The Center for Data Innovation spoke with Jakub Dryjas, CEO of Tensorflight, a Polish-American firm that mixes satellite tv for pc, aerial and floor imagery with synthetic intelligence (AI) applied sciences to assist property threat assessments. Dryjas spoke about Tensorflight’s partnership to assist rebuild Ukraine for the reason that outbreak of warfare.
Gillian Diebold: What is proprietary info and what are its advantages?
Jakub Dry Jacket: Gathering correct property information has all the time been a problem. In the previous, property house owners and insurers had been required to conduct on-site inspections to evaluate the situation of a particular asset. As know-how progressed, insurers started utilizing public data, previous property valuations, and databases based mostly on collections of public data. But it is easy to see how these processes and databases can grow to be contaminated with incomplete, outdated, and even incorrect info. When information is wrong, it will increase the danger that insurers and underwriters will challenge insurance policies and premiums that don’t really replicate the worth of the property.
In the insurance coverage business, insurers can use proprietary info to extra precisely underwrite insurance policies and create extra aggressive insurance coverage merchandise. Property intelligence consists of offering extra real-time, correct information for insurers to behave on and integrating trendy options, equivalent to Tensorflight, to supply that information in an simply accessible manner; for instance straight through API or an internet portal.
Diebold: Tensorflight works within the “insurtech” house. What does that imply and how much information is collected for this sort of know-how?
Dry jacket: Insurtech refers back to the applied sciences that assist to enhance the effectivity of the insurance coverage business. While know-how has remodeled many elements of the insurance coverage business, the tempo of change is gradual and now we have but to see a lot significant impression on actual property insurance coverage. Many insurers nonetheless depend on public data and decades-old surveys to collect property info and inform their selections. As a consequence, too many buildings have gotten underinsured, highlighting the dangers related to guide, time-consuming property inspections and valuation processes.
We imagine one of the simplest ways to assist insurers reduce threat and enhance their processes is by accumulating and analyzing extra up-to-date, related information. We are significantly targeted on geospatial information, together with spatial information, ground-level imagery, aerial imagery, high-resolution satellite tv for pc imagery, and multispectral imagery.
By offering insurers with extra strong geospatial information and making certain that the info is updated, insurers can extra precisely assess the danger of insuring actual property. For instance, Tensorflight can present detailed information on the fabric and situation of a constructing’s facade, building sort, alternative prices, occupancy sort and roof geometry. Insurers can use this information to scale back premium leakage, present clients with immediate and automatic coverage quotes, enhance their claims ratios, considerably cut back on-site inspection prices and deal with claims extra effectively.
Diebold: Your firm helps rebuild Ukraine in cooperation with the Kyiv School of Economics. Can you discuss Tensorflight’s contributions to that mission?
Dry jacket: Since the outbreak of warfare in Ukraine in February 2022, the KSE Institute, a number one assume tank and the analytical unit of the Kyiv School of Economics, has spearheaded a initiative to gather, consider, analyze and doc info on the harm accomplished to infrastructure because of the Russian invasion.
The KSE Institute sought outdoors experience to supply intelligence on a bigger scale and, since April 2022, Tensorflight has been offering and analyzing satellite tv for pc imagery to evaluate harm to infrastructure, significantly inside Bucha, Irpin and Mariupol, on a pro-bono foundation. By analyzing satellite tv for pc imagery and rooftop geocoding info, we are able to present perception into the extent of harm to a given constructing within the areas analyzed, whereas additionally offering the precise geographic coordinates of the buildings.
Our group of AI consultants has educated a deep neural community mannequin to detect and classify war-damaged buildings based mostly on satellite tv for pc imagery, reaching over 90 % accuracy. Our subsequent plan is to investigate satellite tv for pc photographs of Lysychansk, Melitopol and Popasna. However, satellite tv for pc photographs are extraordinarily expensive. We are going through greater than €5 million to scan Ukraine’s 100,000 sq. kilometers nonetheless awaiting harm evaluation. In addition, this information rapidly turns into outdated when additional harm is completed, so we’re at present searching for sponsors and companions to assist fund and proceed the mission.
Diebold: How can your product be of worth to particular person shoppers along with corporations?
Dry jacket: When property insurance coverage corporations endure from an absence of dependable information, it impacts the standard of the insurance policies and companies they will present to their finish clients or policyholders. Tensorflight helps insurers make extra correct and detailed property assessments, enabling them to supply clients extra tailor-made and efficient insurance coverage insurance policies. In addition, by eliminating the necessity for guide inspections, insurers can generate extra correct quotes in much less time, enhancing buyer expertise and finally retention and satisfaction.
Without this correct and up-to-date details about the property to be insured, insurers might not have the ability to correctly assess the dangers concerned and supply acceptable protection. This can result in insurance policies which are insufficient or too costly and depart policyholders weak to monetary loss within the occasion of a declare. Our most important purpose is to scale back threat for each insurers and policyholders, leading to a extra aggressive and sustainable property insurance coverage marketplace for all.
Diebold: What units Tensorflight aside from different proprietary info organizations?
Dry jacket: Tensorflight combines the very best industrial decision floor degree, aerial and satellite tv for pc imagery obtainable in the marketplace to generate probably the most correct property information. This information ranges from info on the situation of roofs to break evaluation and valuation after pure disasters and inside battle zones. In these conditions, on-site inspections will not be doable and different current imaging sources don’t present the extent of element wanted for correct assessments. Our platform may mix current and outdated information, permitting insurers to see property modifications from as little as six months in the past, in addition to how properties have modified over time, enabling them to make extra knowledgeable and knowledgeable selections.
We are a frontrunner within the secure, moral use of actual property information and our international platform offers complete protection of hundreds of thousands of properties throughout North America, Europe, UK, Australia and South East Asia.
Source: datainnovation.org