Bowhead Specialty Holdings Inc. ($BOW)

All right. Cave, go ahead.

Thank you. Good morning, everyone, and thank you for joining us today. My name is Cave Montazeri, and I'll be your host today. We have senior leaders from Bowhead Specialty here to discuss how technology and AI are being deployed at a leading insurance company. Bowhead has built a specialty carrier with both traditional craft underwriting and a growing digital model. What I'd like to talk about today and understand where technology is changing the speed, quality and economics of specialty insurance, and how investors should separate durable progress from broad industry noise. Our panelists today include Brandon Mezick, Head of Digital Underwriting. Brandon has almost 20 years of experience in specialty insurance with leadership roles spanning underwriting, operations and business strategy. Brandon will tell us how Bowhead is using frontier technology to deliver a low-touch underwriting process for SME, E&S customers in construction and real estate and how the company plans to scale this further. Also joining us today is Steve Feltner, who serves as Bowhead Chief Operating Officer. Steve will explain how technology can augment human underwriters in complex specialty lines where human judgment remains critical and can be supported by better tools, data and workflow design. Steve has 15 years of industry experience with a background that spans both technical and operational disciplines, giving him a broad perspective on the evolving needs of specialty insurance organizations. Last but not least, Brad Mulcahey, the Chief Financial Officer, will tie it together, bring the financial implications and tell us why the company is excited about digital tools that did not exist when Bowhead IPO-ed just 2 years ago that are now embedded in the company's daily processes. One little logistical note, please e-mail me any questions you might have that we can -- will have time at the end of this fireside chat to address or if you want to send me an IB on Bloomberg. So let's start with Brandon. For investors who may be newer to the story, how do you define the difference between craft and digital inside Bowhead. Why do you have those two different platforms under the same roof? And where does technology create the most value?

Thank you for the question, Cave. First, I'd like to maybe start by thanking you and the folks at Deutsche Bank for hosting us this morning. We're really excited to tell our story in a longer form. So thank you again for the opportunity. Maybe instead of starting with the differences, I'd start with something that unites both the craft and digital business, and that is our focus on underwriting profitability. That is the essential objective of Bowhead is at the core of everything that we do. But to your question on differences, a simple way to describe it might be in craft, our underwriters are making judgments that a model can't yet make reliably. Where in digital, decisioning is well defined, and we think speed and consistency improves the outcomes. So in terms of technology in craft, the underwriters are synthesizing dozens of signals we get from submissions, from conversations with brokers, from conversations with customers that don't live in any kind of structured data set. So in terms of the value technology brings, it's really about augmentation, giving those underwriters faster access to information like loss histories or the exposure information for a customer or benchmarking pricing data. It eliminates the friction around judgment. It does not eliminate the judgment itself. In terms of technology in digital, the risk parameters here are much tighter. The data is more complete, and we think the decisions then are more repeatable. So in terms of value in digital, technology can be the underwriter. Humans obviously are in the loop for edge cases and portfolio monitoring. But I would note that, these aren't two separate businesses that just happen to sit under one roof. The learnings and intelligence that we gain in each business informs how we think about the other. They really feed each other.

So Bowhead digital business today includes Baleen general liability for contractors and real estate plus Express cyber and miscellaneous Professional Liability. Why were those clients the right place to start? What makes them suitable to a more standardized workflow without compromising underwriting discipline?

We were really deliberate about this, the lines that we started with, as you mentioned, primary general liability in Baleen and cyber and MPL through Express, share at least 3 characteristics that we think make them really suitable for a standardized workflow. The first is data completeness. For these classes, the data that we need to make sound underwriting decisions are largely available at the time of submission. We're not really dependent on broker narratives to fill in gaps. We've got structured, exposure data, loss data. And in the case of cyber, great third-party security data that enriches the customer profiles we're trying to build. The second is a defined risk appetite. These aren't lines of business where we're trying to underwrite something unusual. We have a clear appetite and those rules can be expressed in a way that a system can apply consistently. So it's not dumbing down the underwriting instead, it's controlling discipline. And the third is we're building [indiscernible] portfolios in each of these businesses. At the smaller end of the market, the exposure units are more comparable to each other. For us, that means that things like a pricing model can be more reliably calibrated over time. That's one example. I would just say that this doesn't -- we're not compromising underwriting discipline because when done correctly, a rules-based workflow eliminates the adverse selection that can creep in when individual underwriters are applying slightly different standards to the same risk.

