All right, Welcome to the Third Annual TD Cowen IT Services and Digital Engineering Summit. I'm Brian Bergin, Equity Research Analyst at TD Cowen, joined by my colleague, Zach Ajzenman. Next, we'll have a discussion with CI&T, a LATAM-based digital product engineering vendor with over 6,500 global professionals. With us, we have Cesar Gon, Co-founder and CEO, and Eduardo Galvão, Head of Investor Relations. Gentlemen, thanks for being with us today.
Thank you, Brian.
Thank you, Brian.
You as well. For the audience, if you want to participate in the conversation, you can do so. You just insert questions into the MeetMax platform or you email me, and I'll work them in. With that, we're going to go right into it here, and I think, Cesar, it probably makes sense to start with an intro for the audience, for any of those that are less familiar with CI&T and the story here. Can you kind of provide a brief overview of the company's formation, the evolution, and really those key points of differentiation that you have in the organization?
Sure, Brian. Let's go. CI&T was founded in 1995 in Brazil with a vision of really helping business transform through technology, and the name CI&T stands for Collaborate, Innovate, and Transform, reflecting our vision from the very beginning. Over the past 29 years, we reinvented ourselves several times, I think, and in this way, we experienced consecutive growth. We have 29 years of consecutive growth, evolving from a very niche software engineering company to a global digital transformation partner for large companies across several industries, including names like Johnson & Johnson, Coca-Cola, Volkswagen, British Telecom, and ABI. With annual revenue around $520 million U.S. dollars, around approaching 7,000 people around the world. The U.S. is our largest and fastest growing market, accounting for 44% of our revenue, followed by Brazil with 40%, and then the U.K. with 11% and 4% in the Asia-Pacific region.
Our offering goes from very broad services inside the digital umbrella. We go from digital strategy, customer-centric design, and full-stack software engineering. So with that, we create what we call comprehensive digital solutions that help our clients to stay ahead in this evolving market. Finally, I think the news of the moment is our ongoing reinvention of CI&T around artificial intelligence. I think we are very committed to this new chapter of the digital revolution. We launched it successfully 18 months ago, CI&T Flow, our AI platform for hybrid digital. And I think there we showcase our ability with our clients in this infinite game of exploring tech possibilities and also keep up with evolving consumer behavior trends. So basically, this is CI&T.
Very good. Yeah, we're definitely going to be digging into Flow here as we go through the conversation. Before we get to that, let's just talk about some of the positioning from a delivery standpoint. So certainly, Latin America has become a top destination for custom software development and outsourcing. Talk about what makes the region and its talent base attractive, particularly into enterprise clients in the U.S.
Brian, what I see is basically four factors. First is kind of obvious, a very large pool of skilled talent with low geopolitical risks, and geopolitical now is important, right? Time zone alignment. In Latin America, we are between minus four and plus four hours covering North America, Western Europe. This alignment is crucial for enabling real-time collaboration and minimizing high overhead costs typically associated with onshore or onsite work that support offshoring out of time zone window work, so I think cost efficiency is also related to time zone. Regarding cost efficiency, I think it's a region with very competitive pricing, and this is natural for, of course, there's effects. It depends on the moment. One country is a little more competitive than another, but in general, the whole Latin America is very effective in terms of cost, and the fourth reason is more subject.
I think it's culture affinity. We have similar business practices, business consumer behaviors, including a very high adoption of digital and online services in the whole Latin America. And now artificial intelligence. We see massive adoption in Brazil, for example, of AI platforms and services. So I think it's basically the four reasons that are putting Latin America on the spot as an important source and location.
On the theme of Latin America, have you seen a notable change in client demand for tech services out of the region over the last three years since the war in Ukraine started?
