Federato’s co-founder and CEO Will Ross combined deep AI experience with practical market analysis to create a company that is upending the insurance industry.
Ross has been involved with AI in some fashion for 15 years. Through that period, he’s seen two waves. The first was deep learning, while the second saw retrained transformers based on neural networks, deep learning and large language models that fostered generative vision and language models. That’s the foundation for agentic AI.
Armed with that AI knowledge, Ross looked for use cases he thought AI would be good at solving, because it’s not great at everything. He zeroed in on business processes and asked some questions.
- Where is an ultra-high-knowledge value worker who has an unfair job?
- Would they do that job completely different if they had unlimited time?
- Can an Ai agent do it approximately the way a person would if they had unlimited time?
- Is that approximate way better than they way they do it today?
What Federato learned from Uber
Ross likens Federate and insurance to Uber improving the taxi experience. Taxi dispatchers had unfair jobs as they tried to maximize efficiency with radios, phones and maps. Uber does it better.
“What AI does in Uber… it’s approximate, not perfect,” Ross said. “You don’t know if you had the most optimal ride in Uber and that’s the key. It’s so much better than you used to have that you don’t care.”
Insurance offers a similar path. Ross described it as a probability-based industry where underwriters have unfair jobs. Federato boils it down to its basic concept of capital on one side and hedged risk on the other. Using AI, it attacks a host of problems in the middle.
Manual underwriting involves assess one file at a time for risk and price. Yet, that one price doesn’t matter at the end of the day, what counts is that the overall portfolio is correctly assembled.
“That’s what no human can do well proactively, and AI doing it approximately well can do way better,” Ross said.
Instead of a human agent taking the next file off the pile, AI can proactively identify the best ones to work on, those that can be priced effectively. Ignore the poor bets.
Customers wanted more
That was Federato’s core concept, but customers asked for more. Could Federato’s service be expanded to automate the process beginning with an application hitting the inbox?
Ross knew it could. The ratings process has quite deterministic outputs, yet inputs are up to the underwriter’s discretion. Which 10 out of 100 questions will they ask? AI asks all 100 at once and produces a report with its reasoning that the underwriter can assess.
“It creates this reinforcing loop that makes the human better at point of decision or point of choice while also driving a massive amount of efficiency,” Ross said. He added that timelines can be reduced by 89% as risk selection improves 3.7X.
“This is living our mission,” Ross said. “Our mission as a company is changing how insurance work gets done. This is radical change.”
Human in the loop?
Most fintech AI discussions eventually turn back to the importance of humans in the loop. AI isn’t trusted 100%.
The conversations are different in insurance, where many executives welcome automation. Having done those manual processes early in their careers, they know that much of the work is replaceable, with an approximate AI job being better than a fully-human output. All work evolves, and so will underwriters, who Ross said could become portfolio managers who track business flows.
“In these private conversations, people are telling us to bring on the robots,” Ross said. “We are seeing a very different appetite from the true decision makers in our industry. They think this should change.”
Ross stressed that AI does not change how the product works. What it does is shift how that work is done and gets people to the end point while allowing them to do their jobs better.
“I do see an incredible opportunity for change, and I’m here to facilitate that change,” Ross said. “Insurance is a product we all need and buy. If we’re paying for all that extra time, that just goes into your premiums. That’s not efficient and equitable.
“If you can reduce the cost of reinsurance by originating a more effective platform up front through the underwriting process, which AI can do, you can save a meaningful amount of money.”