The art of integrating transformational AI – key considerations and approaches

AI has gone mainstream

Artificial Intelligence (AI) is no longer a peripheral or “emerging" technology for insurance companies. According to a recent Accenture report, “Across all industries, including insurance, AI ranked consistently as the top game-changing technology in Gartner’s CIO surveys over the last three years (2019 to 2021). Insurers are starting to see the value of AI and adoption is set to accelerate. While less than half of claims executives (44%) say their organizations are advanced in use of automation, AI and machine learning-based data analytics - 80% say these technologies can bring more value, and 65% say they plan to invest more than $10 million into AI in the next three years.”
⁠
⁠BCG put it best: for a business that manages risk, "ignoring AI is risky business."
⁠
⁠The fact that so many insurance companies are adopting the technology is testament to its real, transformative impact on business. At the highest level, it creates efficiency and reduces waste. At Tractable, we’ve seen the positive domino effect it brings the organization – quicker and consistently more accurate decision-making at first-notice-of-loss (FNOL), and at every point in the vehicle repair and salvage process. These faster workflows help save resources which can be used more productively elsewhere. Critically, by speeding up processes, companies naturally create better customer experiences. We’re seeing this right now in Japan with our Property Estimator solution, which is helping settle numerous claims in a record 24 hours, and some in as little as 3 hours. So from the second the claim is made, to the second a cash settlement is in a homeowner’s bank account, there’s no more wasted time hanging around for an extremely busy team of assessors to get through a backlog of homeowner claims. Anxious families facing one of the most difficult times in their lives are able to rebuild their homes and move on quicker than ever before possible.

Change is accelerating 

Looking at the bigger picture, the need for innovative technology that can drive efficiency is more pressing than ever. Increased inflation (8.5% in the US - August 2022),

constrained global supply chains (J.P. Morgan Research’s global car production assumptions have been updated from +4% to -1% for the 2022 fiscal year) and climate volatility (global insured losses from wildfires increased from $8.7bn 2001-2010 to $56.3bn 2011-2020) have created a nasty cocktail of problems that result in ever-increasing costs that need managing. AI’s dual capability of creating efficiency, as well as innovative customer experiences, can not just help, but be transformative. 

Transformation is the key

If AI technology is now a value creator, then the next question is how to integrate it in as seamless, robust and futureproof a way as possible? At Tractable, we see that debate play out across all of our customers. It involves many parts of the organization. What a Transformation Director might want may differ from the sensibilities of a CIO.
⁠
⁠Companies want change with ease. The best implementation is something that will work out-of-the box quickly, while not drastically changing company roadmaps or requiring a complete overhaul of training for the new systems. The onboarding process has to be swift and workflows cannot be compromised. The best solutions are ones that feel like they fit seamlessly and have always somehow been there. They are synergistic to the way people work within that realm and are not adversely detracting from what the day-to -day tasks at hand are.
⁠
⁠Reconciling all those needs can often feel like an unattainable task. But in our experience, the art of successfully embedding technology comes down to four main considerations tied to one single question: what does it take to enable long-term, transformational business results with AI?

1. Adopt an open ecosystem

Let’s define an open ecosystem as a platform that freely supports and encourages integrations and data flows with other technologies, services and platforms.
⁠
⁠Easier said than done. But it’s worthwhile remembering why it should be a goal. The pressure on the insurance sector has intensified and with that pressure comes the need to adapt, by creating more efficient end-to-end processes and more seamless customer experiences, for example. Speed to market is critical to keep ahead of competitors. And technology is the critical enabler to unlock all of that value. 
⁠
⁠Some see the benefits of having a closed ecosystem. Problems may be fixed more easily and there can be a greater level of consistency. But that comes at the expense of accelerated innovation. The more an ecosystem is controlled by a single vendor - a ‘walled-garden’ with limited room for other technologies and data flows to create innovation within it - then the slower the company is to change and the higher risk they run of being left behind by more innovative competitors. Conversely, the more open the ecosystem, the more the reverse holds true. Companies are able to collaborate broadly and quickly to offer groundbreaking and incremental innovation in an environment where data flows more freely. And that drives the industry forward as a whole. It keeps older companies relevant, opens the door for more impactful tech partnerships  and ultimately, benefits the end user.
⁠
⁠As Pablo Mellao, Claims Transformation Manager at Admiral Seguros says,

⁠Of course, this is not a quick or an easy fix. Companies will have invested heavily in their technical solutions. So incremental steps in opening up the discrete parts of an ecosystem to potential solutions are worthwhile - more on that later - while always bearing in mind that AI adoption is not an investment in re-platforming. It’s an investment in real business transformation.
⁠
⁠And at Tractable, our systems can help you either embrace this in-full or incrementally.

