By Marielle Rodriguez, Social Media and Brand Design Coordinator, Triple-I
Triple-I’s Insurance Careers Corner series was created to highlight insurance industry leaders and raise awareness of career opportunities in the industry.
This month we interviewed Sunil Rawat, co-founder and CEO of Omniscience, a Silicon Valley-based AI startup specializing in computational insurance. Omniscience leverages five “mega-services” that include underwriting automation, customer intelligence, claims optimization, risk optimization and actuarial advice to help insurance companies improve their decision-making and achieve greater success.
We spoke to Rawat to discuss his tech background, the role of omniscience technology in measuring and assessing risk, and the potential shortcomings in underwriting automation.
Tell me about your interest in building your company. What brought you to your current position and what inspired you to found your company?
I come from the technology industry. I worked for Hewlett Packard for about 11 years and hp.com grew about 100,000% during my time there. Then I helped Nokia build what is now known as Here Maps, which in turn supports Bing Maps, Yahoo Maps, Garmin, Mercedes, Land Rover, Amazon, and other map systems.
I met my co-founder Manu Shukla a few years ago. He’s more of the mad scientist, the applied mathematician. He wrote the predictive caching engine in the Oracle database, the user profiling system for AOL and the recommendation system for Comcast. For Deloitte Financial Advisory Services, he wrote the text mining system used in the Lehman Brothers probe, the Deepwater Horizon probe, and in the recent Volkswagen emissions scandal. He’s the “Distributed Algorithm Type” and I’m the “Distributed Systems Type.” We are both very technical and have the ability to do calculations on a very high scale.
We see an increasing complexity in the world, be it demographic, social, ecological, political, technological or geopolitical. Decision making has become much more complex. Where human lives are at stake or large sums of money are at stake with every single decision, the accuracy of every single decision must be extremely high. Here we can use our computing power from our knowledge of the last 20 years and bring it to the insurance sector. That’s why we founded the company – to solve these complex risk management problems. We really focus on computational finance and especially computational insurance.
What is the general mission of Omniscience?
It should become the company that executives go to when they want to solve complex problems. It’s about empowering leading financial services providers to improve risk selection through hyperscale calculations.
What are your most important products and services and what role does omniscience technology play?
One of our core products is underwriting automation. We love to solve stubborn problems. When we think of underwriting, we think of facultative underwriting for life insurance, where you need human underwriters. The decision heuristic is so complex. Imagine a 25 year old non-smoker asking for a 10 year policy of $ 50,000 – it’s kind of a given and you can give them that policy. On the other hand, if they ask for $ 50 million, you will surely ask for a blood test, psychological exam, keratin hair test, and everything in between. You need people to make these decisions. We managed to address this problem and use our technology to digitize it. If you take a few hundred fields of data and a couple of 100,000 cases to build an AI model, it quickly becomes utterly unsolvable from a computational perspective. Here, with our technology, we can view all data in all its facets – we automate and use everything.
Once you have the AI underwriter’s brain in software, think from a customer intelligence perspective. You have all of this extensive transactional data from your customers to pre-back, qualify, and recommend for various products. We have also built a great skill in data collection. For employee compensation and general liability, we have the data that improves the agent experience. We can also correctly classify all NAICS codes and help with harm avoidance and the search for hidden risks. We also have a great OCR function. With regard to the digitization of text, we can take complex tabular data and digitize it without human intervention. We are able to do this worldwide, even in complex Asian languages. We also do a lot of work in asset and liability management and can perform calculations that were very weak and imprecise in the past. We can do these calculations on a daily or weekly or yearly basis, which makes a huge difference for insurance companies.
We also work when there is a risk of forest fire. Many forest fire spread models look at a postcode +4 or postcode, and it takes them about four hours to predict one hour of wildfire spread, about 96 hours to predict a day of wildfire spread at a postal code level. California, where I am, had a lot of wildfires last year. If you double the density of the grid, the calculation will increase 8 times. What we have been able to improve is to consider the grid on 30 square meters, almost with an individual plot size. You can look at the risk of the houses individually. At a 30 meter level, we can do an hour of wildfire spread in 10 seconds, basically a day in about four minutes.
Are there potential mistakes in relying too much on automation technology that leaves out the human factor?
Absolutely. The problem with AI systems is that they are generally only as good as the data on which they are based. The most important thing is that we can achieve an accuracy of over 90 percent with every single decision, because we can look at all data and all its facets. You also need explainability. It’s not like an underwriter makes a decision in the blink of an eye and then justifies the decision. What you need from a regulatory or verifiable point of view is that you need to document a decision during the decision making process.
If you are modeling from historical data, how do you ensure that certain groups are not re-skewed? You need bias tests. Explainability, transparency, scalability, adaptability – all of these are very important. From a change management and risk management perspective, you have the AI to make the decision and then you have a human review. After doing this process for a few months, there is a very risk-controlled way to introduce this. Each AI should also express its confidence in its decision. It is very easy to decide, but you also need to be able to provide your trust number and people always need to look out for that trust number.
What is traditional insurance lacking in technology and innovation? How is your technology changing insurance?
Insurers know their domain better than any Insurtech can ever know their domain. In a way, insurance is the original data science. Insurers are very brilliant people, but they have no experience in software engineering or scale computing. The first instinct is to look at open source tools or buy some tools from vendors to build your own models. That doesn’t work because the methods are so different. It’s like saying, “I’m not going to buy Microsoft Windows, I’m going to write my own Microsoft Windows,” but that’s not their core business. You should be using your Microsoft Windows to run Excel to create actuarial models, but you wouldn’t try to write your own programs.
We’re good at systems programming and scale computing because we have a technical background. I would not be arrogant to believe that we know as much about insurance as any other insurance company, but this combination of domain insurance and domain computing enables leading companies in the field to outstrip their competitors.
Are there any current projects that you are currently working on and what trends do you see in Big Data that you are looking forward to?
Underwriting and digitization, cat management and forest fire risk are exciting and some of the work we do in ALM calculations. When regulators ask you every quarter to show that you have enough assets to pay off your debts for the next 60 years, it becomes very complex. This is where all of our mega-services come into play – if you can combine underwriting, claims and capital management, you can become much better at making choices and can decide very dynamically how much risk you want to take, as opposed to a very static route .
The other things we look forward to is wealth management. We are doing an interesting job with a very large insurer. What we have been able to do is increase the return through various strategies. That’s another area we’re looking forward to – it’s growing pretty fast over the next year.
What are your goals for 2021 and beyond?
It’s about helping insurers develop that decade-long compounding advantage through better choices, and we’re just going to move on. We have developed a lot of intellectual property and technology, and we have pilot customers in different regions that have used our technology. We have the evidence and the case studies, and now we’re doubling the growth of our business, be it with the same customers as us or let’s move on to more product lines. We are focused on serving these customers and adding a few more customers in the three areas in which we operate, namely Japan, Hong Kong, China and North America. We focus on methodically implementing our plan.