Exactly. Absolutely. In addition to the conflict in Eastern Europe, which disrupted the sourcing landscape there, we have Russia totally out of, removed from any consideration for sourcing. China is increasingly isolated. So as a result, Latin America is rising significantly as a geopolitical low-risk alternative for sourcing. After over 20 years of my personal experience promoting Brazil and Latin America as a sourcing alternative for the US, this is by far the easiest and most opportune moment, I would say.
Yeah. And then on Brazil specifically, CI&T is a leader. It's a country that's been more of a challenge for competitors to penetrate and scale. So both from a delivery standpoint and local business for CI&T, why has it been the case that you guys have been able to establish this moat and really retain it over time?
So let me try to distill it. Regarding sourcing, Brazil, I repeat that, Brazil has the largest tech talent pool in Latin America by far, but it's also the most competitive. Established players like CI&T, 30 years there, digital powerhouses like Nubank, Mercado Livre, major corporations like Itaú, Bradesco, and a thriving startup in fintech ecosystems create an, let's say, intense competition for talent. So it's large, skilled, but competitive. Additionally, Brazil's labor laws are detailed and protective, adding complexity for companies unfamiliar with local regulations. So in an initial assessment, companies tend to avoid Brazil because of all this competitiveness and the local regulations. Regarding doing business in Brazil, I think we find the same pattern as the majority of emerging countries. High tax on imported services means you need to make local sourcing almost essential.
You need to crack the local sourcing nut before you can serve the local market. So cracking the nut of the Brazilian market first requires navigating all these complexities of local tech talent landscape. And after that, you can for sure gain traction with local businesses.
There's a lot of layers there on delivery and in mining business in the country. Okay. Maybe let's try and move that conversation over to demand. And maybe what you're at, what are you seeing out there as it relates to the enterprise conversations you're having? And how has that evolved over the last, let's say, two to three quarters?
I see the demand environment is slightly better despite the ongoing macro uncertainty. The highlight is that the tech budgets are stable. I keep saying that. This is very good news for the industry, especially large companies where we see less volatility and more visibility. So it's improving, not in a very radical way, but quarter by quarter is slightly better.
Okay. Is there kind of anything you'd call out on the pipeline dynamics? So understanding decision-making, maybe signing or ramping is steady, but is that opportunity set growing while that's happening?
Hard to answer that in general for the industry. But for CI&T, we see our commercial pipeline and bookings improving since Q1, making Q4 our stronger quarter in both terms, commercial activity and bookings. But it's our reality. I don't know if it's a general industry reality.
Okay. And maybe just going a little deeper on demand, anything to call out by industry vertical and/or geo as it relates to outperformers or underperformers?
Yeah. In terms of geo, I see the U.S. in a better mood now with the election result. Of course, a lot of uncertainty for the first months of Trump, but much better than a few months ago. So U.S. better mood. Brazil and U.K., I think, still reflecting the macro. So more complicated.
Yeah, and obviously.
You mentioned verticals. Verticals. Telco and telecom are still below normal levels, I would say. And we see increasing activities in financial services, consumer goods, and retail, especially in the U.S.
Good to hear. And budget planning process underway for calendar 2025 as we speak. Any early insights from some of these initial client budget discussions that you can share?
We are still working on our planning numbers for 2025, and this will be the main subject of our next earnings release.
Sure. Sure. How about large client performance? So as we think about the performance across 2024, it seems to be a pickup, a recent pickup in the top 10, particularly. Is it broadly performing across that cohort, or are there any select few that are driving that?
Yeah, it's nine of 10. The expansion with our top 10 clients last quarter was 10% sequentially and 25% year over year, so very strong. And the key driver, I believe, in this tangible results is our productive gains with the CI&T Flow platform. Remember, we started this move with our main clients, so these are the first companies to really see concretely the kind of efficiency gains we could provide. And for them, we are already providing improvements or showing, demonstrating improvements that range from 50%-200% faster in some context, and it's improving monthly. So I think it's part of our efficient hyperproductivity flow strategy, and we started with them with our largest clients.
