Awesome. Thanks everybody for joining us this morning. I'm Steve Enders, part of the Citi Software team, and I want to welcome everybody to day one of our global TMT Conference. With us today for this session, we got Intuit. Alex, I want to thank you so much for being here.
Yeah, thank you. It's good to be here.
Maybe just to start, I'm sure people know you from here, but just maybe give a little bit of your background and talk a little bit about what you do at Intuit.
Yeah, absolutely. Hello everyone. Alex Balazs, Intuit's Chief Technology Officer. I joined Intuit unbelievably in 1999, October of 1999. I'm coming up on 26 years. I joined a desktop software company, and I was part of the team that was tasked with putting small businesses online. I was one of the engineers on the first version of QuickBooks Online that we released in late 1999. It's been an unbelievable journey at Intuit since then, from a frontline engineer who wrote a lot of code, drove a lot of the early innovations that we had online, product strategy, platform strategy. I spent four years in the Small Business division, really understanding small businesses and how to actually solve their problems, especially as they moved online.
I spent 10 years in a platform organization working on some of the early capabilities around shared data, identity, developer tools, customer support tools, things like that. I spent six years in the TurboTax group as the Chief Architect for TurboTax and ran some of the engineering teams there. Really great experience on the cyclical nature of tax and the urgency of tax prep and really working with the customers there. 5+ years as Intuit's Chief Architect and the Head of Platform Engineering, where I drove technology strategy and a lot of the shared services that are used across all of Intuit's products. I've been the CTO for two years, and it's been a whirlwind two years and an amazing two years.
No, that's great to hear. I mean, with 26 years at the company, I'm sure you've seen a lot of the tech transitions, a lot of background there. Maybe you can just talk about how the Intuit product, how the platform's developed over your 26 years there. Maybe we can use that to jump start into some of the AI agentic conversation that's very top of mind right now.
Yeah, absolutely. One of the more remarkable things about the time that I spent at Intuit is how comfortable the company has been at disrupting itself and reinventing itself. That certainly is something that connects back to Scott Cook, the founder, and I have the great privilege of still having regular one-on-ones with Scott. We're so customer-focused. We're so focused on solving customer problems that we tend to not get over-concentrated on a specific technology choice, a specific platform choice, decisions that were made a year ago, five years ago, 10 years ago. Seeing this company that I joined where I was working on, like I said, some of the early online products and online services, there were small moments of, oh, what are we doing? Like, why would Small Businesses want to go online? They're really happy with the desktop products.
There was this little bit of inertia that prevented us at the moments from moving to the next step. As a company, boom, we just kind of blast through that inertia. That's what I've seen. We're constantly reinventing ourselves. That's one of the reasons why it's been so amazing to be there for such a long time, because as we've hit each of these kind of platform shifts from being desktop to online, online to SaaS, SaaS to platform, platform to data and AI, and now truly becoming this AI-driven expert platform, what I've seen is we reinvent ourselves. That reinvention certainly manifests itself in terms of the choices that we make about our technology stack and how we build products and how we market our products. It also connects to how we actually interact with our customers and monetize and commercialize what it is that we do.
It's going from, hey, this is a really good way to do accounting to this is the all-in-one platform where a business can interact with us and never have to leave. They can run their entire business. It's not just about the back office. It's about every part of their business. From a consumer perspective, it's not just about doing your taxes. It's about credit to taxes to wealth. You can basically use this as your all-in-one platform and you never have to leave. When I think back 25 years ago to some of the original things that we did to where we have evolved as a company to be this all-in-one platform, I think it's been a testament to, as I said, this reinvention and this idea that, you know, the classic book, The Innovator's Dilemma, and that we constantly kind of power through The Innovator's Dilemma.
That's great to hear. Maybe let's talk about AI. Very top of mind.
Yeah.
I mean, you've, I think, been working on building out your own AI capabilities within Intuit for almost a decade now. What have maybe been kind of the key learnings, and how does that maybe set up, you know, Intuit and the platform for the future as we think about the agentic age?
