American International Group, Inc. (AIG) Earnings Call Transcript & Summary

Okay. Good afternoon all for our final session of the day. We did save the best for last. So I want to introduce Peter Zaffino, CEO of AIG. We'll start with some introductory comments, and then we'll jump into Q&A.

Thanks, Meyer. Good afternoon, everybody. I just wanted to take a few minutes just to make some introductory comments, and then I look forward to spending some time with Meyer and some of the questions from the group. It's great to be back here. We appreciate the opportunity to spend some time and present to you. In many ways, 2025 has been a real pivotal year for AIG. One of our defining moments, we talk a lot about it was at Investor Day in March. We set out a few key objectives there, which is really important for this year is we really want to put the past of AIG behind us. I want to spend quite a bit of time providing more details on the company we are today, and I'm sure we'll talk more about that over the next hour and demonstrate the positive impact, the work our colleagues have done to reposition the company. We highlighted and wanted to emphasize a few very important things, which were our underwriting culture. We've developed operational excellence, and that's been from driving end-to-end process to digitizing workflow to what we're doing today with Gen AI, and of course, executing on a very disciplined capital management strategy, all of that has given us a lot of financial and strategic flexibility for today. We also wanted to like introduce not capabilities, but more of what we were doing in terms of Gen AI and actually has received a lot of attention since Investor Day, and I'm sure we'll talk about it today. We outlined ambitious long-term financial targets and remain on track to achieve those. If I could spend a few minutes just highlighting the second quarter. Adjusted after-tax income per share increased 56% year-over-year and that was driven by underwriting income, which increased 46%. Our core operating ROE increased to 11.7% in the quarter and our calendar year combined ratio was 89.3%. So we thought that was an outstanding result for the second quarter. For the first half of the year, our EPS growth is tracking slightly above the 3-year objective. We talked about 20%, notwithstanding a lot of CAT activity, the wildfires in the first quarter. So we're really pleased with the progress we've made through the first 6 months. We also returned $4.5 billion of capital to shareholders year-to-date through the second quarter and delivered a third consecutive year of double-digit dividend growth. We also reached another significant milestone, S&P and Moody's upgraded the financial strength of our insurance company subsidiaries. And it's worth noting, this is the first upgrade that AIG received from Moody's since 1990. We had to go back quite a ways and so that was quite an achievement for us. We retired an additional $830 million of debt in the second quarter, now have a debt to total capital ratio of 17.9%, which is one of the best in the industry. We did talk a lot about, on the recent earnings call, the rate environment. I'm sure we'll cover it today. But generally speaking, Property has had headwinds in 2025. Casualty has actually been strong particularly in Excess Casualty, and Financial Lines has finally flattened out after a period of rate reduction. Property, I spent a decent amount of time talking about actually, what are the components of Property pricing and how you actually derive the rate making and how do you drive profitability? What's the impact of reinsurance? How do you look at AALs? And wanted to make sure that we emphasize the level of diligence that we do in making sure we have a fully loaded CAT cost, and all of our Property costs are very transparent and we know how they move depending on market conditions. So we're well positioned to manage this environment, focus on growth opportunities, but most importantly underwriting profitability. Before I shift to Q&A, I just do want to share a couple of comments on Gen AI. It's been a key focus for us. We want to embed it into our core business and end-to-end processes supporting underwriting and claims. The early results have been very promising. At Investor Day, I spoke about what were really aspirations. We had some pilots. We had some rollouts. But quite frankly, getting some of the tech leaders to join us and talk about what they thought was the work that we're doing. But in many ways, we've actually made more progress than what I thought we could do from when I outlined this originally in March just because of the rollout, tech advancements have been significant. And so we've rolled out a few pilots. It was started years ago, but a lot of companies are still testing. We're actually live now on things. And so I think that was important to mention. We started to roll out a component part of our Financial Lines. And within the year, we'll be rolling out our E&S business, which is Lexington, and then we expect to deploy more of the commercial business in 2026. I do want to leave you with a sense of how fast things are moving. Some of the tech companies and hyperscalers are really driving the pace of innovation at a significant level. Their commitments are substantial in terms of CapEx. I mean, combined Microsoft, Meta, Alphabet and Amazon announced their intent to spend over $400 billion in CapEx on AI just this year. So I think in 2026, we ought to see this acceleration in terms of not only what we're planning, but new horizons as we start to roll all of this out. So those are just some of the highlights. I know coming after a second quarter, maybe not a lot of new information, but I just wanted to emphasize all the progress that we're making, executing incredibly well, momentum into the second half of the year. So that's what I just want to say for prepared comments. And then Meyer, maybe I'll just transfer over to you for some Q&A.