So make that concrete, can you walk us through what digital submissions looks like today from intake to quote to buying? What parts of that process are now fully automated, what parts are low touch, where human judgment is still absolutely essential?

Sure. So it starts with intake in both Baleen and in Express, most submissions are sent in the old-fashioned way via e-mail. The moment that e-mail arrives, our systems do several things in parallel. They scan the applications and loss runs. They pull third-party data to enrich the information we have about a customer. Ultimately, that intake process yields a comprehensive customer profile that then gets compared to our appetite matrix. This is then when the process is probably diverge between Baleen and Express. In Baleen, if the account fits our appetite and doesn't trigger a referral, a pricing engine and the document engines fire off practically automatically, both documents and specimen policies are assembled and those are sent back to the submitting broker without human intervention. In Express, all of that relevant underwriting information is presented on a single pane of glass to the underwriter. As I mentioned in my remarks yesterday, that underwriter's review is structured to take less than 15 minutes per risk. After their review is complete, both documents and specimen policies are assembled and sent back to the submitting broker. And here's where the processes converge again, if and when a customer decides to purchase, brokers visit a self-serve site to complete some compliance information. And once they collect and find an issue, we deliver a complete policy to their e-mail in under 5 minutes. On the point about human intervention, we think it's essential in at least 4 places. First, in setting and updating the business rules that the system runs on; two, in reviewing edge cases that the system might not be confident about; three, evaluating risks that trigger those referrals I mentioned for things like loss activity; and then fourth, for portfolio monitoring at the aggregate level. I would just say we think a lot about that workflow in the context of the experience we're delivering to brokers. We know brokers get especially frustrated when underwriters take a long time to even acknowledge their submission or offer any response. It's worse when an underwriter takes days to respond, ask a bunch of questions and then ultimately declines the risks. Our brokers do not have those experiences. If a risk is out of appetite, we tell you, but most often, we offer bindable quotes and we do it quickly.

Let's turn to Steve. On the craft side, how much of an underwriter is still consumed by lower value administrative work rather than actual risk selection and broker dialogue? Or have you already made the biggest gains in reducing that bridge?

Thanks, Cave. Like Brandon, I'll just take this chance to thank you and Deutsche Bank for hosting this event for us. But to answer your question, in craft, we think about this a bit differently than digital. We've deployed technology in our Casualty unit around submission intake, data enrichment and triage, but the objective isn't to abstract the work away from the underwriter, it's to set them up to do it better. The goal is a true single pane of glass. We want the underwriter to have everything they need in one place and not Google searching for basic information on an account. In digital, the technology often summarizes and streamlines decisions for speed and consistency. In craft, we take a different approach. We tee up the data and structure it, but we preserve the underlying detail because these are more complex risks and the edge comes from the underwriters' ability to work through the nuance, not around it. So while some might call that administrative work, we don't. In our view, that is core to disciplined underwriting, and that's the balance we're after in craft, efficiency around the work, not simplification of the work itself. So building on that, as triage capabilities improve, where does the value show up first? Is it better time allocation, stronger puts you bind on the submissions, you do touch with the ability to handle more top of the funnel volume with the same team. For triage, you actually see the benefit in that order, and that sequencing matters. So first, it shows up in time allocation. The earliest and most immediate benefit of better triage is simply giving time back to the underwriters. You're cutting out the noise, meaning low quality or out of appetite submissions so that they can spend more time on risk that actually warrant judgment. Second, you see it in the quote to bind quality. Once underwriters are more focused on quality accounts, the hit rate improves. They're quoting more quality deals that are better structured, better priced and more aligned with our appetite. That's where you start to see real underwriting leverage come through. Third and only then does it scale capacity. With cleaner intake and more efficient workflows, the same team can handle more top-of-funnel volume without adding headcount. The goal is to scale [indiscernible] not just volume for volume's sake. Efficiency is nice, capacity is valuable, but better risk selection and stronger quote to bind is what ultimately drives margin and long-term performance.

Can you give us a concrete example of a process change or a technology enhancement that has made Bowhead measurably faster, more accurate or more efficient? And one example of the craft side and one of the digital side.