Yeah, absolutely. I remember, I can't remember the exact date, but it's like 2015, 2016, and I was in the TurboTax group. It was the first time that, at least that I'm aware, that we actually leveraged AI inside of our products. One of the biggest drop-off points in TurboTax was someone doesn't know, should I take the standard deduction or should I itemize? They go through the entire trouble of going through itemize. They do all the work, you know, an hour's worth of work, and then they realize a standard deduction would have been better, and they get upset, and they either leave or they go to a tax preparer, or whatever the case may be. Obviously, we wanted to keep those customers.
We said, boy, wouldn't it be great if we could predict if this customer could actually benefit more from being an itemized filer before they give us all their information? That was the first time we had trained a model to say, hey, let's see if we can predict, is this person going to be an itemized filer or standard deduction? To our surprise at the time, it was like 97%, 98% accuracy we could actually predict before they gave us all their tax info. That was kind of this eye-opening moment for me and for us as an organization to really understand, boy, the power of data. Certainly, the AI is amazing. We have a lot of AI folks who work at Intuit. We leverage a lot of amazing AI technology partners.
Really, when I think about the lessons that we've learned about leveraging AI, creating impactful AI, the first part of it is, boy, the data has to be there. With the large set of data that we have at Intuit, hundreds of thousands of attributes about every business, about 60,000 attributes for every consumer that's on our platform, we can actually leverage that data. Public models are great. Public LLMs are great. We use them, and we'll continue to use them. The power of the data, the data is so important.
The investments that we've made in data in terms of making sure our data is clean and organized, our data services to make sure our data is available, and we can get data both locally in our ecosystem and go get it anywhere where it is on the internet and make it available to our customers, and then how you actually leverage that in your AI, critical, critical part of our AI strategy. The second thing that I would call out is the AI is great, and the AI can solve many, many problems. The combination of AI plus what we call HI, human intelligence, is even more powerful.
For us, that manifests itself in our live services, TurboTax Live, Assisted, QuickBooks Live, Assisted, where it's the combination of what the AI can do and what an AI-enabled human can do together to actually go from, this is a really easy product where I could do it myself, to this does all the work completely for me, and I don't have to do anything. That AI plus the HI for us is something that we continue to lean into. In fact, as we released our agentic capability, our capabilities that leverage LLMs to automate entire workflows, one of the really critical features that we built into it was that at any point in this agentic workflow, it can pass you off to a human being. If the agent doesn't know what to do, the AI agent doesn't know what to do, it passes you to a human.
The human can pass you back to the AI agent. Along the way, we keep track of everything that's going on, so the human agent knows what the AI agent does, and the AI agent knows what the human agent did. AI plus HI to us is such a huge part of our strategy, and we're continuing to lean into both. The last thing I'll call out is we're very particular as to how we set up our teams to actually leverage AI. We want to make sure that they have unbelievable building blocks, that we can have very small teams who act like they're startups and they solve problems.
The challenge that any big enterprise has, any big company has, is how do you take the power of the company that you are and take the capabilities of you as a big company, but then act like you're a small company, act like you're a startup, be extremely agile, take risks. What we did is we specifically, in the age of AI, knew that we had to create a setup, and that's what we call GenOS, our Generative AI Operating System. We had to create a setup where our developers, and really it's more than developers, developer product design, AI folks working together, how do they act like they're a startup yet leverage the power of Intuit?
Creating these atomic services and capabilities allows them to move fast, to experiment, to launch experiments in a matter of days and weeks instead of months and years, and then collect all the metrics about it, see how it's performing, see how the customers are interacting with it. Are they engaging? Are they re-engaging? Are the predictions accurate? All of those things. We really made sure that this, what AWS refers to as the control plane, we made sure that we invested in this control plane so that our developers and our builders can iterate very quickly and solve for our customers. In the old days, 10 years ago of AI, it was about big investments and big projects and, hey, let's plan roadmaps and let's roll this out. Now in the age of AI, it's about how fast can you move? How fast can you learn?