Perfect. And yes, and I should point out 2 things: one, that Peter is providing an update every year. And every year, the update is dramatically beyond where it was the prior year. So there's been a tremendous amount going on there. Related to that also, I think my question list is going to be all over the place because again there are so many things going on at AIG that are worth talking about. So let me start with one. We spent a lot of time talking about talent and recruitment. You recently announced that John Neal is joining AIG as a President. What does the President of AIG do? And what should we, on the investment side, expect from John?

Well, I was the President, so I'd say we actually did quite a bit. So he's got some...

Well, what's left?

John is one of the top executives in the insurance business. I've known him for a very long time. Actually, when I was at Guy Carpenter, I used to work with John before he even took over QBE on a lot of the reinsurance that is. I've known him very well for a long period of time. He's incredibly capable. He's had some unbelievable experiences, whether it's running QBE or most recently running Lloyd's. I've talked to some of the investors earlier today, I think there's not a lot of companies that are really global and have a broad footprint, but John actually has a lot of global experience. So he's going to fit incredibly well within AIG. Strong track record of underwriting strategy. He's kept incredibly relevant with all of the moving trends and has a tremendous following. So he's going to give us a lot of bandwidth in terms of the path in which we are going to continue our underwriting journey. But he can do so much more than that. He's really well known. So he'll be working with a lot of our stakeholders, whether it's investors, regulators, reinsurers and expect great things from him. And he's very additive to a very capable team that we already have. So we're anxious for him to join us later this year.

Right. And it's November.

Technically, it's December. I'll leave it at that.

Fair enough. That's good enough from my perspective.

I'm good at getting people to work before they're on the payroll.

When you talk about -- so AIG Next, another success in terms of like getting ahead of plan, both in terms of time and amount. You talked about not just cost savings, but also increasing internal efficiencies where you could process business faster. And some of this may get back to the agentic AI, but can you update us on where those efficiencies are playing out?

The original plan for AIG Next was actually mostly just cost takeout. We actually had said on some of our earnings calls that it's not going to be like AIG 200, where we had 10 different operational programs, and the baseline for AIG 200 was to improve the company. This was about getting rid of redundancy, getting a parent company, expense structure that fit with the size of the organization that we are. But we had some very terrific observations and insights. Actually, Mia Tarpey, who's here today, actually led AIG Next. And we saw halfway through that there's a lot of things we can do on the operational side that can improve the organization in addition to getting cost out or cost into the business. And that was connecting a lot of the functions, better end-to-end process, allowing opportunities for better insight on data and also what we're able to do on Gen AI today is based on the end-to-end connectivity of the organization. And a lot of that was done through AIG Next. So ended up being a continuation of AIG 200, which was terrific because we had originally just planned for cost takeout, but we were able to improve the company significantly during that period of time.

Okay. And in your introductory comments, you talked about agentic AI being expanded to other product lines. When you said Lexington, is that every line of business that Lexington writes?

It's -- yes, it will be predominantly -- it may not be every single line, but it's going to be the Property and Casualty, which is -- and Financial Lines, which will be the predominant portion of it.

What are the data points that you track to demonstrate that it's working as expected?

Well, we focused on -- and again, I'll try not to make the answer too long. But when we had embarked on this journey, it was all about how do you take in data structured and unstructured, how do you actually use data sources that you're not using today? How do you equip an underwriter with more underwriting criteria? And how do you reduce the cycle time substantially to be more efficient and to be able to look at submission flow and workflow to be able to drive growth? And then we didn't spend a lot of time on this at Investor Day, but we've been doing this. It's evidenced through what we did with Blackstone and Lloyd's and creating a special purpose vehicle for own reinsurance is how do you allocate capital and optimize the portfolio. So like how do you connect all of that together? Lexington became a very big opportunity for us because we showed that the submission count -- and it's still happening in E&S. Yes, there's been a rate environment in Property that's coming off. But quite frankly, it's been coming off in a lot of the property market admitted as well, but submissions are still flowing in through non-admitted and through Excess and Surplus Lines. The volume is very hard to handle when you start getting into hundreds of thousands of increased submissions for our own business and whether we were not efficient at the beginning, or where we are today. I don't spend a lot of time thinking about that. What I do know is we have 10x the amount of submissions as we once did, and you're not going to bring in 10x the amount of underwriters. So -- and then you look back to what's the bind to sort of submitted ratio, and it was 6%. Like so it's not as though it's 20%, and it's like, oh, you want to get it to 25% or 30%, that might feel hard. When the volumes start to increase, when the business started to get more demand, our buying ratios went to 2%. And so yes, maybe we're more selective underwriters, but 2% of what you're seeing seems largely inefficient. And we felt that getting that data and getting things more prioritized and having the underwriting cycle time decrease is going to allow us to grow significantly in that business. So that became the next evolution in terms of scale, which we should be able to have it done before year-end. And as we said, if we can just get back to our old bind ratio, it's north of $1 billion of growth, and I think that, that would be a low bar for us what I think we can do over time.

