AI, Security, and the Courage to Reinvent with Smartsheet CTO Cynthia Tee

Speaker A: All right, Cynthia, thank you for being on the Shift AI podcast. I am really, really excited to talk with you today.

Speaker B: Yeah, thank you, Boaz. Thank you for having me.

Speaker A: Well, before we start, you know, I, you've had a very interesting work history and uh, you know, you were just finishing up as the CTO of smartsheet and you know, before that you've been in technology a really long time. Can you talk to people a little bit about your background and kind of what led you to end up being the CTO of smartsheet and then kind of what's happening and what you're doing now?

Speaker B: Yeah, yeah. Well, I have always been in the technology space for a while. I graduated with a computer science degree way back when from mit, which by the way taught me very little about how to go through the disruption, uh, of the Internet than cloud computing, than AI itself. But uh, I think the most important thing I learned was you just have to pivot and learn fast and grow yourself really, which I think has really helped especially, especially with AI. So, um, I've worked at Microsoft for a number of years. I thought that was a really great place to land early in career. There was just a lot of support to grow my skillset, both as an individual contributor in addition to a leader. And I worked on a range of products there, the operating system, Windows being one.

Speaker A: Yeah.

Speaker B: But also a lot of consumer facing products. And so I worked a lot with consumers, enterprise workers and developers is my audience, my customer base there. I then went to run a nonprofit called ADA Developers Academy which took women and uh, gender and gender expansive community into the software industry from no computer science degree and so non traditional backgrounds. And really that was a great experience for me to learn how to make the industry more accessible to people who have just been systemically told that they couldn't do it and also didn't have access to that.

Speaker A: Yeah.

Speaker B: And so that was, that was a great learning experience for me and the ability to really increase the diversity, especially in Seattle, but nationwide. The companies that participated went to work for Nordstrom, immersed myself in the world of retail and back end commerce, which was super educational. Did a short stint with Playfab, which is a game developer platform that got absorbed back into Microsoft. And because I did not want to go back to Microsoft, I basically chose to do something else. I found Prairit Garg, who I think you've had on this podcast.

Speaker A: Yeah, yeah, PG's been on the show.

Speaker B: Went into SaaS with all my learnings, all my learnings from Microsoft in terms of how to build enterprise grade, global, secure, manageable software, but in a very different world now. Right. Um, so it was, it was great going much of that time with him to really turning smartsheet into, you know, bringing it to the next level of scale.

Speaker A: Yeah.

Speaker B: For high capacity workloads that our customers were expecting and also turning the crank on its user experience to make it easier to learn, easier to use. And of course, AI came in there too, and, uh, I had the fortune to take advantage of that also.

Speaker A: How long were you at smartsheet? Cause you, you were on a, on a journey with them, weren't you?

Speaker B: Yeah, I was there for four and a half years. A little more than four and a half years.

Speaker A: Yeah.

Speaker B: Yeah.

Speaker A: And then when you became cto, you were part of that transition back to private, right? From, from public to private.

Speaker B: Yes, I became CTO right as that, as the private transition was becoming official. Um, and had really learned a lot.

Speaker A: Yeah.

Speaker B: About that transition and really managing the, um, organization through that change.

Speaker A: Yeah, absolutely.

Speaker B: Yeah.

Speaker A: Well, I want to dig into the, some of those learnings and just the fact that you've been at such a large enterprise and enterprises, I mean, when you're at Microsoft too, and the implications all of that now, you know, within the AI world. But before we do that, I do want to ask you, because I like to ask this question on the show about your first job. What, what was your first job way back where you actually got paid? And what state did you grow up in? Like what, what part of the world did you grow up in?

Speaker B: Yeah, I grew up in the Philippines, actually. I grew up in Manila. My first job was when I was 12 years old and it's a summer job. I went to work for the library. So, you know, didn't have computers back then. You had to read paper and books. So people borrowed and returned a lot of books. So every day I was looking up where they needed to go and sorting them. We didn't really have a computer used in the library back then either, so I was looking for them through card files. Amazing things like that. The library system and library science is very rich in terms of these cataloging things. They weren't digital yet at that time, so. Yeah, that's a great, that's what I did.

Speaker A: That's a great fit. Yeah.