It's really one of the biggest things I've been pushing as CTO is we have to have this builder culture where we're building and not talking. We're not in meetings, not talking about process. How do you build, get things in front of customers, see what wins, and then when it wins, scale it and commercialize it.
That's great to hear. I want to dig into the portfolio a little bit more. Before we do that, I do want to ask about, I guess, the own internal initiatives. I think you touched on it a little bit, but maybe how has all of the changes that you've made there, how is that beginning to translate in terms of the efficiencies within the R&D org? How is it translating into the pace of innovation and how much more quickly you can build products?
Yeah, absolutely. Our initiatives around productivity actually started about 10 years ago, where we actually, step one of being more productive is to measure how productive you actually are. We have a lot of automated metrics gathering that basically figures out how long a certain task takes for a member of the workforce. Certainly, we've invested more in areas, larger areas of the workforce, so our engineering development community, customer success. For each of these workflows that they do, we know how long it takes. Step one was to say from a productivity perspective, what can we put into place to actually grow productivity? In the tech organization, we talk about it as development productivity. In the past six years, we've improved our productivity 8x. We can actually, we have 8x more throughput from an engineering perspective now than we did five years ago.
Wow.
Probably about 5x of that was just standard development practices. The last couple of growth areas of productivity began in terms of our investments in AI. When you think about AI-based productivity, everyone tends to go to the engineering first because we create the tools, and we tend to aim it back at ourselves first. I was actually chatting with Dario, the CEO of Anthropic, two weeks ago, and he was talking about how development productivity, so much of it can be driven by these AI tools. Our early experience in terms of productivity for our engineers is when using these tools well, their productivity goes up by as much as 40% for coding activities. That is early. That is based on where the tools are today. We are constantly pushing on productivity, but it cannot just be about the engineers. For us, it is about the entire workforce.
We actually have a program where every single functional role in the company, we have a team who is dedicated to say, how will AI revolutionize how this person works? You have engineers, customer success, the most obvious ones, finance, HR, marketing. In each of these roles, we are now going through the same process and saying, how long does it take to actually execute certain tasks? How do we make these tasks more efficient? I am actually very excited that in two weeks at our Investor Day, up on stage, I am actually going to be demoing a couple of these capabilities, these tools that we have built that are completely changing, for example, how marketers work, completely changing how customer success works, how salespeople work, and having it be all AI-driven. This is a big, big area of focus for us.
The great thing about, I think, the way that we are approaching this productivity is we are pushing really hard on productivity. What that does is it gives us a lever. With that lever, we can choose, well, what do we want to do with that productivity? So far, with engineering, for example, we poured it back into engineering and we have said, let us do more, let us create more product, let us create more services for our customers. We can make those choices independently. We can choose to say, hey, is this a place where we are going to take the productivity and do more with the same number of people, or is this a place where we will actually recoup some of that productivity and invest it somewhere else? It is a big, big area of focus for us.
That's good to hear. A good plug for the Investor Day in two weeks on Thursday. Maybe it's a good time to start talking about how you're building AI into the product. I think you mentioned you're thinking about the personas of customers and users. How does that kind of inform how you build out the range of products for your end customers?
Yeah, I think in order to answer that, I have to first talk about what's the wrong way to solve the problem. Across the industry, there have been a lot of examples of the wrong way to solve the problem. The wrong way to solve the problem is to say, I'm going to take a specific feature and just turn it into AI. When you do that, it tends to not resonate well with the people that you actually did the work for. They'll say, you move my cheese, or I don't understand this, or I don't want, I'm not here for AI. I'm here to manage the finances for my business. I'm here to do my taxes. Why are you showing me AI? We completely changed our approach about a year ago where we said, what are end-to-end customer workflows that the customer themselves, they don't want to do?