Okay. Fantastic. Are there bottlenecks in terms of rolling it out even more broadly to every AIG region, every AIG line?

I wouldn't say bottlenecks, but how much change do you want to go through in an organization? I think we'll be very aggressive in terms of the rollout in 2026. We have to realize is that even since March, the companies that I mentioned, I'd add in Palantir, Anthropic, other companies that are pivotal to our partnerships. They've accelerated their capabilities in the last 6 months, it's unbelievable. And how they can extract data, where they can extract it from, what we've seen is a lot more discussion around agents or individual large language models that can do specific tasks and how do you actually roll that out. So I'm very optimistic that the rollout does not need to be as linear and that you can do it in multiple parts of the business at once, and that's what we plan on doing in 2026. You do need -- look, we're not having it make decisions for underwriters, but you need the underwriters, if I would say not a bottleneck but a significant amount of heavy lift work is if an underwriter is looking at 100, 125 pieces of data, what are those? And what are the best sources that you can get in terms of property data or casualty data or financial data. And one of the, I think, significant advancements that the large language models have made is that you can now find out where the source is coming from. So in other words, you get financial statements that came from, a rating agency or it came from something that's credible versus something that may be scrapped that you don't know where it came from, that's where there was more hallucinations. So I think the data accuracy is better. I think the underwriting criteria is better. The issue -- it's not a problem; it's an issue to be aware of is that how much of these large language models become intuitive and how much further do they go in an organization than what you're asking them to do. If you start to implement a lot of different large language models at once, that ontology gets super complicated. And so we're working through that as well to make sure that we understand the rollout. Again, it's not making decisions, but making certain that the organization can catch-up to how we want to operate in the future.

Okay. Fantastic. When you look forward 5 years, this is maybe not a great question, can AI make underwriting decisions not today, in 5 years?

I think they will be able to, what type of decisions and what you want them doing is really the question. But like even today, a simple example would be looking at sprinkler systems, is it water or is it suck the oxygen out of the room or is it foam or knowing how to train large language models as to what's a better outcome on some of the subcomponents of underwriting, I think, will be very important and very helpful.

Okay. So to me, that sounds like maybe the prioritization that you talked about where the best submissions were at the top, that gets more refined.

Submissions at the top and information that could be more binary in terms of how you train an agent, I think can be very helpful to the underwriting process.

Okay. Fantastic. You also mentioned the rating agency upgrades. Practically speaking, there's nothing -- I don't see anything wrong with those. What difference will those make?

I mean for our business, probably nothing. I mean -- but from a perspective of this journey of bringing all of our stakeholders along, we had to get an underwriting portfolio that was one of the worst performers to get to a top performer. We had to get limit management. We had to develop operational excellence. We had to bring -- we were getting no ordinary dividends out of our subsidiaries because the regulatory environment was one where AIG just disappointed. So it just -- it was kind of a last like stakeholder that had not sort of voted for the change in the improvement that we made and to see that affirmation. And then going back and saying, like in 1990, I was a trainee at the Hartford, and there's no such thing as cell phones, voice mail. The world was different back then. So it's a long time. And so it was just more symbolic of we've come a long way. People are proud of it, and it was just another last sort of vote in terms of, yes, we believe in all the things you're doing, and we believe in the financial strength of the company. So it was a good moment for the company.

Yes. No, there's -- and in any way...

But no, in terms of trading, it won't matter.

Okay.

But we're going to keep telling everybody, we got upgrades since 1990. I liked it.

It's great line. One -- okay, so we're, I mean, heading to Monte Carlo soon, start talking about reinsurance. AIG buys a lot of property reinsurance. And we're in a presumably softening part of the property reinsurance cycle. So I know the timing of this question isn't great. But is AIG over reinsured? Are there ways of optimizing that, or do we take advantage of changing -- not changing, declining pricing to further benefit shareholders, benefit the company?