Speaker B: Yeah.

Speaker A: Well, let's dive in a little bit to smartsheet in that time. And also the ways that you guys were thinking about incorporating or the ways that you are in thinking about incorporating AI into the product. One of the things that I think is really fascinating, especially as AI is experimental for a lot of people, whether it's the agentic stuff that's happening now or even assistants getting integrated into companies, is that people aren't operating at the scale that you guys were both at the, uh, enterprise scale, and then the security issues related to that. So can we. Let's. Let's talk about that a little bit. Like, what came up for you as you were trying to figure this out as cto?

Speaker B: You know, the one thing about AI that was super invigorating in its disruption is that there were quite a few capabilities that we wanted to build a certain way where when AI came along, all of a sudden we realized that we don't have to build it that way anymore and you can provide it in a very different manner. Right. Because of all the LLM capabilities. So I will say Smartsheet's features on formula generation and visualization, which were part of the first generation of features, they've since evolved, especially after also since I left, are really great examples of. If we didn't have generative AI and LLMs, it would be, uh, just very challenging to make an easier interface. But because of that, it just became so much more delightful to provide that to our customers.

Speaker A: Right.

Speaker B: And so we, we really took advantage of that and our customers, you know, received it very well, especially those who really were. Were using our dashboards as well as picking up on our formulas. I think, uh, you know, AI is. Is exciting for a lot of our customers, but also pursued with caution, at least at the time when it was released. Yeah, we had to be very diligent about communicating to our customers exactly what information we sent in and what we didn't. What the LLMs were doing. It was. It was a big black box to them. And so we really had to be very specific about how the information, uh, what, uh, we save what we don't, things like that.

Speaker A: And this is sensitive information that's going into these spreadsheets.

Speaker B: So, yeah, yeah, we didn't send all the information from the sheets to the LLMs. You know, we actually by design made use of very little precisely because we are also very conscious that we didn't want to move too fast in using the. We didn't need to actually. We. We basically decided, you know, what is enough to provide the value, um, and just use that much. And we disclosed that to our customers so they would have control over it as well.

Speaker A: I like that use case, though. I was down in Dreamforce recently and listening to the keynote last year, and A lot of it was around how to make the integration of Salesforce tools much m simpler by using generative AI to answer those questions, like, hey, spin up this integration for me. Where in the past you'd have to become an expert, now, it was much easier. I think about the thing in the same way as I do for formulas. Think about all the formulas that we had to kind of scratch our heads and figure out how, uh, to do. And that complexity was erased when I can use English to ask for what I want. And that's how you were thinking about it right at the beginning?

Speaker B: Yeah, yeah, that's where we started. And of course, that was a while back. I mean, I know the, the, the team has really evolved these capabilities. Um, I. And I know they recently released integration with Claude through MCP, which you can find the LinkedIn posts. I, I know the customers are really excited about it, and I am too. I mean, I am no longer affiliated with the company, but I use smartsheet with cloud all the time.

Speaker A: Yeah, that's cool.

Speaker B: Yeah.

Speaker A: Explain to the audience who don't know about MCP how that works in the context of Smartsheet.

Speaker B: Uh, the best way to explain MCP is it's really, it's really the way you talk to applications. It's the way Claude or any LLM would basically communicate with applications to get the action done. And so something like Plaid is very good at interpreting what you want. Right. That's its, that's its specialty, is you talk to it, you tell it what you want, and then you get to a point where you actually want insights or you want something done, you want something completed. Right. And we have a whole slew of applications that can do stuff for you. Google can generate documents, spreadsheets, slides. So can Office. No name any piece of SaaS suite or SaaS software. They do a lot. And so MCP is basically the way an LLM will talk to these applications, to basically translate the intent to action. Right. And so, um, I could use mcp, for example, to create a project plan in smartsheet that matches what I need. And the LLM will be very good about saying, okay, you want to do, let's say, a home remodeling project. I think you're going to need this information. And through mcp, it can basically generate a smartsheet with the right columns, the, you know, provide status, change things for you. I'm trying to think of a good analogy like, uh, maybe Google Translate is a good analogy. You just basically tell it and it basically translates it to what the application needs.

Speaker A: Yeah.

