’re going to take important software, we’re going to add on AI and we’re going to cover that today. And the person I have on this particular opening keynote for the day is someone that I think understands this extremely well, someone I’ve come to know recently.

It’s Mark Mader, CEO of Smartsheet. Joining me here. Mark, welcome to the Six Five Summit. Thank you so much for joining us.

Mark Mader:
Thank you, Daniel. Good to be here.

Daniel Newman:
Yeah, it’s really great to have you here. It was fun to have the chance over the last couple of months to get to know you. Joined you in London, been tracking and trailing the company, doing great things. Just last week your earnings came out really good results, but we’re here today to focus on big picture, the future, the impact of software. You heard my preamble and we’ve talked a bit here and there. But you know what, before we dive in, before I hit you on the day opening tough questions, I like to say that just to keep everybody on their toes. Give me the quick rundown. Smartsheet, people have probably seen you if you watch CNBC, if you’ve been following… You guys have had a really active campaign. Maybe you’ve seen Lando winning one of the F1 races in the car with Smartsheet on it. But give me the quick kind of 30 second, one minute overview of Smartsheet for everybody out there that maybe isn’t familiar.

Mark Mader:
Yeah. So Daniel, whether it’s Lando and the F1 team trying to upgrade the car or it’s a rocket company trying to build and launch rockets or it’s an imaging equipment company trying to install new equipment at the world’s leading hospitals. The common thread across each of these is there is a process involved, there’s a program involved and there’s a project involved. And what has set us apart and why we’re in a hundred thousand plus organizations worldwide is that it is targeted at business teams who need to move really quickly. So this notion of how can we graduate with technology on the latest and greatest without being burdened with complexity and dev needs. That’s who really built the company and that’s… We’re taking advantage of now.

Daniel Newman:
Yeah, it’s a great way to explain it. And as someone that’s invested and looked at different platforms and built, we’ve been very impressed in our interactions and with the Smartsheet software, Mark. So let’s talk a little bit about kind of AI, the big trend. I mean, I don’t know if you can even get five minutes into a technology conversation anymore without talking about it, but in your opinion, I think AI provides this kind of open door to being impactful, driving business results and not necessarily requiring the same technical depth. I mean, is this what you’re seeing is sort of this double trend line of, hey, we want to incorporate AI, but we also want to do so without making it become so technologically complex that it actually creates a slowdown before a speedup?

Mark Mader:
I think Daniel, that’s where we’re trying to take people. I think what the world right now is trying to work through is this narrative of unimaginable complexity. We hear stories about multi-trillion dollar market cap companies, people procuring hundreds of thousands of processing units so they can scale these and build these very sophisticated models. And what I’m finding with a lot of executives, they’re like, I see the headlines. How does this relate to me and my teams? And I think the false narrative is for me to be successful with AI and for my teams to use it, I need to build a model. I need to tune a model. I need to hire all these resources that by the way, are extraordinarily scarce in market. I think what’s lost in all this is there’s an unbelievable amount of software coming online right now that enables you to take advantage of AI right now.

And what I love is when you introduce AI to someone in the context of something they already do. Now Microsoft obviously has a huge investment with Copilot going, but I really am encouraging executives right now to help understand and to learn what they have available to them in their already approved systems. So when we introduce AI to our customers, it’s already an approved system. Their workflows’ people already know how to do and we’re making them better. And I think that’s almost the best way to introduce someone as opposed to starting at the complex end of the spectrum, which is build infrastructure. It’s really not pertinent to most companies out there.

Daniel Newman:
Yeah, it does feel like the infrastructure build out is sort of limited to the biggest of the bit. Now again, I do think there’ll be this pivot from the sort of open widely distributed, and I kind of call them table stakes now, LLMs, that then need to be tuned and customized to be useful for businesses. And then some companies are just so big that maybe they need to build models that are industry specific. But I think a few companies are going to build sort of largely capable models and AI solutions that are going to be usable by everybody else. And if I’m hearing you right, that’s kind of what you’re saying is you at Smartsheet are sort of looking to democratize it and say, look, this is the workflow you have anyway.

So imagine if AI could power, it can do something generatively to interpret what you’re building out, or it can be used… I mean, Mark, forget that there was a whole other AI that we’ve been doing for four decades. It wasn’t generative, it related to analytics. And so how do you think about education there then? So not everybody needs to buy GPUs, not everybody needs to stand up data centers and servers. Not everybody needs to train a model, but how do you think about training and educating and empowering teams then if they are more consuming it in their existing app ecosystem?