Okay. Yeah, that makes sense. So I guess that's encouraging that it's flow-related and then potentially they have otherwise stabilized. Sounds like sequential growth, you're starting from a decent base to build upon as you go forward.
Yeah, we are confident in delivering, first, a very solid Q4 as we guided, and under the current market conditions and assume, let's say, effects neutrality, we expect to continue growth sequentially throughout 2025.
Okay. Just to follow up on the large client dynamics, the former top client, Anheuser-Busch, was obviously a notable headwind in 2023, given their own challenges that kind of hampered their spend. So maybe just a progress report update on ABI. Is it still a top client and any potential avenues for growth over the medium term that you can share?
Sure, sure. We are proud to continue to be the main partner with ABI on the BEES platform, their largest digital initiative by far. And it's still a top client for us. So it's stable. And we see a long-term partnership continue to happen with them.
Okay. Let's go a little bit higher level now as it relates to structural demand. So kind of backing away from top clients, backing away from just the '24 performance. If we think about the category, it has been historically perceived to be a 20% plus organic grower. That was certainly the case pre-COVID. Clearly, a lot of moving parts the last couple of years. But as you kind of step back and assess your growth opportunity and the industry, the digital engineering category's growth opportunity, is there anything structurally different going forward that wouldn't allow for that? Are you thinking about what drove it before, where we are now, where do we go from here?
Yeah, I can talk for hours about this, Brian, but I will try to summarize here. Basically, the three pillars of what I call the first chapter of the digital revolution. First, tech advancements. We got internet, mobile, cloud, big data, and social media. This is huge, and companies had to move. Second is customer expectation. I think the demand for personalized experience journeys, digital journeys increased a lot, and the third factor is competitive pressures. Business raced to stay ahead in this digital landscape, especially this battle among digital natives and incumbents. This is basically the equation of this first amazing area of investment around digital, and now, I think we are on the verge of what I call the second chapter of the digital revolution, boosted, of course, by artificial intelligence.
And if we look back at the pillars, tech advancements, we start the golden rush for efficiency, for hyperproductivity with AI that I call the act one of the AI revolution. Customer expectations in a timeframe of two-to-four years, we're going to see a radical evolution in the human-machine interaction. We're going to see a new layer of AI-based interaction being created, natural language based on natural language gestures and replacing the screens and buttons. So this will be huge. And I call this the act two. And finally, it's always important to see the competitive pressures. I believe macro environment and geography plays a role here. Incumbents will accelerate their investment in digital when they see concrete signs of competitive risks from native or other competitors.
And this is going to happen differently by industry, probably starting in more competitive customer-facing industries and moving backwards to B2B or resource-intensive verticals. So I see the beginning of, I would say, a much more radical digital revolution ahead built on the pillars of this first chapter I mentioned. So I forecast that after this short cycle of post-pandemic adjustment, we will rejoin a cycle, trend of inevitable high investment in digital and tech.
Okay. Okay. All the more reason then why some of those deep relationships with big enterprises are really important. Okay. Let's talk about Flow now. So a lot of optimism here, a lot of excitement. You've made a big investment in GenAI via the Flow platform. But for those that aren't really familiar about that, I want you to talk about the launch of that, maybe the value proposition, and the components of the platform just to give people an overview.
Sure, sure. First, we made a clear strategic decision in early 2023. Instead of just creating a new business unit for generative AI service, CI&T committed to transform itself into an AI-first tech service company, and to make this vision concrete, we partnered our five of our largest clients in different industries and geographies, and we created the first initial version of what we called the CI&T Flow platform. We launched it publicly. We launched it in July 2023, and we designed Flow as a layer on top of the best large language models, enterprise-ready in terms of security, privacy, and reliability that would allow the teams to create and evolve AI agents they need to solve specific problems in each one of our clients, so fast forward to this end of 2024, we reskilled more than 80% of the whole CI&T team, which now we call AI-boosted teams.