Speaker B: And of course, under the hood, the application does have to provide, you know, this gets more technical, but they have to provide interfaces to do everything that needs to be done. And so what I think is really great about this, it produces a flywheel for those applications to provide those capabilities. Because at some point they're like, oops, you know, I can't quite do that, but I'd love to. Right?

Speaker A: Yeah.

Speaker B: So when you get that customer data, you know what to put on your roadmap to improve. And by the way, I send all my requests to all the peeps I know what smartsheets 1 to 5. Here's where I got, here's where I couldn't move further. But I'm very excited because at that point, once you, once you have that integration and once you have the customer feedback in, then it's, it's pretty exciting for a company to then keep evolving. Absolutely, keep evolving it.

Speaker A: What do you think? There's a lot of executives that listen to the show, Uh, a lot of, uh, technology executives as well. What are some of the lessons learned from your time, uh, at smartsheet and what should they be thinking about when they're trying to implement, let's say, folks that are just starting to implement at scale, both from a security standpoint and just from an integration standpoint that you learned, what were your takeaways?

Speaker B: Are you talking about executives who run other SaaS companies or executives who are customers of these SaaS companies?

Speaker A: I mean, I think in this case, because you guys were integrating this so closely into your product, probably executives of SaaS companies that are trying to figure out how to make this work for their products.

Speaker B: Yeah, yeah. You know, I think actually coming up with the capabilities and the features you want is the simpler part of the problem, the harder part of a problem. I would say there are a, um, few things. One is, you know, trust and transparency and governance. Control are super important. You know, at some point you're going to have to submit a ton of documentation and really explain the pathway and what your APIs are doing. If you think about GDPR, for example, the level of transparency that it demanded, you should assume that that's going to happen. You should assume that customers will want the trail of what they queried, they will want to know their information, will want to control what to turn on and off. Uh, so that level of governance is strong. And I think just like you think about how you would interact with an LLM, sometimes you're like, that doesn't sound Quite right. Can you tell me exactly how you got to that answer? What's your source? Because you do catch it, make mistakes. As a SaaS product, you have to be able to explain exactly what happened there, because it's not always going to be right. Right. And so I think a lot of customers will feel like, well, but it's not always right, so why should I use it? So you really have to. Perfection is not necessarily the goal. I think coming up with a productive outcome for customers is a goal. And I think as long as you can use the LLM to understand, okay, what's the confidence? How can I also be very transparent with my customer about what I'm. If they want, give the opportunity for the customer to tell you whether you were right or wrong so that you can adjust. And I think when, um, customers see that you are building towards what's right for them, that is a very important set of capabilities to really build the right foundation for. The second thing is that you can come up with these capabilities. Your engineering team will now use tools to generate a lot of stuff. And then there's the rest of the process that you have in your company for releasing software. Right. And so if you do not also think about how to evolve all of that. And I'm talking about everything from reviewing the feature. You know, engineers can accelerate everything within their realm all the way up to the deployment. But for example, what is your process for enabling the sales team or the customer support team to learn and support this? What is the process for setting it up for pricing?

Speaker A: Yeah.

Speaker B: Like, if you don't think about how you want to also accelerate that, then you kind of get stuck.

Speaker A: Yeah.

Speaker B: Somewhere.

Speaker A: Especially in a, In a world where it's so easy to create features.

Speaker B: Yes.

Speaker A: You know.

Speaker B: Yeah. So I do think it's. It's read a very great post on LinkedIn that was recently posted about this, which is like the rest of your process and the rest of your organization ecosystem also has to think about it. So I think about it as, how do you, how do you accelerate the whole life cycle of producing a feature? Right. And that's everything from the engineering to even the roles. Like, I was always confused by why some of my engineers kept waiting. I'm like, you can pick up lovable now. Uh, as some of, some of, you know, you know the product so well, you can actually prototype things yourself. Right. One of our designers while I was there, worked her way to the point. Point where she could almost check something in.

Speaker A: Wow.

Speaker B: And so I think that you have to really the second thing I would say is you really have to think about, if you run a big SaaS company, your roles and responsibilities and how you want to change them. Obviously, if you change them too fast, it's chaotic, may not be that manageable, but I think you have to be open to disrupting some of what these roles do and don't do.