Over reinsured, under reinsured, it is a strategy and a philosophy. Each company has sort of a different view. If you're running a business in a global business that is running in the low 80s combined ratio with what would be technically considered a lot of reinsurance, I like that strategy. I'd like the predictability of volatility containment. And I don't like to be at the whim of what happens with the weather or major hurricane or wildfires. We have a risk tolerance and that's how we set up the entire company. When the market started to change a lot a couple of years ago, we stuck really hard with lower retentions because, again, it was within that volatility containment, and you either load the cost in for the reinsurance or you don't. And so my own view is that and I try to break this down because I get a lot of questions around this in the second quarter, which was how do you actually break out the components of Property? And I have this philosophical argument, I'll stay away from it, but like why would reinsurance cost be that much more expensive than retaining it yourself? I mean you could take -- there could be a couple of philosophies. One is, well, I'm not going to take a single year view, I'll take a 5-year view. Okay, well, that's been wrong each of the last 5 years to the -- like it's been higher than what would have been predicted. So that doesn't seem like the right way to play it. There's a cost of capital issue. Somebody has too much of a margin, like whatever, that's in the eye of the beholder. But my view is this should be like fully loaded into the cost. And so if you think about a market that we're going into, what I try to break out is that a reduction in reinsurance costs benefits us because that is embedded into our product. If you just funded it net, it's an AAL, you don't reduce your AAL because the reinsurance market is going down, that's your expected losses and it has to stay the same. So if the rates are coming down, and you don't have anything to offset that like reinsurance, like that's not coming down commensurate, then you have to have margin compression and a meaningful amount of margin compression because you still have the same CAT net, you still have your same risk net. And if you're getting total premium coming off the top, well, that has to still fund that. So it goes really into your attritionals. If our sort of cost of goods sold, I call it, is that of reinsurance is coming down at or more than the original pricing, well, then really, I'm only focusing on the attritional. Same thing on the risk is that then it's much more manageable because that CAT cost, which would be self-funded with AAL otherwise is coming down, and so we get the same benefit for lower cost.

Right.

And so I think like in a market, look, I don't root for an active CAT season or benign one, we manage through the cycle. But if the reinsurance costs are going to go down that benefits us and doesn't put as much pressure on pricing as it would if you kept substantially large nets.

Right. Okay. No, that makes a lot of sense. This is only tenuously related to that concept. But I've had this thesis for a while that when we look at lines of business where pricing has gone soft over the last few years, so D&O, Cyber, most recently Property, there's been a suddenness to those decreases compared to past cycles. And I know it's a mistake to assume that cycles would be identical, they're not. But how -- if that premise is true that cycles manifest themselves faster, how does that impact your ability to plan for the consolidated AIG?

You have to look at each line of business on its own merits. And again, the topic right now is Property. And you have to look at the cumulative effects of that. And so you have to have a plan there. I mean, in other words, with Property like the cumulative rate increases have been north of 100%, combined ratios have been excellent. You can look at -- for us, we have so many different points of entry into the Property market, whether it's through admitted in the United States or non-admitted through United States or through Europe or through Asia. We could hub in Singapore. We see it through Lloyd's. So there's so many different places and you want to take a look at -- you don't want to manage it as an index, but there's opportunities in different parts of the world at different times. And so the planning is really what's going to happen if the rate environment is going to be down like we've seen in Property. Well, okay, what kind of margin do you have? What type of combined ratios do you have? And can we manage through it? I think you also have to take a look at where you are in the market, are you lead in pricing? Are you setting the terms or are you just capacity following outcomes, that becomes a different outcome. And then you have to have a point of view as to how long you can sustain that. So I think, look, is the Property market more aggressive in terms of some of the pricing coming off than we would have thought at the beginning of the year? Yes, it is. Dramatically? No, but a little bit more. And so we will see how it plays out through the rest of the year, but we have perhaps a different plan in '26 if the rate environment continues. In Financial Lines, we did the same thing, which is we reduced our writings when they started to become cumulative rate increases in commoditized layers, high excess, mid-excess, and we just started to nonrenew business. And there will always be pockets of where there's opportunities for growth. We're seeing that in Casualty now, non-admitted as well as admitted. And I think some of the specialty classes have had some pressure on pricing. But again, the same dynamics as our Property book, which has had really strong combined ratios. And then you have to be prepared to grow when there's opportunities that exist in the marketplace from either dislocations or opportunities in different parts of the world and geography. So that's part of the planning process. And I think we're in a really good place in terms of understanding what the economics are for each of the lines of business that we trade in.

Right. Okay. Fantastic. I do want to look around the room if there are questions. I want to make sure that you're getting the information that you need out of the session. So please raise your hand if you do have a question, and we'll get the mic to you. In the absence of that, so another large commercial CEO has suggested...

Who?

Evan.

I need to know.

Has suggested that bigger companies are positioned to grow faster. And the backdrop to this question is we've seen huge market share concentration in personal lines among the biggest companies. Some of whom became bigger because of skill, but that dynamic has played out. We haven't really seen it in Commercial Lines. And the suggestion is that because of a number of factors, we should see that basically better growth at bigger companies. Are you seeing that? Do you expect that? How are you addressing that potentiality?