Mark Mader:
I think there are two dimensions to it. One is this notion of education is how can you make it real for somebody. And making it real can come in a couple forms. One is using the tools that are broadly available today. Whether it’s ChatGPT, whether it’s [inaudible 00:06:16], offerings from Meta, from X, from Amazon, and that’s fine. But oftentimes people understand that those are available, but they don’t understand or have confidence that they have the permission to use them in the context of their work. So there’s an aspect of education and there’s an aspect of permission. And a university president asked me last month, she said, “Mark, what’s the first investment I should do in getting my population educated?”

And I said, “For both faculty, for administrators, from students, I would go all in on understanding the core set of tools that are in market and understanding how to ask a question of AI. What is a prompt?”

And we might chuckle when I say that. Most of humanity doesn’t understand this today. And in the absence of understanding that construct, people are frozen. So go all in on the base education. That is a huge enabler. When you have that understood, you can then take that next step more confidently. The second thing you can do is look at the tools you have today. In our case, we have a hundred thousand plus organizations who have access to four really valuable AI skills in the context of our platform today. So if you’ve had an analyst building reports and visualizations in Smartsheet manually for the last two years, and you show them that with AI, you can do that in 5% of the time, they will understand that. And the cool thing about learning AI in the context of a system you already use, there’s a better contextual understanding.

So you will know what to ask. The challenge with those AI models, those blank prompt windows is what do I ask it? When it’s done in the context of a business tool, the person knows where to start and they usually have a better sense for whether the result is accurate or not. So I really love introducing on that concept. I think again, we need to fight through this belief that AI without tuning, without model development is not really worthwhile. And that I could not disagree with that statement more. Start where you are today, get those early wins, build from there.

Daniel Newman:
Yeah, I think what’s going on is we have these sort of two concurrent shifts going on, Mark. And I’d be interested in your take on this, but there’s the one shift that’s sort of the way we’ve gotten used to search, whether that’s… You’re hearing a lot about RAG, which is Retrieval Augmented Generation where it’s able to access historic data and quickly generate tags. And then of course we’re hearing about tuning, which is a much more in-depth and expensive way to do it. Then of course we kind of just have search two or three, whatever generation we want to call it, right? And the idea now you’re seeing Bing incorporating it, you’re seeing Google incorporating it, you’re also seeing challenges getting it right. But we’ve gotten used to searching a certain way. So when you do a Google search, just say we search a certain way.

Prompting to your point to get the best results, it’s not necessarily done exactly like search. There is this first generation of just getting people to move from how we searched in the last era to how we need to search today. And the second thing I thought you said that was super prudent, Mark is accuracy. So right now we’ve got the kind of large language models everyone’s accessing is all the internet. It’s basically been scraped the whole internet, the high value stuff where something like a Smartsheet can really make an impact is all that proprietary data combined with language.

So you can take a LLM model and combine it with all the great insights that exist inside of a company’s instance of Smartsheet and start to get language and text to generate alongside data and insights to create outcomes. And that can be done fairly seamlessly in a tool like what you’re doing. So that’s really interesting. So it sounds to me… And I mean I just kind of… Let me run this into the future is you’re sort of indicating to me that you’re promoting curiosity. You’re promoting that people dive in, take risks, break things. How do you do that? Gain an edge, but also make sure that you’re right. There is a lot of criticism right now in the market about LLMs and things that have been done very quickly, but some of the outputs have been raw.

Mark Mader:
And I think the word right is one we should pause on for a second because no business leader who is charged with growth is anchoring in the notion of being right all the time. If you are right in every decision, you’ll be moving really slowly. So it’s this notion of super high velocity, super high inquiry, and then being able to check your sources to make sure that you have the probability of you being right is high enough. And I think this notion, again, the narrative is many answers will be wrong. The scary words like hallucinations. If you can present the responses and retrievals with sources where someone with critical thinking can review something and say, “I think this is right enough, high probability that it’s correct.” Then they should move forward very quickly. You mentioned the word curiosity also, I think there has not been a time in my working history of 30 years where there’s ever been as high a return on curiosity.