If you could take this entire end-to-end customer workflow and actually automate the entire thing, then all the customer knows is that it was done for them. How do you approach customer workflows? When I sit down with various teams, in fact, I was just here at our New York office yesterday, and I was meeting with some of the engineering teams, they were asking me, like, how should they approach integrating AI into our products? What I told them was, don't think of it as you're creating smarter software. Think of it as if a human did that for that person, how would they solve the problem? That is a much better way to leverage AI. The way you leverage AI is you say, if you delegated this to a human, how would the human solve the problem?
A very specific example is if you're inside of QuickBooks and you're a lawn care business and you go on a job and you are quoting some big commercial lawn care project to do all the lawn maintenance for some big office park. You write a quote, you write an estimate. Under normal circumstances, what would happen is you would take that estimate, end of the day, you'd go back to the office, you'd probably give it to someone, that person would transpose it and put it into an email, that email would be sent out to the customer, the customer would decide, okay, do I actually want to take this estimate? They send it back and say, okay, I'll do it.
Another person takes that and puts it into the accounting system and creates an invoice and they send the invoice and then a payment comes in, or actually a payment probably doesn't come in, they forget to pay it. You have to send a reminder, right? These are all human activities. Everything I just described, our agents do automatically on our platform. You create an estimate, you can write it onto a piece of paper, you take a picture of it, that estimate automatically goes into our platform, the estimate is sent out to the customer, the customer digitally can accept it, it comes back to our platform, it automatically sends the invoice, they approve the invoice, if the time goes by and they didn't actually pay the invoice, we'll automatically send the reminder. Along the way, the business owner has the ability to control various aspects of it.
They can control the voice and tone of the messages, they can control the frequency, but the work is done for them. That is an amazing use of AI, right? Because now, instead of that business owner, not to mention the fact that one of the things that we've learned is that I believe it's 80% of the time, the first quote that goes for a business owner, the first quote wins. Now imagine you go back to the office and you send that quote, that estimate late in the night, and by then, two or three other people have sent estimates. You've just lost that job. When our customers see this, they truly understand the power of AI, but they don't see it as AI. They see it as you're doing the work for me so that I don't have to do it.
I'm more efficient, I win more work, I make more money, and therefore, with that value proposition, I'm willing to pay you more money.
That's interesting. I guess with that kind of perspective, we think about what the Intuit product portfolio maybe looks like a few years from now, with agentic experiences maturing, with the generative AI product set becoming more powerful. What does that look like? Maybe it helps to frame what is possible today versus a few years ago and how that begins to progress even further.
Yeah, so Intuit spent the first 40 years of its existence creating products to make it easy for you to do it yourself. We have spent the past two plus years and now going into the future creating products that will do it for you. That is the best way to kind of describe where our strategy is going. That manifests itself that instead of these standalone products, it is a platform. I talked about it in the beginning, about this concept of the all-in-one platform that you do not have to understand different tools. You do not have to know, under this circumstance, I have to use this tool, I have to find this data, I have to implement this workflow. The work is actually done for you. The evolution of our product strategy is this all-in-one platform, and the platform will do the work for you.
We have already seen early manifestations of that, certainly in the business platform with the agentic capabilities. We released these 10 agents in July, and we are continuing to invest in these end-to-end workflows. Over time, what you will see is quite directly, the amount of user experience, UX, starts to go down. The interaction modes tend to be something that are more simple and familiar. They could end up being voice. They could end up being video avatars or avatars. They could be, you know, the things that are more familiar to the customer. In fact, to answer your question about what is possible today and what will be possible in the future, a lot of this is actually possible today. We have early versions of some of our mobile products that almost completely eliminate the UI. Almost the entire interaction is just voice and text.
If you are going out on a job and you work for a mid-market business and you are doing construction, when you go out to the job, the mobile app will summarize everything that you need to know. It will tell you who the customer is. It will tell you what the action items are. It will tell you if they owe you money. It will tell you what the next part of the job is. If you want to interact with it, you can type, you can talk to it. Is all of that available today where we can completely eliminate that?