Sure. This is Brandon. I'll start with digital and then turn it to Steve for example in craft. The best example I can give you on the digital side is how we built the automation layer for Cyber Express. When we launched that business, we didn't start by telling that platform what to do. We started by watching what our underwriters actually did. We launched on a low-touch basis for all risks. And behind the scenes, we studied the underwriters' decisions, which forms they added or remove, which businesses they reclassified where they may be credited or debited a risk. Once we had enough data to see those patterns clearly, we encoded that into a rule set and then applied that to the simplest, most homogenous segment of the book where decisions were the most consistent. Those risks became fully automated, the more complex risks in Cyber Express paved in a human's hands. I think this approach matters strategically because we know why the rules are what they are, we can update them when the environment changes, which it often does in cyber liability, and we have a baseline of human behavior to audit against as we move forward. Steve, do you want to talk about examples in craft?

Yes, sure. And I'll provide two examples, if that's okay. So we recently deployed a tool to our underwriters that reads underlying policies, extracts and organizes terms and conditions and presents the information in a clean structured dashboard for the underwriter. Historically, this was an extremely manual process, underwriters would spend considerable time reading through lengthy underlying policy documents to determine what coverage they were following, where exclusions were endorsements differed and how the structure compared across carriers. That work is important, but it's also time intensive. What the technology does is accelerate the understanding, but not eliminate the judgment. So instead of spending hours locating and organizing the information, they can now spend their time interpreting the information and making better underwriting decisions. Also, I just mentioned the platform we use in our Casualty unit for triage amongst other functionalities. Before committing to this platform, we ran a controlled pilot within our Casualty team, one group of underwriters using the platform and another continuing in the existing workflow. After just 1 month, the results were clear in the data. Underwriters using the platform produced roughly 30% more quotes than the control group. The real insight wasn't just higher output, it was better allocation of underwriting time. Getting to a no faster allowed our underwriters to spend more time reviewing quality risk and engaging with brokers, which in turn generally leads to more submissions.

[Operator Instructions] So back to Brandon. One structural advantage Bowhead had from the start is that you are building a new organization rather than inheriting decades of legacy systems and [indiscernible]. In practical terms, what does that allow you to do faster today, whether that is launching products, adding data sources, changing rules or integrating third-party tools?

The short answer is yes to all of those. We think the advantage is real and compounding. When you build a modern modular tech stack from day 1 without a legacy layer or trying to retrofit, we believe it's a huge advantage. Every capability we've added has been additive and not a replacement of something that isn't working. So just working through some of those dimensions first on product launches. Once we complete the product design, which really takes a lot of time and requires extensive collaboration between disciplines, bringing a new digital product to market can happen in a few weeks rather than a few quarters in terms of the technology implementation. The infrastructure we have in place for things like intake pricing, issuance, data capture already exists. So a new product is essentially a new configuration on top of that infrastructure plus some distribution work, which is really important. We usually start with a soft launch to ensure that technology works as expected and to make sure there's a strong product market fit. The second is around data sources, as you mentioned, adding a new third-party data source is a day's long integration, not like a month's long IT project. It matters a lot because the data ecosystem for underwriting is evolving quickly. Things like new cyber risk signals are available. There are new financial health indicators that are available. All of that help with underwriting decision-making and being able to plug those in quickly is a real advantage. Third on rule changes, when market conditions shift and we need to update things like appetite or pricing, we can do that in the platform without messing with the underlying code. The underwriting team can change how we're treating a class or a geography and seeing it flow through submissions the next day. Ultimately, I think this is what the market has long chased its speed without sacrificing underwriting quality. As I mentioned yesterday, most approaches have traded something away to get their large teams of low-cost labor, legacy systems being forced to do things they weren't designed for or asking people to work harder. Baleen was built on a different premise, being a digital-first company means that one of our competitive advantages is organizational agility. We make decisions faster. We iterate on our models more quickly, and we have really strong alignment between claims, actuarial, underwriting, technologists, product operations professionals. They share this common view of risk without the internal friction that might come with size. The carriers with the biggest budgets are playing in a very different game. We are not trying to out-scale them. We are, however, trying to out-think and out-execute them in the markets that we've chosen.

How do you measure if what you're doing is actually working? What are the most important metrics internally? Is it turnaround time, underwriter productivity, buying rates, expense efficiency, broker adoption, loss performance, anything else?

