Podcast Recap with Marcus Ryu, Battery Ventures

25 Insights About Insurance Technology

1. At its very core, what is the goal of insurance technology?

Marcus: Terrible things happen in the world. We don’t think that they will happen to us individually, but in the aggregate, they’re certainties. It’s a question as a society, do you make the people who are unlucky suffer those losses, or do you as a civilization syndicate that risk and protect people from disaster. I think that’s a noble mission.

2. How is this period of technological innovation different from what came before?

Marcus: We all have a sense that in our consumer lives, in our professional lives, that history is accelerating, that the pace of technology change is at a breakneck pace, and that all of us are revisiting fundamental assumptions about what our work is and what the boundaries of tech are.

3. And how is this period of technological innovation similar to what has come before?

Marcus:  When Guidewire was founded in 2001, it was a hard market, but most of the time during my tenure from roughly 2004 to the late 2010s, there was a soft market, and the industry worried about its return on equity and its core profitability… One of the main value propositions that we would put front and center in saying why you need a new policy administration system was to say no one’s getting any younger here. The people who understand your core systems, the underwriters, the expertise, all the tribal wisdom, that’s a scare resource. It’s harder and harder to hire young professionals into this role, and you need a new technology foundation to not be completely out of position. We were saying that 20-some years ago. I would say that’s all utterly true today. The questions are perhaps more profound, but they’re not different in my experience.

4. How is the promise of what AI can do in terms of routinization different from what happened with enterprise software?

Marcus: I think it’s the next frontier technology that is expanding a decades-long motion to say how do we have machines automate things that otherwise humans would have to do. The humble claim file was already an advancement in just being able to aggregate information in a structured fashion. The batch processes that were instantiated in cobalt code were ways to routinize tasks. The insurance industry is hundreds of years old. AI has expanded the frontier of what can be routinized.

Insights into the Potential of Insurance Technology

5. What’s an example of something AI can routinize?

Marcus: The submission process. In commercial lines in particular, you have large corpuses of very unstructured information that has to find its way into structured form. That transition used to be mediated by the gray matter of a clerical worker or maybe a junior underwriter. It was outside the scope of what a computer could do reliably. Now AI has rendered that a completely solvable task. I think you’ll see that across almost every single process that an insurer does.

6. How is AI changing policyholder interactions?

Marcus: I think this is being completely transformed. I don’t think this is an incremental change. There’s been a step function increase.

The interaction between policyholder and insurer used to be something that had to be mediated by a human being. There was no one else that could translate the intentions, questions, and anxieties of an insured to what they wanted their insurer to do other than a human being in a contact center. We now know that translation can be effectuated just as, and in some ways more effective and with equal empathy, with instant pickup, with high resolution, with faster resolution

Insights into Disruption in the Insurance Sector

7. The insurance industry encompasses many roles, from agents and brokers to MGS and carriers. Who stands to benefit from AI development?

Marcus: In my entire time serving the insurance industry, there has bene an enormous energy and hope that’s put into how we can accelerate the movement of risk through the value chain… How do we facilitate the information exchange and the risk transfer and accelerate the throughput. And everyone might be better off in this context. There might be winners and losers, but overall, everyone would be better off.

8. Where will AI cause the most disruption within the insurance industry?

Marcus: AI has disruptive impact on broadly speaking two kinds of vectors. One is the potential to assimilate information and make decisions on the basis of that information in a much more systematic fashion, maybe to augment a knowledge worker, or maybe in some cases to do it straight through.

The other opportunity it has it to routinize otherwise manual work. Of course machines and computers have been doing this for many years, but now there’s a possibility of expanding the perimeter of what you can routinize and automate with much great flexibility, fidelity, and reliability.

Insights into the Impact of Technology on Insurance Jobs

9. Will AI replace humans?

Amrish: In the AI world, we talk about a spectrum. In the AI and insurance world, people often fall into this trope of AI will replace humans. So, it’s binary. It’s humans right now, and AI in the future. And the reality is what we’ve seen with the AI world is that’s not the truth.

10. What does technology change mean for professionals starting insurance careers?

Marcus: The industry is incredibly resilient. It serves a fundamental societal and financial services need. It’s not going anywhere. However, I would say the nature of insurance work, what it means to be an insurance knowledge worker, to be a professional whether you’re the underwriter, an actuary or a claim specialist, is a dynamic question right now. Those questions of the meaning for these professions and how to be valuable both within your organization and to your insured and business partners has never been more dynamic.

11. In the near future, will there be billion-dollar companies with just one employee or a small handful of human employees?

Marcus: I struggle to see that happening. I think maybe that’s just my own hubris from having gone through the company building journey. There’s just too much to be done. But I think we are going to see very dramatic differences, unequally applied depending on the market segment and the function.