No, but what we are working on is the capabilities, like if that is our North Star, that it is completely done for you, then what we are working on is how much of this interaction, how much of these more traditional clickable UIs can we actually eliminate in lieu of actually doing the work for you and making the experience and the interaction be more natural and more human-like?
Okay. As you think about that, how do you, I guess, then kind of go to the next step? How do you monetize that? How does it actually maybe show up in terms of how you generate revenue from it? Is it about driving cross-sell and driving some of the broader portfolio? Is it about layering these capabilities in and using that as a different lever for driving price over time?
Yeah, I mean, it's all of the above. What I would say is that we've already seen that in terms of our product lineup, and as you kind of go up the product lineup, the customer value that you describe that is delivered by AI is reasons why people will trade up. What they won't trade up for is, hey, this is the non-AI version and this is the AI version. Many of our customers are like, I don't know what that means. If they say, we'll automatically do invoicing for you, it will automatically generate all your marketing campaigns for you, it will automatically run payroll and collect all your payroll information for all your employees so you don't have to hire a payroll clerk. These are the things they understand.
When they look at a product lineup, the monetization, commercialization comes from getting customers to actually move up the lineup. The other thing that becomes extremely valuable, which you touched on, is that because it is this all-in-one platform and because it is these end-to-end workflows, the traditional way to interact with a customer to sign on for these additional services very much feels like an ad. It's not supposed to be an ad. I mean, to us, it kind of ends up being an ad because it pops up and then they click yes and they sign up and we make more money. That's kind of the definition of an ad. In reality, the reason we're showing it is because we actually think the customer can benefit from it.
When it's an end-to-end workflow, and as part of this workflow, they have to do something, but then if they sign up for this thing here, then they don't have to do it anymore, to them it doesn't feel like an ad anymore. That's actually some of the early feedback that we're getting from our customers for in-product discovery. How do customers actually discover these things? What that allows us to do is to basically, you know, in some sense, the traditional land and expand, right? You kind of land, the customer joins your platform and they show up for a specific reason. There's a specific job that they want solved. Over time, you expand. The way that you expand is that you show the customer, hey, here's a report that actually shows your projections of what your revenue will be over the next three months, six months, 12 months.
The first one they get for free. Then they say, boy, I'd like to see that projection again. Oh, you want to see that projection? You basically have to sign up for this service. They’re more than willing to do it because these are the kinds of cash flow projections and reporting and automations that make their lives easier and allow them to actually concentrate on running their business instead of the mechanics of running their back office.
Okay. I think you talked a little bit about maybe changing the interface for how customers interact with Intuit or interact with the platform. How does that maybe change the go-to-market approach? How does that change some of the upsell and cross-sell mechanisms that, to your point, it traditionally looked maybe like an ad, but you're trying to change that. How does that look different today or moving forward?
Yeah, so it's interesting. A couple of weeks ago, I was sitting down with the marketing team and reviewing how they went through the process of inspecting and understanding all the AI capabilities that we were building. They actually did an amazing job. They ran a bunch of experiments. Obviously, we love running experiments. There's no substitute for hearing from the customer, right? What do you have to do? You have to run a lot of experiments. We ran a lot of experiments in terms of our commercialization strategy. We put things out there like, hey, if you have this version of QuickBooks or an AI version of QuickBooks, which one would you buy? Doesn't resonate. Hey, if we deliver this and this specific feature is automated versus not, will you pay for it? Didn't really resonate. What really resonated is the benefit.
It just always goes back to customer benefit. For us, no work, complete confidence, more money. When you connect it back to the benefit, then the customer actually understands it. The way it ties to our marketing strategies is that if you look at turbotax.com, quickbooks.com, some of our marketing front doors, six months ago, nine months ago, they tended to be product, product, product, product. If you look at them now, it's benefit, benefit, benefit, benefit. It's more about the benefit and having them choose a product, having some carousel where they have to make decisions. Those are things that are actually becoming a thing of the past because the customer doesn't actually even really understand them, to be quite honest. What they understand is what benefit am I going to get out of it? Our marketing campaigns and therefore our commercialization campaigns have been tied to benefit.