Speaker A: Yeah, right.

Speaker B: And challenge people. Um, and in fact, I think that whole upskilling of what you would define as your traditional skill set is one of the most exciting things about AI.

Speaker A: Mhm.

Speaker B: Right. Um, and this is why I think young people, young people, they have no conditioning yet. They don't know. They don't know what a product manager is. They don't know they might grow up in the world where they don't need to know. Which is great.

Speaker A: Yeah, it is great. Well, let's try. I mean, that's a great segue. I know. You know, we both have kids and they're both, you know, at college age, studying and doing things. I mean, uh, that's really, the learning aspect of this has really changed. I mean, one of the questions that I inevitably get at every presentation is what should I tell my kids to study? How should I help them understand what this world looks like? How do you think about that, both for your kids, but also just for the engineers that were in your company and all the young professionals that are coming up in this new world?

Speaker B: Yeah. You know, I think there's a tension. I'll talk about a tension that I'm not sure what the, what the, uh, best answer is to. And I'll talk about the opportunity. The tension is, is that young people go to college and usually some of them will be very interested in the craft and learning the craft. You know, I'm. My, I, uh, know my son is very interested in learning computer science as a craft, coding as a craft. So, you know, he'll try using Claude to do his homework and then realize, oh man, this made me lazy. M. I don't like, I'm not learning anything. I'm not learning anything. So I said, well, while you're paying the institution to teach you something, don't ask AI to teach. Maybe you ask it to help you. But I know when, once somebody starts paying you, you really need to then know how to use the tool. So, you know, I don't know what the right answer is there, but I ask him to balance it out like, yes, learn the craft. But in learning the craft, how do you use AI to make you more efficient? And that is A harder problem when you go to industry. Right. Because increasingly, I think, you know, the norm is to say, like, hey, we don't need the junior engineers. But you do realize that the more senior engineers at some point were, you know, you need to. You need to train the next generation. So I don't know what the right answer is there, but I do. I do remember some very, very, very, um, young engineers at smartsheet that had the hustle, I call it the hustle. Right.

Speaker A: Yeah.

Speaker B: They had the determination, their curiosity, they learned and were coached by more senior people. And I think they would say that they are now much more proficient with using AI than they were before. And they'll still make mistakes, but they'll know very quickly how to recover. So that sort of hustle, I think to be always curious and to be willing to try stuff, go out of your comfortable norm is, I think, a very necessary skill.

Speaker A: Yeah.

Speaker B: On the other hand, I think the opportunity is enormous for people who didn't know how to do something and then all of a sudden got something done that they never thought they could do.

Speaker A: That's right. Get on block. Get something done. It is. It's incredible.

Speaker B: Yeah. I. I read a LinkedIn article from one of my former colleagues whose daughter, just literally, you know, she just made a wonderful resume website that adapted itself depending on who came. You know, she did this herself. She was not a technical person, and she did this herself because she was curious. She dove in, and the tools have helped her. Right. So there is so much opportunity. I was talking to another friend of mine who is not technical at all, and she built an inventory system for vintage clothing to sell vintage clothing because that was her passion. And she said, look, I just pulled up Claude and. And it did stuff for me. And now she's thinking of selling it. Right. Because there are other. So. And now she's asking me, what do I do? I'm just one person. So, um, it's so interesting to me how there is so much opportunity for people to reach out and do something they've never done before. And so for me, that, like, is. It opens a whole new world of businesses that people can get into that they've never gotten into, and. And skills people can pick up that they've never been able to do on their own or would've taken a lot. Yeah.

Speaker A: And it's funny, too, because so much of the focus now is people are afraid that they're gonna lose their jobs. They're afraid of becoming irrelevant. But there's also the flip side of it, there's amazing opportunities out there. And you know, one of the things about software engineers that I think is interesting is that the hustle that you talked about and like hitting your head against a wall on a problem and like pushing through the bugs or you know, going into documentation like they used to have to and like figure things out, now I see entrepreneurs doing that too. You know, they have this tool Claude, but then inevitably something happens, doesn't work. They need to go into Claude and ask questions and figure it, you know, figure it out and they push through. And then on the other side, the other. With my kids, I'm seeing a kind of hustle that allows them to go to market because one of the things that I see a lot of is it's easy to build things, right. Everyone's a builder now, but it's hard to really message and sell and get customers and you know, have conversations. And so I'm seeing that's another side of the hustle that young people are starting to tell me that they're learning and it's, it is uncomfortable. Uh, but it's something they're pushing on. I wonder what you think about that.