All of those guys. I would say that the expense ratio that you mentioned, it definitely matters and investors are right to focus on it. But internally, our view of technology effectiveness is a little more granular. We think about this across 3 dimensions. The first is maybe throughput and speed. Here, we're measuring things like time from submission to quote, how often we're converting those quotes into binds and how many risks are going straight through without human intervention. All of those are leading indicators on the platform's efficiency. The second dimension is the quality. So here, we're looking at things like the accuracy rate on automated decisions. Here, we measure those against what a senior underwriter might do if they review that same file. We also look at things like the rate of midterm endorsements or corrections that we need to make driven by data errors or early loss indicators on the digital book versus craft. Here, back to quality, technology just shouldn't be faster. We think it should also be more accurate. And then the third dimension we focus on is broker adoption and friction. So here, we're looking at things like the repeat submission rate of brokers and the speed that we're rolling out access across all of their networks of offices. The brokers aren't coming back, something about the experience isn't working, either speed or coverage or communication. We treat the way brokers are adopting us as a signal of the platform's quality. Maybe just to wrap this up, we do measure a lot. There are two metrics that I would highlight that define the broker experience, and that is speed and quotability. In the first quarter, as I mentioned yesterday, we responded to 75% of submissions within 15 minutes and 100% of submissions within 1 business day. We also quoted 75% of our new business submissions in the first quarter. So to me, that means brokers are getting fast answers they can actually use.

Great. Let's talk about financials now. So your expense ratio fell below 30% in 2025, outperforming your previous guidance of low 30s. You're not expecting the expense ratio to be below 30% in 2026. So Brad, how should investors separate the benefit of scale from the benefit of actual automation?

Yes. Thanks, Cave. I think if we back up a little bit, our older guide, if you will, was the low 30s, so call it, 31%, 32% expense ratio. That came as Baleen had just started and Express was nearly an idea at the time. That previous target of low 30s, low 30% had some assumptions on continuing to scale. We are still early in the investment mode with digital and automation in general. We knew the digital strategy should be accretive to the craft expense ratio when it's fully mature. But as we invested in the digital platform before the premium was earned, it actually added pressure to the expense ratio. So recently, we've trended below 30%, in fact, it was 28.4% in Q1. But as you know, we try not to overemphasize any individual quarter to look more at the longer-term trend. This recent trend below 30% is mostly from the continued scaling and tech initiatives on the craft side. We're still in investment mode in digital, but already seeing the benefits of those early investments Steve and Brandon mentioned. In fact, I'd say our confidence in the digital strategies impact on the expense ratio increases with each passing quarter, I wouldn't be surprised if we land well below 30% once digital is fully mature and up and running.

So there's a lot of excitement around AI. But the practical question is where it can be trusted what still needs guards. When you think about Bowhead's road map, where do you see the most credible use cases for AI in advanced analytics? And what governance standards need to be in place before these tools can play a more meaningful role in underwriting or claims?

This is Brandon. I'll maybe start with some use cases in underwriting and then turn it over to Steve to discuss the same in claims. Within underwriting, I think there's this continuum that we think about based on credibility or maybe said another way, how much the output can be verified. So on one end, we see some really high credibility use cases. Those are things like document analysis, data extraction, like reading submissions, processing endorsement requests, generating loss runs. In all of these cases, the output is verifiable. You can catch the error if they make one and the efficiencies that you can gain are significant. In the middle of that continuum, there are some use cases that we think are best operationalized with guardrails. One example might be a chat-like interface or an internal tool set, so helping underwriters the book or generate comparisons or even draft some endorsement language. Importantly, for all of those use cases, the output is going through a human review before it gets implemented or before it affects a transaction. And then on the other end of that continuum, there are cases that we believe are least credible. And so an example there might be fully autonomous AI-generated terms on anything outside of a very narrow rules-based digital context, plus I think the regulatory clarity over those use cases just simply isn't there right now. Steve, do you want to maybe share some insights on claims?

Yes, sure. And if it's okay, I'll use a similar framework to Brandon because claims like underwriting really operates along a credibility continuum as well. So at the high confidence end, you automate aggressively. So these are repeatable rules-based processes where speed and consistency matter the most. Examples in claims would be first notice of loss and taking classification, acknowledgment communications, drafting standardized correspondence like coverage letters. These are areas where automation makes process fast and more consistent without sacrificing judgment. In the middle, that's where AI becomes a force multiplier. So here, technology is augmenting, not replacing the adjuster. Examples of this in claims would be risk and severity assessments, summarizing the incident defects, supporting coverage analysis with relevant policy language. In these cases, the adjuster remains fully in the loop, but is better informed and faster to act. And then there's some low credibility cases at the other end of the spectrum. These would be speculative use cases like AI predicting claims outcomes and autonomously paying or denying claims. Like Brandon said, we don't view these as credible today. There are some regulatory hurdles to get through there as well, particularly in our complex specialty lines. So in summary, I think our approach here is very deliberate. We automate where confidence is high. We augment where judgment matters and avoid overreach where credibility just is [indiscernible].