I think over time; I think that will happen. I don't know that we're seeing it right now because, again, it's always what you see in this marketplace where either some MGAs or capacity that has come in that needs premium, needs volume until something happens that tempers a little bit of the growth. But I think over time, scale is going to matter significantly in some ways more than it ever has whether it's balance sheets, ability to have diversified growth, quality and expertise in underwriting to be able to scale a business. But -- and again, I'm trying not to make every answer about Gen AI, but I think in 3 years businesses that have invested, the pendulum is going to push those much further faster and those who have not been able to ingest and be able to accelerate cycle time will be, I think, very much disadvantaged and you have to have scale to make those investments, you have to have expertise in order to be able to operationalize it. And I think larger companies with scale are going to benefit from all of that significantly.

Okay. When you look at your competitor base, and I'm asking this because I don't get the sense that you compete a lot with smaller regionals writing small standard commercial policies. How do you see your advantages playing out there?

The footprint for every large insurance companies is a bit different. And so it's in pockets. I mean, obviously, in London, it's a big specialty. We're one of the world's largest specialty underwriters. And so like we're competing with Lloyd's, we're competing with other large insurance companies that may have some expertise in specialty classes. But I think we are the most sort of prominent amongst organizations in terms of size, scale, expertise, I believe that. In the U.S., it's a little bit more in terms of how we scale. We've done an amazing job with Lexington in terms of its relevance to the marketplace and size and scale. So we had some formidable competitors there. But I do think that there's a way that you bifurcate that with size, expertise and ability to grow versus more commoditized. It's not all large players, but there are specialty insurers that compete in that area that really differentiate themselves and Lexington is one of them. I mentioned Property and the many ways in which we can access Property from so many different parts of the world. We have a very different business in Asia, I mean, we're the world's largest nondomestic insurance company in Japan where we haven't fully seen all of the benefits and I think potential of that business with some of our personal insurance, particularly A&H and some of the digital investments we're making there to fuel growth, that will be on the come. And India, I know it's not a consolidated business, but it's a big business and one that we think has one of the biggest growth potentials in terms of value, size, relevance and also when it gets out of its domestic sort of capacity, meaning it can actually do global business through the network of AIG. That could be very advantageous for us as well. So like we compete with a lot of different companies in a lot of different geographies. But really, I believe, bring expertise and value differentiation in the markets that we trade.

Okay. That's very helpful. You've talked about in the context of M&A, certainly, an openness to M&A with incredibly strict hurdles for financial, cultural and strategic upside. So the financial and cultural, I think, would very much depend on a potential acquisition. Can you talk about where there could be strategic advantages to AIG from M&A?

Well, I think we've demonstrated through a lot of the work that we've done in restructuring the company that we would know how to embed an organization, deal with the operational complexities and get that synergies that needed to be realized very quickly, but also embed an organization to be able to make certain that they are more benefited from being part of AIG than not. And that's dependent on a lot of different things. I talk about like there may be segments we're not in, in SME, geographies that are additive to our global network, more scale in businesses where we already have size. A&H would be a great example, if we can find something that we thought was high quality that we could bolt on to a big business that's profitable that we could accelerate, that would be something that's very additive. But I think that there's a lot of -- I keep mentioning a lot of strategic and financial flexibility. I want to be patient, which doesn't mean wait forever, but I want to make sure that whatever we do acquire is strategic, is ROE accretive, but also is a better fit for them and for us being together. And I think depending on the market and depending on the challenges of growth, that may come earlier or later depending on what happens with market conditions.

Okay. So if I can just respond -- not respond, but a follow-up to that. Should we think about that strategic benefit of making AIG better or AIG making other businesses better consolidated with it?

Both.

Okay.

That would be our hope.

All right. One issue that's come up a lot, and I like this issue because it gives me the opportunity to talk about something besides pricing. And on the sell side, that's 95% of my conversations is speed of processing for smaller applications. And I'm asking this question really in the context of Western World. How does it compare? And what are its prospects for developing further competitive advantages on speed and accuracy of response to brokers?

So Western World, for those who don't know, is part of our Excess and Surplus Lines business. It's really services the small end of commercial. And it's done remarkably well, whether it's been new products, the speed to bind is very impressive and developing a risk appetite, but also an ease of doing business in terms of ingesting data. We know what our risk appetite is in those businesses. And it's actually penetrating into a market that we really haven't had much presence. So I think the framework of that business is great. The strategy is evolving, and our execution has been very good. That's how we -- sometimes the submission count there and the flow and growth has been tremendous. But I think in this world that we're in and that we're going to be competing with in the future, it's only going to get better.

Okay. Fantastic. And that's, again, you said you don't want to use AI as the answer to everything, but it's going to have broad relevance.

Yes, that AI can help decision-making there a little bit more just because it's much more small commercial volatility around decision-making is not going to be great. And so hopefully we'll be able to advance it. But even today with the technology we have, it does serve our distribution very well because they can find things in minutes, not -- and so like getting to seconds doesn't really matter. But making sure that we can continue to enhance our ability to expand risk appetite when we want to or focus on specific industries and be able to target those submissions better.