12. Will AI create new jobs?

Amrish: We’re seeing an opportunity for new kinds of jobs to emerge even in the insurance industry. For example, we have this thing called the AI supervisor. We have people who today are in the front lines answering customer requests and resolving these requests. They now become not just a manager of their individual job but also a manager of a group of AI agents that do the job for them.

13. Why will humans still be needed once AI advances?

Amrish: You will always need human oversight. It is a highly human industry. Insurance is a human construct. It’s deeply emotional.

Insights into the Future of Insurance Technology

14. The insurance industry is adopting machine learning and predictive analytics. Will the insurance industry ever be able to predict risk accurately?

Marcus: It’s terrible to think of it in terms of one unitary convicted prediction. That’s a mistake. And the best forecasters don’t do that. What they have is a kind of Bayesian probability machine where they say there’s a mosaic of things that are happening. They’re conditional probabilities between them. We’re constantly getting new information, adjusting our priors and trying to make a prediction empirically from that basis. That’s how the insurance industry workers at its best.

15. Will policyholders embrace AI?

Marcus: Human beings will gladly talk to a voice agent and have a satisfying conversation, and a good number of them will do it in preference to talking to a human because they get an instant pickup, 24/7/365, and they get faster resolution of their task. To me, that’s a binary change. There’s a moment when we cross through the uncanny valley and now this is actually laying claim to being a superior experience.

16. Where will AI be in 20 years?

Amrish: It’s potentially conceivable that in 20 years, our understanding of what’s possible would totally change.

The promise of AI is real. The opportunities are real. But it’s a little bit of an unknown what really happens next.

Insights into the Success and Failures of AI Projects

17. How can organizations set their AI projects up for success?

Amrish: It isn’t just about deploying a technology. It’s about changing the mindset of the organization and applying real thought into change management and organizational transformation. It’s changing the processes.

18. Why do some AI projects fail? Is it because AI advancement in insurance moving too fast?

Marcus: I don’t think too fast. We have to be humble about what we can predict about the capabilities of technology. There’s another kind of humility which is even when we have internalized and understood it accurately to translate that into something that really works that closes the last mile and really creates value – that’s another place where there’s lots of ways to go wrong.

19. Is AI overhyped right now?

Marcus: There’s a kind of political pressure in organizations to declare victory quickly, especially in the tech industry. There’s been some really striking announcements by tech CEOs… I think all of these are aspirational. Maybe some of it will come true, but it’s hard to know. However, all of that said, I think more than any other time in my career, there are possibilities for very dramatic impacts in large-scale processes that happen within an insurer.

20. Is AI adoption an imperative for businesses?

Marcus: I don’t think any organization that serves clients in any industry in any financial service has the luxury of saying things can be exactly the same. Even if you’re convinced that your client base is happy the way things are, your cost structure to serve them is a competitive consideration because you have peers that can serve them just as well or maybe even better to their preferences at a completely different cost structure.

Insights into Solving the Insurance Industry’s Challenges

21. Can AI solve the problems faced by the insurance industry?

Amrish:  It’s not instantaneous. It’s not that you deploy AI and suddenly everything – the talent crisis and the profitability crisis – will get solved. But what I’m noticing the winners doing is finding these intermediate stages that demonstrate the value that AI is bringing. You will do a large transformation, but you will see piecemeal value.

22. Will AI help with the insurance industry’s talent crisis?

Amrish: There’s some estimate that 50% of the insurance workforce is going to retire in the next 15 years, and CEOs are worried, chief human resource officers are worried about what we will do next. This is where the promise of AI has been so powerful. You can take a 30-year underwriter veteran and codify their knowledge in AI, or you can take an SIU expert and codify their fraud knowledge in AI.

23. How can AI help with the insurance industry’s talent concerns?

Marcus: I think AI is the first technology since the written word where we have a way of instantiating our expertise, our knowledge, our taste, into a persistent form.

Insights into the Risks of AI Adoption

24. Will younger generations become too dependent on AI?

Marcus: I think this is a deep and universal question, but I don’t think it’s new to AI. I mean, how many of us know the way that a combustion engine works? How many of us know the way our water processing system works? Most of us subsist in a technology-infused world, most of whose workings we don’t understand and cannot individually reconstruct. And yet on top of that foundation, we’ve been able to exercise new capabilities, new potential. I think that will happen in AI as well.

25. Will insurance take on too much risk as it adopts AI?

Marcus: I have no concern essentially that the industry will be reckless in its adoption and start empowering poorly trained, poorly supervised agents to do crazy things. If I have a concern for the industry, it’s that it will be too cautious.

Want more insights? Watch the full webinar recording. 

Answers have been edited for clarity and flow.