We are impacting more than 100 clients with efficiency gains, starting from what I said, a minimum of 50%, but goes from numbers that are scary. It depends on the context, and it's still evolving, so now Flow is the house of more than 2,000 agents growing exponentially, and our initial focus was to streamline the end-to-end flow of producing digital solutions from assessing customer needs to understanding business requirements, coding, testing, deployment, operating digital products. Everything we do, we saw as an opportunity to boost with AI agents, so this is basically the first sprint of CI&T Flow.
Okay. Yeah. So you're all in for sure based on that 80% plus of the workforce there being trained up on it. When you kind of thought about.
I say we burned the ships.
There you go. When you thought, when you guys the concept of the platform, how did you go about leveraging, maybe, third-party open-source models versus thinking about developing something natively? Maybe, is there an aspect of training that you have to put on top of that too? Can you talk a little bit about that?
Yeah. CI&T Flow is agnostic, basically leveraging a range of proprietary foundational models. I mean, GPT from OpenAI, Claude from Anthropic, and Gemini from Google. And also very good experience with the open-source models like Llama from Meta. But regarding developing new models, given the current state-of-the-art architecture of foundational models, training a PhD-level model requires what I call a pentagruelic amount of data and GPU power. So I don't see now a viable alternative with the current. Of course, everything can change, right? But with the current architecture of the large language model, I think a better path is fine-tuning, customizing existing models, whether proprietary or open-source. I think it's a far more viable strategy than trying to develop an entirely new native model.
However, we never know with a few, and we would need a few scientific breakthroughs that could change this game, making smaller domain-specific models economically feasible. This is not the reality now, and I don't see that happening probably in two to five years.
Okay. Maybe shifting to the demand side of GenAI and Flow specifically, any notable changes in the volume of programs moving into full-scale deployment more recently? Those are mostly still one-offs, and maybe you can just highlight one or two representative client program examples that are scaling just to give us a better sense of what they may look like.
Sure, sure. We are publishing the evolution of our metrics, including client testimonials in a quarterly report we call AI Pulse. We can check it out, and everyone can check it out on our website. Our key metrics, I mentioned, 80% of the teams, including coders and non-coders, are intensively using AI daily, more than 190, an extraordinary case. Flow now represents a new way of work. So in terms of software development, we can say that it's ready full-scale in production. But we also are highlighting use cases. We are developing and scaling with our clients. In our recently earnings calls, we highlight powerful stories with big names like Bradesco, British Telecom, Itaú, BASF, YDUQS, Audi, B3 in financial education, several others. I think our ratio of putting POCs in production scale is tremendously high.
I believe it's because of our approach and the way we are guiding our clients to leverage these opportunities.
Got it. And what are key areas of client tech spend that Flow can unlock in the near term and why? Like within client engagements, what area have you seen shown some of the strongest efficiency gains?
Top hits now are what we call hyperproductive teams, where we replace like 10 squads of underperforming competitors by five squads of AI-boosted CI&T teams. We call this hyperproductive teams. Application or legacy modernization is also huge. A lot of things that would take three to five years to rebuild or recode, not lift and shift, but recode or redesign for a cloud-based architecture now can be done in six to nine months. So the return on investment is finally there. So companies can really face the dragon of the legacy systems and design or reshape their whole architecture to be cloud-based and prepare for this new AI world. So I would say that hyperproductive teams and legacy modernization are the big hits now.
Has Flow impacted win rates in a meaningful way? Or like you mentioned, consolidation opportunity there as an example. What are you seeing so far from a competitive standpoint using Flow to either get new clients or wedge other existing relationships or existing vendors out of the way?
I think it's by design in our strategy, right? It was what we were looking for. And I think, yes, it's been a very good year in the commercial front, mainly because of Flow and the way we are proposing Flow as a way to leverage efficiency first.
Okay. Good. And then just on the, if we think about Flow within the context of the software development lifecycle impact, how do you see agentic AI impacting the software development lifecycle? Just maybe talk about where in the lifecycle you've seen early success, and then conversely, where are you seeing some limitations?