And when I think about the skills we’re looking for today, the two things. One, massive return on curiosity. And then two, it’s almost like the Renaissance age for the generalist. We spend years and years hiring these uber specialized people. Now there’s a resurgence of people who are articulate, can frame their thoughts well, pose questions, synthesize, incorporate, and determine back to your point, is it right enough? Is the probability high to move? And it’s almost like this balancing effect. The specialist had an amazing last 10 years, the generalist is making a recovery and now it’s more of an even waiting. It’s a really exciting time, but I would strongly encourage people to incorporate those aspects into their hiring profiles as well. Because I think that’ll be really important for this next phase.

Daniel Newman:
So as more of a generalist than a specialist, I appreciate your sentiment and of course also your subtle debate of what I said. I mean obviously there’s parts across AI where right is super important. And I don’t think we… I’m not trying to litigate it, but I guess I’m saying if you’re creating a sales document that’s going to be auto-generated out of a series of tables and customer data coupled with language, you want to make sure it’s the right customer name, the right address, the right sort of offer, the right appreciation for compliance, privacy, security. But at the other side of this, what you’re saying is with every great technological innovation and disruption, there’s been this fail fast sort of mentality, right? Where you got to sort of fail fast, fail forward, we’re going to learn, we’re going to see. This is why legislation will never keep up with innovation because they can’t put the policy behind this stuff because we’ve got to move faster.

We’ll fall behind in the world, we’ll fall behind in market leadership technology. So we’ve got all these things kind of going at the same time. So I got a CEO question for you, since you’re one of the day openers, and this is something I love to ask CEOs. But how fast, Mark does this move from here? Because I feel like it’s absolutely been… It’s been eye watering and blazing forward since November ’22. I mean, I can’t slow down. I blink and there’s a new model, there’s a new GPU, there’s a new software release, there’s new economic data and growth numbers. I mean, how fast is this moving? How fast are your customers?

Mark Mader:
There’s a huge distribution and I would say the median organization out there, it’ll be measured in years, not months, not quarters. And that’s fine. You’ll have an early flight group who moves very, very quickly. I don’t sign up to the notion of that early flight group will dominate the world and everyone will be out of business. I don’t think that’s the way it plays out. The median organization will come along though. So there’s no question in my mind that through apps, not through infrastructure, but through apps, it’s going to become a natural part of everyday fabric. They’ll be an advantage to people who adopted earlier. And I think the degree to which people feel comfortable to start to utilize their information in these models. I’ll give you an example. We internally had these hundreds of Slack channels that people were pretty familiar with. But I can tell you our employees weren’t rifling through 200 open Slack channels searching for answers.

So what we did was we melded the Slack channel content with HR documents, policy documents with important websites. We unified this in a searchable AI-ified way without writing a single line of code using Amazon Q. Those are the kinds of things that if those become more popular and people become aware of those, I can see that bringing the median forward faster. If it gets relegated to this concept of building and complexity, we don’t get over that. I think it’ll elongate that process. So again, part of what we’re doing is evangelizing for, AI is not scary. AI is totally accessible.

You can be successful in participating in AI without tuning a darn thing. That is the message we need to feel comfortable with. And I recently came back from a conference in Europe and each of the AI tracks was well attended. Every CEO in attendance was scared to ask a question. It was fascinating. And one experienced CEO looked at another CEO and said, “You’re the one who’s doing a lot of these discussions.” He goes, “How did you get so good at this?”

His answer, Daniel, YouTube. “I watch YouTube videos”. And it was actually a pretty brilliant answer because this person who came across as highly fluent, capable, informed, he was able to make recommendations. He had trained using publicly available content. And not that that is our single source, but it is a source. So again, we have to change this narrative and bring people along.

Daniel Newman:
Well, I mean I would say that that is sort of indicative of the educated of the future. You’ve seen the ones that like those that can learn, unlearn and relearn rather than those that kind of attend, learn a thing. You went to school, thought computer science, everyone told you learn to code. Now you see text to code and image to code. And I’m not saying… I think there’s still always going to be a place for highly skilled developers, no question. But to your point, the way that most will consume and the way that most will likely code and the way there is going to be a lot of ways that AI accelerates this when it becomes democratized simple and people see that path forward.

So we’ve got a couple minutes left. I got one question that I’ve sort of alluded to. Love to get your take and then I got a fun question to take you out of here with. The more serious question though is you heard me sort of comment on policy responsibility, security. Any thoughts from where you sit on the balance of how we should be thinking about security and privacy in the era of AI?

