Defining The Broker's Role with AI Near Its Soul: A Discussion with Oxbow Partners

Speaker 1:
Welcome to TD Cowen Insights, a space that brings leading thinkers together to share insights and ideas shaping the world around us. Join us as we converse with the top minds who are influencing our global sectors.
Andrew Klingerman:
Hi, I'm Andrew Klingerman, the insurance analyst at TD Cowen, and today it's my pleasure to introduce Manmeet Bawa, partner at Oxbow Partners. Just so you know what they do, Oxbow Partners is a specialist management consultant exclusively serving the insurance industry. And I'm excited to be discussing today with Manmeet AI in insurance brokerage. Manmeet, welcome.
Manmeet Singh Bawa:
Thanks, Andrew. I'm excited to be here.
Andrew Klingerman:
So I'm going to kick off the questions, and Manmeet will do the heavy lifting. So if AI can automate much of the analysis, research, and administration that brokers perform today, where will brokers create value in five years time?
Manmeet Singh Bawa:
Andrew, that's a great question. My view is that AI does not eliminate brokers; it does eliminate large portions of the brokerage work. Historically, brokers have played three roles. They've been information intermediaries, they've been market intermediaries, and they've played the role of a trusted advisor. The first role is under significant pressure in my mind, where gathering information, preparing submissions, comparing quotes, reviewing policy wording, et cetera, are all increasingly tasks that AI can perform better, faster, and often more accurately. In my mind, the value is increasingly shifting to areas where judgment, advocacy, and influence will matter way more.
So in five years from now, I think the highest value brokers will feel more like strategic risk advisors where they might be helping clients understand emerging risks, whether it's cyber, climate, AI-related exposures. They will also start being capital advisors where you're helping clients navigate increasingly complex insurance and risk transfer markets that start to merge. They become better claim advocates such that when a large loss occurs and the client really needs the comfort of that trusted advisor fighting for their best outcome. And that's a very deeply human role.
I'd say the last area where I see brokers increasingly focus would be orchestrating relationships. By that, what I mean is bringing an ability to access markets, influence outcomes, and negotiate solutions that are bringing together the right stakeholders. An example that I might take for the benefit of your listeners would be a large manufacturing client looking to renew a global program. Today a broker might spend weeks gathering exposure information, building a submission, comparing codes or whatnot. But in the future, AI allows them to do this work in a matter of minutes. But where a broker's value becomes much more about helping the client answer how much risk should we retain, or what market should we participate in, or what are the right trade-offs, or what are the trade-offs that exist between premium and coverage, or how should we structure our overall risk financing strategy? I see that as just a fundamentally higher value conversation.
Andrew Klingerman:
Really interesting, Manmeet. So that brings me to: which parts of the broker value chain are the most vulnerable to AI disruption, and which are most likely to become more important because of it?
Manmeet Singh Bawa:
So carrying on, Andrew, from what I was describing, I believe the most vulnerable functions are the ones which are largely related to information processing. So whether that is a submission building or a market matching or providing a comparison of coverages or renewals, a lot of these activities in my mind are just going to be done more efficiently by AI. I think the pivot, again, is going to be towards things that are complicated and inherently more human. So whether those are negotiations, whether that is the claims advocacy that I was referring to, whether that is now morphing into a risk consultant rather than just a broker, I feel like those are aspects where you see the unique differentiation of value of the broker emerge far more than the traditional administrative stuff.
Again, let me take an example. Think about cyber insurance. AI can very easily analyze controls, industry benchmarks, claim trends, differences in coverages, but whether a client needs an extra $50 million of limit is ultimately a strategic business decision that is going to heavily rely on judgment, appetite, and board-level considerations, which is inherently human.
Andrew Klingerman:
And then that brings me to: will AI strengthen the position of large brokers with access to vast amounts of data or will it create opportunities for smaller specialist brokers to compete more effectively?
Manmeet Singh Bawa:
Andrew, that's an excellent question, and it's a very strategic question in my mind because the answer is both. It depends on what the broker's unique value proposition and competitive advantage is. The large brokers will win because scale tremendously matters in AI. The fact that these brokers have decades of placement data, claims data, carrier interaction history, pricing information, industry benchmarks, that's a significant advantage. A broker with millions of such records can generate insights that smaller boutique firms simply cannot replicate. It's very similar to why Google wins with the access to data that they have.