Speaker B: I think the tools can also help you there. Right. It won't always suggest something m. That you're comfortable doing, but if you have worked with it enough, you know, to push it right, you know, to say, well, I'm not comfortable with that. Can you say that differently or well, what about if I do this and what about if I do this instead? Like criticize this for me objectively, give me the pros and cons. So I think, I think anybody who is inherently self critical and very curious and willing to hear the other side of the story, you definitely will get the most out of an LLM and not just take its suggestions. That's actually not the way you should work with it because it can get, it can get pretty loony every now and then. So I think, um, it does help people do more than just build. I think it's a great tool for marketing. I think it's a great tool to gauge and predict reactions to something. I work with social venture partners right now to advise a couple of nonprofits and I encourage them to use it all the time. I say, hey, you're supposed to write this pitch, right? Let's, let's start to use the uh, LLM to figure out like how you can adapt this to social media, how you can adapt this to a different audience. I think it's great for things like that.

Speaker A: Yeah. Now I Know, um, you've been and are very passionate about social causes. And so now that you have the time and the opportunity to work with some of those. You know, one of the stories you told me about ADA and about giving people the opportunity to get into software that they never ever in a million years thought they could and just the impact that you had on their lives. Can you talk about some of the other social projects that you are passionate about and that you may get more into now that you have some more time?

Speaker B: Yeah, one is one that I had gotten into with an, uh, organization called svp, Social Venture Partners. They started. Started off actually, in Seattle, and they have a program where they pick a cohort of nonprofits who have an annual budget no more than a couple of hundred thousand dollars a year. So they're smaller, and they're at that pivot point where they need to learn a set of skills and also need to cultivate a next level of donorship to sort of take them to the next level. Right. And so the program has a set of advisors, I being one of them that works with. I partner with somebody to work with two nonprofits and really help them sort of go through the whole cycle of building a case for asking for donations, describing their program, describing the problem that they're trying to solve. We walk them through a financial workshop, and then we also do a storytelling workshop, and then there's an event with SVP's community where they can fundraise and, you know, make a pitch.

Speaker A: Yeah, amazing.

Speaker B: Um, and a lot of the funds donated that arrive from that event basically are distributed to the organization. So I love that because it's a opportunity for me to learn more, actually, about nonprofits and who they serve. They are really a wide variety of. Of really important causes. Um, and I also like that I can, you know, just give them my time and advice where they need it.

Speaker A: Yeah. So if people are interested in learning more about that and reaching out, is there a website for. For that, or can they reach out to you directly?

Speaker B: They can reach out to me directly also. Uh, I wish I memorized svp. Seattle Venture Partners.

Speaker A: Yeah, Seattle Venture Partners. People will find it.

Speaker B: Yeah, yeah. And you could just get more information on their programs. You could be a partner. They describe how to do that. We are, I know, running a next cohort.

Speaker A: Okay.

Speaker B: And so, you know, I. I think that if folks are interested in giving their time to be advisors, let me know. I know that we're trying to recruit more folks.

Speaker A: Nice. So, you know, I always ask this question at the End of the show. And I know you're, you're ready for it, but when you think about all that you've been through and you think about the future of work and the future of AI, uh, how would you describe that in two words and then you can elaborate on it.

Speaker B: Uh, the first word I always would use is disruptive.

Speaker A: Mhm.

Speaker B: The second I would use is bold.

Speaker A: Okay. Disruptive and bold. All right.

Speaker B: Yeah. So disruptive. Because I think there are, there is negative and positive disruption that happens from something like this. Right. You know, it is very real. Like, I don't want to belittle the fear and anxiety about your job going away because LLMs can do it. Companies who question their identity because now, uh, it's been commoditized.

Speaker A: Totally.