Zach, in the next 10 years, not true. Hold your horses, as I say for our clients. Software engineering will be turned upside down end to end. Not a single task that we believe is a traditional way of developing software will remain unchanged. On the other hand, I think more challenging than just optimizing each one of these individual tasks will be to redesign a new way of a new methodological way of creating digital solutions based on AI. We have this ambition also to be a protagonist in this area. I think we did that in the first chapter of the digital revolution. We created the Lean Digital approach. I think it is a huge contribution of CI&T for the industry. We understand that we will have to really redesign the way we produce software and digital solutions.
But we are in the early beginning of this curve. It's a 10-year task to see a totally different landscape in software development.
Gotcha. And then internally, outside of client engagements, like software development, are you using this technology in other areas like hiring, resourcing, or training? I'm just kind of curious if you are where you're seeing efficiency gains, and is there potential for structurally higher levels of efficiency gains in areas like utilization?
Zach, I love this question because I think this was one of the main things we got right in our AI strategy since day one. Just a few weeks after we internally launched the big Flow vision, I also launched a program we call UPrompt, fostering that every single CI&T from any functional area, I mean, marketing, HR, finance, even Galvão from investor relations, should start a journey of becoming an AI-boosted professional. And now fast forward 18 months, we have the same level. We measure that. We have the same level of daily utilization of Flow in our internal support areas as we have in our digital production frontline teams. This is an amazing achievement. And to give you a result, a concrete result, recently we mentioned we onboarded more than in Q3, more than 500 new employees.
And our utilization, normally during this large onboarding, you have a decrease in the utilization rate because you need training to prepare these new employees for the projects. But our utilization rate increased from 85%-89% in the quarter, mainly because we were able to use AI to reduce the onboarding time, or we can call it the time-to-build ability for every single new employee. This is one internal result of UPrompt and applying or dogfooding our vision around AI.
That's a good proof point right there internally. Run to clients with that and win more work. Okay. We got a minute or two left here. I got a question from the audience. The question here, does the liberalization of Argentina impact CI&T sourcing strategy and Latin America domestic business focus? So understanding your Brazilian base and leadership there, but what about Argentina here and any dynamics there?
Yeah, I think we see Latin America as a huge opportunity. We are now fostering Colombia as a second large talent center for CI&T. We are, of course, looking for Mexico, Argentina, and other countries as a good next step for increasing our diversification of sourcing in Latin America. So we are following. I think this is, we hope the economic stability will allow us to count on Argentina as a good sourcing location. The current agenda or the 2025 agenda for Latin America is fostering Colombia as much as possible.
Okay. Last question for you. Just circling back to GenAI, the business impact considerations. So as you think about it, you mentioned 10 years down the road, end to end, but even sooner than that, things getting more productive in these relationships. How is that going to potentially change the model for the services business? What do you have to do differently, contracting terms, things like that?
Yeah. As you know, the digital service industry is now heavily based on time and material commercial models, and I see the possibility of an incremental move or more room for fixed-price deals or output-based deals or performance-based deals. This is a dream for the vendor part of this equation now. But it's a kind of structural change in the procurement areas of very large companies, so if you consider our cohort, so it will take a while to become relevant. Facing the reality that large companies don't change their internal processes from Friday to Monday, it will take a while, but I think it will evolve. Of course, with smaller or digital native players, you can easily test new models, and we are doing that, but for CI&T, 90% of our business comes from large or very large companies.
We are proactively addressing them regarding moving from rate cards to a more smart way to assess the players. It will be an incremental education, I think.
Okay. And so gradually, so time to navigate as that occurs. All right, very good. Well, we are out of time, Cesar, Eduardo, and Stanley behind the scenes. We appreciate your time today. For the audience, thank you for joining. Everybody, have a good day.
Thank you. Thank you for having.