Okay. Fantastic. I do want to scan the room again just to see if there are questions. I don't want to make sure I'm not overlooking them. Quick question on underwriting and claims talent and maybe other segments that I'm not thinking about also. But whether we're talking about the relentless formation of MGAs, whether we're talking about brokers looking for talent, whether we're talking about underwriters, there does seem to be an elevated battle for talent right now. And I was hoping you could talk about where AIG participates. What do you do to attract, to retain? And from a different perspective to develop talent, so that you have superior underwriting and that persists or claims handling, I always....

Yes. I mean, we don't have the same dynamic as brokers do where teams get lifted out or there's significant war for talent in terms of groups of people because I think underwriting is different in that maybe business travels with brokers, but it doesn't travel with underwriters. So I think you have a real opportunity to develop the culture that you want an underwriting organization. We've done a remarkable job with just terrific leadership that we have with an AIG, great practitioners. We had to add a lot of underwriters in '18, '19, '20. We have a tremendous training program and so getting out of university into our analyst program and then getting individuals working through the overall company has been tremendous. And so we supplement what we bring in from the outside with talent that is developed within AIG. And so we're always looking for ways in which we can train and advance and enhance the skill sets that we have within the organization. And that means actually training to where the business is going, not where it's been.

Right.

But yes, we're always like -- we're always going to be in the war for talent, retaining, attracting. Look at AIG, I don't fully understand it to be honest, is like it gets the headlines for everything. I mean there's been people that have left and like, who is that? Like I mean -- and it just gets headlines. And so our attrition has been very low. Our underwriters are very attractive in the industry just based on the experience that they've gotten in a sort of once-of-a-career opportunity to work through a turnaround and reunderwrite portfolios, be part of that. But I've been incredibly pleased with the quality of what we have, the culture we've developed and our ability to attract talent is tremendous.

Okay. And that attraction flow, if that's the right word, how does that compare not to 5 years ago when you really needed to rebuild a bench, but like 2 or 3 years ago?

We have been very selective because it's really important now that we have -- there's different skills needed to reunderwrite a portfolio than to grow a portfolio or to be in an underwriting group that's not going through that level of rapid change. We're always looking to fill in technical expertise depending on like in London and Lloyd's, there's different classes of areas where we want to compete that we're adding and supplementing talent. But generally speaking, it's very specific as to knowing positions that we want to develop. I mean, look, AI is a big one right now. It is bringing in people from different industries. We just brought in somebody to be the Chief Digital Officer, Scott Hallworth, who has got a lot of banking experience in terms of being a Chief Digital Officer, but he was the Chief Actuary at Travelers. So coming in with a totally different lens of financial services, but being grounded in data and how our business operates. That's been a tremendous and will be addition to the organization. And so finding different capabilities that complement where we are today is a big part of what we're doing.

Okay. And no $200 million signing bonuses?

No.

Okay. Fair enough. On the MGA side, you guys do have a little bit of delegated underwriting authority. How do you manage that? I know -- I think there's investor concern that's probably a little exaggerated right now, but it does require a slightly different administrative skill. And I was hoping you could talk about the AIG approach.

Yes, we don't do a ton of delegated authority. That is not a priority. You're right, it's a different requirement in terms of skill set. If a -- now Glatfelter was a company of a great example where they were writing on behalf of another insurance company at the time, but like they had their own capital. They consider themselves an underwriting company and their track record proved that. But they also had a distribution in terms of voluntary fireman that was, and you couldn't really penetrate it. So do you want that expertise within the organization? So that was an example of an acquisition. I think we use the same criteria for where we would do MGAs or MGUs, companies with experience, track record, skin in the game, alignment in terms of underwriting outcomes, not looking to drive top line growth at the expense of underwriting profitability, and you'd rather do it with fewer than sort of spreading it out and watching it very carefully, to be honest, because that can -- and it's happened to AIG and many other companies in the past where that can go pear-shaped pretty quickly in a market where you don't have a real accurate view of what the accident year loss ratios are, the growth and the development of the book of business. So I think there's been a lot of delegated authority. In my view, it's going to end in tiers over time for some of the organizations, but that, unfortunately, in our business takes a little bit of time.

Right. I mean that's -- I would argue that, that's true even on the traditional side with maybe slightly misaligned incentives.

Yes.

I want to spend a little bit of time focusing on reserves, really from 2 perspectives. We did have some older accident year strengthening in the second quarter. And I want to get a sense is your comfort for other legacy mass torts for lack of a better word. And then one point that I think got under -- that's been underappreciated is the fact that when we look at your quarterly results, and this is true, even in the second quarter, there's favorable development and its favorable development besides ADC amortization or ADC gain amortization. So I want to talk about those 2. I know they're somewhat contradictory, but they are 2 different realms.