When the customer perceives benefit, so we are giving them benefit, we know they will give us benefit. That's when they're willing to pay for these things.
Okay. That makes sense. I'm going to shift gears a little bit and just, you know, I think we hear a lot of concerns from investors on, you know, AI impact to software, what that means for SaaS. I think we hear about the death of SaaS sometimes. It would be great to kind of get your perspective on how you think about, you know, the market. How does Intuit maybe, you know, stay ahead of the competitors in this space? How do you think about, you know, building out a product set when there is this kind of era of disruption right now?
Yeah, there is no middle ground in this space. You're either going to be disrupted or you're going to be the disruptor. There's as much as any time in my career that I've seen and reading the tea leaves as to where things are and where they're going to be. We talked about things like cloud. When cloud came out, it was like, oh, if you don't go to the cloud, you're going to be disrupted. Yet there are still companies today that probably aren't running on a public cloud. In this area of AI and SaaS and business logic, you're either the disruptor or you're going to be disrupted. That is absolutely the way that we are approaching this. What does that mean? It means that the days of hand-coding business logic into SaaS applications are numbered.
That is the reason why we are investing so much in allowing ourselves to build this platform where AI and agentic capabilities become the business logic. The business logic learns from the data. It learns from the training that we give it. It learns from the experience that we have. That is why we firmly believe, I firmly believe, in this area of disruptor and disrupted, we will be the disruptor. Because we have 100 million customers, because we have all this data, because we have our platform, we have the ability to be the disruptor, and we will be the disruptor.
The power, the investments that we've made in our platform, where we have all these backend capabilities, are extremely valuable because the business logic that sits on top of it, this AI, this AI is only as good as the tools and the data that it has access to. What has Intuit invested in the past 10 years? Tools and data. That's exactly what we've invested in. Now we've also invested in the AI. Absolutely, there is nothing sacred about any part of any of our products. Every single experience we have, every single app we have, every single product we have, we are self-disrupting it. Going back to the first thing I talked about, because you're absolutely right. You're either going to be disrupted or you're going to be the disruptor, and we are aggressively leaning in so that we are the disruptor.
Okay. That's great to hear. I know we're kind of getting near time here, but just, as we think about the last couple of minutes, what does maybe the future of Intuit look like as you build out these broader agentic capabilities? Maybe how is the portfolio different a few years from now, five years from now, if we're talking here at Citi 2030?
Yeah.
Yeah. What does Intuit look like? How is it different?
Yeah, I think what you're going to see is the scale of the business will change. One of the biggest ways the scale of the business will change is the most obvious one, which we've already had the strategy around going up market in businesses, going to the mid-market, much larger businesses. This is definitely a place where AI helps us. The future of the business will be as Intuit goes up market, and the more you go up market, the more verticalized businesses become. It used to be, boy, the barrier to entry to verticalize and across all these different spaces, the barrier of entry is very high. With AI, the barrier of entry is significantly lower. What you'll see in the future of Intuit is a company that has moved up market, and we're solving for significantly larger businesses. We're solving extremely well in different verticals.
You'll also see us go horizontal. We'll solve more of those horizontal jobs. As opposed to being kind of pigeonholed into the historical problems that we solved, AI and our ability to integrate with external data sources and external tools that are running on our platform will allow us to basically solve every job. It doesn't matter how big of a business you are, how complicated of a consumer you are, or what job you're looking to solve, we'll have the depth and the breadth as a platform to actually solve your needs. In 2030, what you'll see is an Intuit that is less about the individual products and more about, wow, this is the platform that is truly the one-stop shop where as a business or as a consumer, for my finances, I don't have to go anywhere else.
Okay. I think it's a great place to leave it. Alex, I want to thank you so much for being here and the broader Intuit team. Thanks again so much.
It's a pleasure to be here. Thank you.
Thanks, Alex.