Speaker B: So I think there's very real negative anxiety where I think it is more positive. And I think this is for people who have the grit, the resilience to really muscle through. It is the pace of reinvention and upskilling that it really imposes on people and organizations. And you do have to pace it. You do have to be very thoughtful about it. But I think we've gone through disruptions before, but this is just, I don't know, 10 times faster than ever. Right. On individuals and on teams. And so I think both on the positive and negative and when you look at a product that you're building, it's, it's sort of the same. Right. You look at it and you're like, oh, maybe some of my things become irrelevant, but, oh, there's still customer value I've provided. How do I provide it in this new world? How do I monetize it? How do I make it viral in this new world? So think, think. That is a disruption. And the bold is, you know, I mean, I, it's very personal to me because I tend to be a little more conservative. It just my nature. Right. Uh, you do need to be willing to just take more risks because you just don't know. Right. What was hot six months ago is going to be old. Right. And so if you spend six months building something that is no longer the cool thing, you got to be willing to set it aside and evolve a different way and evolve very quickly. Yeah, right.

Speaker A: Yeah.

Speaker B: So that's probably, you know, gives people that are not risk takers a heart attack practically on the spot. You have to be willing to do this. You have to be willing to be bold and experiment and pivot really, really quickly.

Speaker A: Yeah, no, I love that, I love that. It's true. And it's a great lesson for leaders, too, to sometimes just have to lean in and take some chances and be bold. Well, how, um, what's the best way for people to reach out to you? Are you pretty active on LinkedIn? If, uh, folks are interested, how should they get in touch with you?

Speaker B: Yeah, you, uh, people can reach me on LinkedIn. That's probably the easiest way. I am always on it. Uh, yeah. And that's a great way to reach me. Did they talk enough about security? I feel like we didn't. We may not have.

Speaker A: Yeah, let's, let's touch more on security. Tell me, tell me. I am very interested in the way we talked a little bit about it, but the surface area has just really expanded in this new world, especially in the last six months with agents. And can you imagine, uh, unleashing agents in your enterprise in the ways that they're getting right now? I mean. Yeah. So talk about security from your perspective.

Speaker B: Uh, there are a couple of things that come to mind. One, it does not m. It does not relieve you of really nailing the basics for your software. Right? Your ability to support secure logins, your ability to secure your data, make sure it's accessed with the right credentials, your ability to, you know, depending on what business you're in, you encrypt data in motion or static, like whatever that is, whatever your customers have been asking you, classification, uh, of data, you gotta nail that because AI accelerates the transfer of information, right? And so if you don't have those basics nailed, it is going to be so painful, so painful to catch up with it. So really, I would say accelerating the completion of those basics for your software and a lot of the demands makes any CIO and it much more confident in the deployment of, uh, your software.

Speaker A: What more the AI and that's data and security.

Speaker B: Yeah, Yeah. I will say that the kind of, you know, we always get security training even within our own environment. And this speaks to protecting your environment from security breaches. Uh, it gets pretty scary now, what you can and can't tell apart. And so I would say to any company that is, you know, has efforts to protect the boundary of their company that you have to, you have to really get on top of how you catch phishing, how you catch any attempts to, to break into your door, right? You know, phishing, uh, I used to get trained all the time with security, and then they would send me these phishing tests and I would fail them and then I would get scolded. Right. But this is not the time to keep scolding me because I will fail at some point. The things are getting so good. Getting good at impersonating things where you can no longer really tell. Right. So I would say to any. Anybody who has to, who's in charge of protecting their boundary, your it, CIO or C. You gotta figure out how to use the tools at the edge.

Speaker A: Yeah.

Speaker B: To really, really scan and protect what's coming in. And, um, you know, I know I introduced you to Ravi, so, you know, these are the kinds of things he and I talk about all the time, is how do you protect the boundaries of the company and then actually how do you get the product capabilities. Good about protecting customer data.

Speaker A: Yeah, absolutely. And identity is such a big piece of that, too, these days, so. No, I'm glad you brought that up. Well, Cynthia, thank you so much for taking the time and being on the show today. I think it's just such an interesting perspective given the size of the organizations that you've been at and what's happening today and how these are trying to make their way into enterprises that, in a lot of cases, aren't ready. And so I think it's a really good perspective. I think the audience is going to get a lot out of the conversation. So thank you again for being on the show.

Speaker B: Yeah. Thank you, Boaz, for having me. It was fun.