We're very focused on making sure that we continue to strengthen the balance sheet. We take a very conservative view on reserves. I wouldn't read into the back years at all, nothing emerged. There was nothing that said we have to do something on mass torts. We decided to do it and have built some conservatism into that. We've been very focused, and I've said this on the first quarter call where we've built in some margin in some of our accident year loss picks, not from emergence, not from concerns on the accident year loss ratios, but just being conservative in terms of our view of the current environment. There was the ADC amortization, a little bit of release. But again, I think that was all very orderly. We've been doing it quarter-after-quarter. And Keith is spearheading a little bit more of getting away from these DVRs and looking at reserves in a more thorough way every quarter. Yes, there'll be deep dives in certain quarters on lines of business, but just getting a much more balanced view throughout the year instead of waiting for the third quarter or waiting for certain quarters for certain lines of business. We've been working hard on that since the end of last year. And I think it's just patience and making certain that we are watching emergence of each of the accident years. We study Schedule P probably, and I don't want to compare it to other companies, but what has happened to the industry of whether it's other liability or other areas where there's long tail sort of claims activity and watching that emergence and watching our own. I mean, you know our absolute loss ratios are so much higher because they needed to be based on some of the underwriting that happened in the past. But we are looking at a variety of different metrics to make sure that we're being conservative and observant of what's happening in the industry.

Okay. Fantastic. And again, I want to look around just to make sure that I'm not overlooking anything. One, there's some new -- occasional news articles on a shift of longer-tail casualty policies to more of a claims-made basis because it's not that much in the short term that the industry can do about social inflation, but you can reduce the tail size, which would limit the compounding risk associated with that. How tenable is that? How much of AIG's book could and should be written on a claims-made basis that hasn't been so far?

Well, I mean you pointed out the benefits of doing claims made. I don't know that in the segments that we trade and that the clients are going to be as receptive to claims made as we may like. And certainly, when you're underwriting an occurrence form that has like a really long tail that making sure that you have the appropriate pricing for what could be loss cost inflation that may or may not be contemplated today is, of course, a concern with the class. What you do there is you focus on terms and conditions, you focus on attachment points, you focus on limit deployment. We have -- again not every answer is reinsurance, but we're only willing to take a certain amount of net depending on the gross limit deployed, which is much smaller than what the historical portfolio would look like. So some of that emergence became from maybe it wasn't great underwriting, but it also was significant net limit retention. That gets you in trouble when you have sort of vertical exposure and hadn't fully contemplated. So I do think that there are going to be pockets of industry classes that we will push pretty hard for claims made. I don't think it's going to be a predominant trend within the industry. Some companies are trying to put stuff together. I don't know what the take-up is. I don't think it's significant. But you have to really manage that through underwriting with the terms and conditions and focusing on how much limit you're willing to deploy. And are you following someone we lead, we set the terms that's really important.

Okay.

And I think what our reinsurance [indiscernible] too is not that I sat up here whatever, 5 years ago, you're 2 to 3 years into it, you've got to play it out. But like we're like in the year 8 or 9 in terms of taking low nets within Casualty. We started sort of in '18, '19, but we also did the unearned portion of the book. And so we just don't have big attachment -- sorry, big retentions. We have lower attachment points and so as each year goes on, you're going to get more and more confidence in terms of how we think about ultimates, certainly on the back years, but even on some of the more recent years.

Okay. No, that's very helpful.

I mean I'm speaking to an actuary, so it's dangerous but that's my view.

Well, [ it's not supposed to be. ] But yes, no, we're trying to get the sense. And I think it's been pointed out that there are a number of periodicals to focus on the insurance industry. And sometimes it's tricky to filter out where they have these stories, is this a real shift? Or is this a one-off? So I think that insight is very, very helpful.

One like, you got to do the work. When you look at companies, again, you go through Schedule P and see whose loss ratios held up from their original -- even go out like whatever, go through year 2015, start year 2, say everybody got year 1 wrong, '15, '16, like you can see companies that traditionally get it wrong. And you see companies that traditionally get it right, like -- and so I think that's important in terms of experience, looking at triangles, understanding, again, limit deployment and potential ultimates, and it's a complicated line.

And not to jump too far afield from what you're saying, but one of these analyses I've been working on and this is not going to come as a surprise to anyone is that the best predictor of whether a company will have adverse development on accident year x is if they did the prior year. So for me, it all fits into why I don't trust book value as a valuation metric. But your point is something that I'm more than happy to say I've seen.

Yes. Book value is as good as the quality of the reserves, right? Like -- and so if you have a view of what ultimate looks like, then you'd have a good view of book value.

Okay. Which may not be what companies are saying.

Correct.

Where are we on global personal in terms of the shift to an MGA model, the shift to adequate pricing? And what are the long-term prospects?