Let's talk about cyber. So very interesting line because the risk environment can change very quickly, especially with the new development in AI. As you automate more of the smaller risk cyber workflow, how quickly does the technological feedback loop adjust to new risk entering the market?

This is Brandon again. I think that's exactly the right question for cyber, and it's why the way we design the system matters enormously. We've addressed this in a few ways. First, the cyber appetite rules are structured in a way that they can be updated quickly. As I mentioned previously, underwriters can change a parameter overnight. So when something like a new type of ransomware emerges, we're not waiting for an IT cycle to adjust eligibility criteria, for example. Second, we use cyber risk signals at intake that update in near real time. So the risk assessment on a new submission always reflects current signals, not just what an insurer reported on an application from 90 days ago. In a class of business where exposures can change a lot between application and buying, we think that really matters. And third, we're monitoring our cyber book continuously. If we see claim activity starting to cluster around a technology provider or a sector or industry, that triggers underwriting review. And we have this continuous feedback loop between claims, underwriting and actuarial that ensures we're making any necessary adjustments quickly. Lastly, just on cyber, I appreciate that there are different views on cyber across the industry, different views make a market. Some carriers avoid large risks because they typically experience the largest claims. Other carriers may avoid small or medium-sized risks because of concerns about the sophistication of their cyber defenses. At Bowhead, we're in market for cyber risks across the customer continuum. We have platforms to underwrite both small risks quickly, which were designed by expert underwriters, and we have platforms to evaluate large complex customers, which are handled directly by some of the best underwriters in the industry. We don't change the way we underwrite based on size. Our essential approach and standards for things like eligibility are shared between craft and digital.

Great. And frankly, my last question before we open up for Q&A. In specialty insurance, do you think technology becomes a true moat or do vendor platforms eventually make everyone look more similar? Where does Bowhead believe this real edge comes from?

Yes. We think about this a lot. It's like the biggest long-term strategic question for us. I think the honest answer is technology alone is not a moat. The combination of technology, data, underwriting expertise, assembled and integrated over time is the moat, and that is really hard to replicate. Vendor platforms have done a good job of commoditizing certain activities like managing workflows or processing documents. And I do hope that over the next 5 years, the baseline technology standard across the industry does rise. It's great if it does, it does not worry us. But what vendor platforms can't replicate. First is proprietary data. That proprietary data that we collect compounds every submission we process, every policy we find, every claim we pay creates a feedback loop that informs how we evaluate and price the next risk. The second thing that it doesn't commoditize is the underwriting culture and decision-making framework that we've built, how our underwriters interact with each other and with the platforms, what type of risks get preferred? How do we use data to help with decision-making? That institutional knowledge is ours. The tools might be available, but the underwriting judgment embedded in how you use them isn't. So in summary, I would just say that our actual edge isn't any one piece of technology. Instead, it's a combination of decades of craft underwriting expertise and the technology infrastructure to scale it.

Operator, should we open it up with Q&A?

Sure. We have a question from Scott.

Scott Barishaw from Deutsche Bank. I just wanted to ask about, obviously, the focus of the call is technology use. And it's a couple of things around where do you see Baleen being as a percentage of total premium. So in the quarter, you guys did $11 million or so of premium, 300% increase. It's -- you guys do about $200-and-change million of premium. Like where can we think about this in the next couple of years? I looked at Cave's model, it kind of goes up a little bit in percentage. But like is there the possibility that this becomes a really bigger percentage of the total premium in the next few years?

Scott, I would say, yes, I wouldn't put any particular number on it. I would say that we expect that contribution to grow meaningfully. The drivers of that growth are real and durable. We have -- and we're activating more broker relationships. We are seeing more submissions. We're expanding the product pipeline. So I would say the 7% or so of total Bowhead GWP is our starting point and not our ceiling.

Okay. And so maybe just jump into sort of some of -- like we've spent a lot of time with some very tech forward insurtechs where they're on thousands of brokers, desktops. Like how is that process going for you guys? And are there any holdbacks from brokers saying not -- what -- has there been any pushback from brokers, I guess, would be the question for Baleen?