Mark Mader:
I think it is our duty as providers to in plain speak, explain what is happening to someone’s content. This can’t come under the guise of a simple click through agreement. It needs to be presented in plain speak at the highest level. We need to explain to customers whether their information is leaving the boundary, which is our service. If it does leave the boundary, how is that information used to improve someone else’s content or model? These are very basic concepts. The good news is the education on this is happening pretty quickly.

Much like when the early days of SaaS. There was a lot of question around security and can I trust it and availability and SLA is totally fluent today. I think the privacy dimension of AI will be a very quick learn by many companies’ legal teams. And I think that will permeate into the business teams who are making decisions on these things. So that is I think a very hot topic today. It’ll remain very relevant, but the fluency will get very high, very fast in my opinion. And the good news is the companies that don’t conform to it, Daniel, they’re not going to be around very long. People will not move forward with them. So this is a great forcing function for companies to operate with great thought.

Daniel Newman:
Yeah, it’s a good moment for this and I’ve said for a long time that I feel that people have sort of sacrificed too much. And in the very least when it comes to the context of the business and private data that we use as businesses and that we have and hold to create AI and to utilize AI, I hope that we can see a kind of a bit of a change of the guards to take it a bit more serious. Because you see how sometimes it slips and slides. It definitely did on the consumer side. So fun question to end. I mean, it’s no secret anyone that followed me on Twitter, I love McLaren. I am a huge Lando Norris fan. Smartsheet is a sponsor of that particular F1 car. But that car has long-tailed, a certain other car, a certain other driver. Just in your sort of eyes as you’ve been around it, you’ve sponsored it now, how much… We won’t talk about the specifics, but how much are you seeing something like an AI pivoting the future of a sport like F1?

Mark Mader:
The teams, what’s amazing about that sport is that a car at the beginning of the year, which is a winning car that doesn’t apply any changes to it, is the losing car by the end of the season. I mean, it’s remarkable. So this notion of implementing change quickly because you’re in a very tight calendar, I can’t think of a more relevant sport where AI is going to take root. They’re on very, very tight timelines and unfortunately once you get into the season, your ability to make change and apply change is under great pressure. So anything to accelerate that is a huge advantage.

So you’ll probably have teams who are considered to be the haves and the have-nots are the teams that are not investing in this area. That’ll become very, very clear here in the coming years. We have a huge format change coming, I think in 2026, where things sort of reset with a whole new engine configuration. So I’m sure there’ll be trying to figure out every single advantage they can get to improve it. So I think highly relevant. I can’t think of an industry where it’s not going to be relevant, Daniel. It’s going to be in most. This is one though that has a lot of capital behind it, so they’re going to be able to put full focus on it.

Daniel Newman:
It’s highly capitalized. The amount of data they get used is immense. I remember in ’23 the car was struggling. I went to the British GP and it was like one upgrade and the whole tide had turned and the entire season changed. But it was so data-driven. Everything is data every turn, every corner, every tire. The amount of space between the car and the ground, everything is done. And by its all adjusted in real time. And it is just so cool, Mark. And-

Mark Mader:
It is, Daniel. What was so gratifying to us this last season, the upgrade that was applied in Miami that led up to Lando’s first ever win, that entire upgrade was managed through Smartsheet. Every single project component, every single dependency on the build and upgrade of that car was run through a Smartsheet. The team principals had dashboards. They knew which teams were on Mark, who were a little bit behind, and they had to get that thing applied before that race. And in the video that some of you may have seen on CNBC where the principal says, “We’re making the changes and the results will come.”

Well, the results came like the next weekend. So we took great pride in that victory. A lot of people high-fiving here at Smartsheet.

Daniel Newman:
Yeah, absolutely. I may have had a tear in my eye and anyone that knows me knows I’m not a crier, but I’ve been suffering for a long time. And anyone also that knows that when you’re a fan, it is all about the suffering that creates the joy when success comes your way. Mark Mader, I want to thank you so much for helping us open up our first day here at The Six Five Summit 2024. You’ve been an excellent guest. It’s been a lot of fun. Congratulations to all the success at Smartsheet and I look forward to tracking and continuing to discuss the AI industry with you. Let’s have you back soon.

Mark Mader:
Thanks Daniel.

Daniel Newman:
All right, everyone, here you are. We’re kicking off our day. That was Mark Mader. CEO Smartsheet.