So -- are we talking about high net worth or like just the general global personal?

So I was going to ask both. Starting off on that overall.

Yes, I would say, if you start with the sort of bigger one, which is global personal, I would say, very good businesses, we've divested businesses that we didn't think were core. We need to grow A&H more. It's profitable, but it has not grown to the level it needs to grow. We have to have a little bit more profitability improvement in the personal, defined as personal auto and homeowners that we do in international, not necessarily in the U.S. But the big bet is going to be, can we get the high net worth in PCS to profitability, and we will. It's been slower than I would have liked just in terms of some of the headwinds that have existed changing a model that -- I still believe the way we've structured this will work out in the long-term, but we need other carriers to be working with us in the MGA to be able to underwrite the business alongside AIG. That was the real driver of the quota share, which we now see negative growth in 2025, but that is bringing in strategic partners that are going to learn more about the business that will either provide paper or will come off business. I mean we don't need the quota share for reinsurance ceded or -- look, it's a benefit in the short run on profitability because the ceding commission is higher than the expenses. We were able to lower the MGA expenses and get an enhanced ceding commission. Okay, so what? That's a 400 or 500 basis point benefit. That's not going to be a long-term business strategy. But I do think that the opportunity is going to be seen. We're growing in E&S more than we are in admitted, of course, but not as much as we need to. We have been off to a slower start than I would have liked. And in the rest of '25 and as we get into '26, we're going to have to accelerate that growth or think about going more into an open market environment where we can see more submission activity. So it's working. The profitability improvement is there. It's happening this year, but the absolute performance of the high-net-worth business is not where we want to be, but we're heading in the right direction.

Okay. Once the profitability has been addressed, what's the global market potential of high-net-worth homeowners, high net worth personal lines?

We won't go outside the U.S. I wouldn't expect to see that strategically in the next 3 years. I think the opportunity is I'm going to give you a number and then you can hold me to it. But I think it's substantial. I mean I still believe that going into the sort of E&S environment, problem with admitted is if you're in the ultra-high net worth or high net worth you have to put out large policy limits. The accumulation in CAT exposure gets pretty heavy, very fast. If you look at areas that have always been problematic, there's been more TIV in those areas. Think of Florida, Texas, California, all have different dynamics. I still think the Northeast is the most exposed because of how weak the PMLs are and the TIVs are massive. And also if a Category 3 ever found its way up this way, flooding construction type, we're not ready for it. So look, there could be things that happen that accelerate it. But I think it's -- the opportunity over a longer period of time is well within the hundreds of millions, maybe even greater than that.

Okay.

Because I just think that there's not -- the void isn't getting filled. You hear antidotes, but they're real, which is like we've seen -- look, I'm not going to call it a trend yet, but let me get through the full year. But even with what was happening with the wildfires is looking at sort of claims development because you want to be observant on inflation, supply issues. Is there more demand or replacement costs? Like what does it look like? I mean a lot more of our clients took the check and aren't rebuilding. And so like forget it like I'm not doing this because I can't get insurance and I'm not going to be in a place where there's going to be more wildfires not protected. And so that's not really solving a problem. I think -- and again, we haven't seen something significant in Florida. I think there's more limit to be purchased and there's more demand if we are there to offer the product.

Okay. Fantastic. I had one -- and again, if there are questions in the room, please let me know. But I want to focus on investment income, you talked in the Investor Day and again, not abruptly but taking more risk. Where are we on the process? How should we think about the time line?

I didn't say, I think Keith said that. Hi, Keith. What we are alluding to is that when we were fixing the underwriting side, we were also very conscientious of being conservative on the investment side. And so we went from a -- when we were part of, AIG was life, retirement and P&C, the general insurance, we had $325 billion of AUM, and then we began to say, okay, is an outsourcing model more advantageous? Well, knowing we're splitting up, I'm not going to spend time talking about Corebridge, but had Blackstone come in, Nippon come in, we outsourced a lot of the fixed income to BlackRock, both Corebridge and AIG. But when you're $100 billion or thereabouts of AUM, developing a more outsourced model, not for the investment strategy, not for the asset allocation, but for the execution we felt was an appropriate model. And that's what we did and I think it's worked out really well. Now along the way, we started to get out of hedge funds. We started to get out of some of the alternatives. And Keith has been working with the head of our investments, Pat Boisvert in terms of repositioning the alts or the reinvestment that we've had in fixed income. And so I think there is an opportunity for us because we were so conservative in terms of the overall percentage of asset allocation alternatives that there's ways in which we can get perhaps not chasing it, but just executing better with a more balanced strategy where we can get a little bit more return 3.8 years, not the longest duration, not like life, but it's also not that complex.

Okay. And then last chance to -- as I look around the room. And if not, I am going to thank Peter. Please join me in thanking Peter for a very informative session.