However, on the flip side, there is going to be a space for specialists as well, typically the boutiques. So while they could never afford the large analytics teams, the armies of data scientists, a dedicated research function, sophisticated technology, AI has dramatically reduced the barrier for those. Now, a 15- to 20-person specialist brokerage can actually access capabilities that previously required much larger number of employees. So think of a niche broker that's focused exclusively on, let's say, renewable energy risks. Historically, a global broker would've been your default choice because they would've just overwhelmed this niche broker through scale. With AI, your research has become cheaper, your analytics has become more accessible, your marketing has become more efficient, and knowledge management has improved dramatically. So the specialists can spend a larger amount of time advising clients rather than managing administration.
So if I was to be a betting person, I believe the biggest winners would be the large data-rich global brokers, but also the highly specialized niche brokers. I think the challenge will be for the ones that are kind of in the middle where their unique selling point or value proposition is largely unclear.
Andrew Klingerman:
That makes a lot of sense. And that brings me to the last question. As AI becomes capable of recommending coverage, pricing risk, and producing client insights, how should brokers balance automation with human judgment and accountability?
Manmeet Singh Bawa:
Yeah, it's a very interesting question, and I think this is one that several industries are grappling with. I have a very simple principle to think about it. AI should advise while humans should decide. So that fundamental human in the loop with an augmented AI resonates very well with me, at least in the insurance context. It continues to be a business of trust, as we all know. But how do you make your advisors more productive is where AI could truly shine.
The mistake I see many industries make is that because AI can make a recommendation, that humans should stop exercising judgment, i.e., do we choose to become lazy just because we have a tool at our disposal? I think insurance is fundamentally different. We are often dealing in scenarios where either the information is incomplete or information is just tough to gather or uncertain. Trade-offs are often not very clear, and so you're operating with a lot more judgment, and there are human consequences at the end of this. What that means is, and you and I can reflect on our insurance buying journeys, we really do hold our brokers accountable. If an AI recommends a structure that proves to provide inadequate coverage following a loss, it's hard to sue the algorithm.
So the operating model in the future should largely be thinking about AI as a means to analyze information, identify options, highlight risks, generate options and recommendations, and create that healthy challenge to assumptions. But you need the humans to validate those outputs, to apply judgment in areas where that model does not have adequate input, drive much better explainability to decisions, and accept accountability. I think that is what's going to separate the winners and the ones who are kind of in the quasi AI race.
Again, I always think about this from real-world examples. Think about an AI system recommending a client to reduce their D&O limits by 25% because that's what the benchmarks tell the model to do so. Now, that recommendation could be statistically correct, but may not be factoring in scenarios specific to the client like, are they taking up an acquisition, or is a shareholder dispute emerging, or does a board have a lower risk appetite than its peers? These contextual factors matter. A good broker will combine this AI insight with human understanding.
To summarize, the larger challenge or the critical issue is not technology in this case, it is governance. What I would encourage your listeners and clients to consider is, if you think about becoming the leading brokerage of tomorrow, you really need to focus on three key questions. Be very precise about where AI can act autonomously, be clear about where human approval is required and where it is essential, and then how do you attach accountability when things go wrong? Firms who are explicit in their answers, not just internally, but also to their clients, are the ones that are going to build much better trustworthy relationships. I think, again, I'll go back to that adage of insurance is a business of trust, and how brokers end up using AI to enhance their trust profile with their clients will eventually determine their success in market.
Andrew Klingerman:
Manmeet, these were truly outstanding insights, and I've learned quite a bit here. So I want to thank you for the time with us.
Manmeet Singh Bawa:
Thank you, Andrew. It was wonderful to be with you. I certainly think AI is going to be a big differentiator in how our industry evolves, and brokers are a critical part to it. So I'm very excited to see what the next generation of broker and brokerages look like.
Andrew Klingerman:
Absolutely. Until next time, Manmeet.
Speaker 1:
Thanks for joining us. Stay tuned for the next episode of TD Cowen Insights.