There hasn't. And I think one observation we made in our research and development phase was that brokers have portal fatigue. They do not like having to go to one or many portals to get one or many quotes. Their process, by and large, for the kind of business that Baleen targets is to, if you can believe it, send underwriters a submission via e-mail. And so our approach was to meet them where they are. But to work at the speed that those platforms portals can deliver, which is why we're so obsessed with this 15-minute turnaround time or less in Baleen and Express. I would also say, Scott, the appetite of markets on most portals are limited. They might not have the same willingness to write hard-to-place customers in terms of nature of operations, venue or loss history. And so I think that's where some of the platforms struggle and where Baleen really fills the gap.

And then I guess my final question is thinking about Baleen as a part of Bowhead, is that -- I mean, you guys have an iconic CEO in Stephen Sills. Has that -- like can Cave and I go out and create Baleen? Or like are you guys at just such an advantage of being a part of such a unique company of Bowhead and having a CEO like that. Maybe talk about just sort of the benefit of being a part of Bowhead.

I think Baleen has been able to see the success that we have because we're a part of Bowhead. And I'm also happy to report that I think Bowhead is benefiting from the investments we're making in digital, as Brad mentioned. The way that we created Baleen within the company, I think, is a really unique story. Stephen said, I want you to effectively go on Mars to build this business. That approach, it sounds simple, but it is awfully hard to do in larger companies where there is a magnetic force to use what you have and to play by a certain set of rules. You're right, Stephen is an icon, he's a visionary. He's been incredible to build Bowhead and Baleen. But I do think Baleen benefits from being a part of Bowhead and vice versa.

Well, congrats on the success so far and look forward to seeing it continue.

If I may follow up on Scott's last point. Can you talk about Express because yesterday, during the call, you did mention this kind of like positive feedback loop where Express is on the craft business, some business that craft has [indiscernible] but couldn't do it because it was too small, not profitable enough. But then also Express, you could help also get more broker mind share and helping the craft. Could you maybe talk about that and give us a bit more details on that little feedback loop?

Yes. The example that I give is in previous versions of Bowhead, we would go and we launched our cyber offering initially. We were a market for access and for large customers. We would walk into a broker's office because we're wearing the Bowhead jersey, 25 brokers show up. to hear our pitch on cyber. And when we say we can offer access for large customers, I suspect we lost 30% of the room because they weren't handling excess for large customers. Our story and our pitch to brokers today is very different. We say sort of like I did, we're a market for cyber. If you've got small risks, think of us, if you have large risks, also think of us. I think we are -- the message is simpler to brokers. I think it's more compelling. And we see the impacts of that in the craft submission numbers, which I think are being helped by the Express offerings and vice versa. So it is sort of this halo effect that we're experiencing. I think it all has to do with the simplicity of the message and how relevant our offerings are to brokers. do you have any more.

Do we have any more questions on the line?

Not at this time.

All right. Well, we've got one question by e-mail that I can address to Brad. So Bowhead was founded in 2020 as a remote-first hybrid company. How does that shape the company's culture and the type of talent you've been able to hire? Do you think you have attracted employees that are more likely to embrace change and new technologies?

Yes, good question. I think if we go back to 2020 and you think about when Bowhead started and the world was sort of ending, the type of people that came on board early were join a start-up in that environment were obviously highly entrepreneurial, thriving and change. Once that base was there, the tech person, I would say that we hired, I would say it's considered -- it could be called a builder, someone looking to make a difference, have an impact. We recruited a lot of people from larger organizations with really deep expertise, but there's too much bureaucracy, not enough accountability at those organizations. So that type of person obviously embraces new technology by definition. More recently, we've been able to hire people who prefer our remote-first hybrid workplace over the return-to-office mandates that we're seeing out there. Obviously, this allows us to hire expertise regardless of their location. We trade large rent expenses for smaller travel expenses that we still get together regularly. But the secret to our success, as Brandon mentioned, people that are remote is applying that culture of expertise over the technology that we're building, especially on the digital strategy. The expertise we bring in on the craft business informs and perpetuates the digital business which then culminates in a better broker experience in both craft and digital. So we believe that better broker experience ultimately results in higher growth for the company.

That was a great way to end this little panel. So Brandon, Steve and Brad, thank you very much for your time and looking forward to keeping in touch. Operator, we're ready to close the Q&A.

Thank you.