2. Calibrate to your business

When it comes to integrations, there are a whole host of technical options. The holy grails are “plug and play,” cloud-based or server-to-server API integrations. Those options may work well for a mass market, horizontal SaaS analytics platform. But if we’re talking about applying AI to drive a business’ transformation agenda, then we need an alternative approach. 
⁠
⁠Expectations for the performance of AI and computer vision technology are high. At its best, the tech transforms your business results and improves specific customer experiences. In the case of companies working in the property and auto repair ecosystem, this is complex. For example, repair standards, parts, labor and contractor costs all differ by country and company. So do data formats that any given company supplies an AI system to ingest, from images and faxes through to PDFs. And on the other side, there is an eye-widening array of other platforms and considerations that the AI engine’s output has to integrate with. So while a vendor, such as ourselves, may have built a finely tuned, performance-ready AI engine (let’s call it “the core”), the system needs some calibration to your business’ and your countries’ (repair) standards, as well as to your input and output norms. Without this calibration, true automation will never be achieved. 
⁠
⁠Here, your investment should be carefully considered - you need to look for evidence of AI being effective within large businesses and that the AI should be effective from day one. You need a team that comes with the AI, to calibrate, operationalize and build success, supported by a level of expertise based on both what the AI will learn to improve on and the embedded base of expertise that will drive it. The handover from partial automation to full will never happen with vaporware, yet a strong proof of concept (POC) might say otherwise. Be mindful of what the steps to success are and ask the right questions.

3. Verify core capability 

There’s no doubt about it: AI hype is big. We’ve written in the past about how to test and verify an AI system. But at its core, it comes down to data, modeling and problem solving.
⁠
⁠The advanced state of Tractable’s AI is down to a purpose-built engine that uses different techniques, such as natural language processing, computer vision, machine learning and fusion to solve multi-layered, complex problems immediately. We built this, in close collaboration with customers. Now ‘generations’ in, it’s continuously being refined by internal domain specialists working in tandem with research scientists. Added to that is an enormous bank of continuously growing global data inputs (such as images of damage). The combination means our damage assessment reports and repair estimates are more robust, accurate, useful and trusted by our customers, and their customers, in their specific geographies. That foundation and the culture that built it grow stronger every day.
⁠
⁠It means that when we integrate our tech into customer systems - as we have done for over 30 customers across the US, EMEA and APAC - we have a uniquely strong head start before we calibrate our system even further to customer and country repair standards.

4. Partner on delivering ROI

The case for business transformation is necessarily broad. And there might well be an appetite to go big and broad at first. But, as we’ve discussed, technical barriers often mean integration of new technology into existing systems is a challenge. And while case studies of results with other companies can be compelling enough to get started, there needs to be real empirical evidence that the AI can work within any given business before it’s rolled out broadly. This agile approach to rolling out software is, of course, best practice and to be expected. So, when conducting a Proof of Value or an initial integration, it’s better to start small and focus on getting a return on investment (ROI) as quickly as possible by opening up specific parts of your ecosystem focused on specific use cases. But before you even get to that, a critical part of the process is making sure the right people in the organization are aligned on metrics of success to start with.
⁠
⁠For example, we and our customers understand that changing the way their business estimates property damage can’t be done overnight. But on a relatively small scale and recent tests, we have proven that our Property Estimator product can dramatically cut down homeowner claim cycle times. The average settlement time from making a claim to money in a homeowner’s account is just a single day.

⁠On the auto side, our Subrogation application can help drive huge business efficiency. Our AI can cut down the time it takes for a manual subrogation claim from 15-30 minutes to under a minute, which is a x15 to x30 return on time spent. Our Total Loss application can  identify 97% of all losses up front with up to 95% accuracy, which results in x10 faster cycle times. Claim Review can reduce cycle times by up to x10 and finally, Speedy Payment can drive 3.5x ROI. These performance metrics indicate real, meaningful ROI on AI investment.
⁠
⁠The key to delivering ROI by integrating AI across your business is very much a two way street between your business and your AI partner. It has to be done by a strong accompanying service layer to onboard teams, bring the business along, spot opportunities for improvement and drive the transformation you will be seeking. These teams bring the experience of having leveraged AI across the business with similar use cases. That knowledge and understanding can’t be attained through impersonal, corporate, cookie-cutter programs. It takes personal attention, hands-on service and cross-team collaboration. It’s something we, and our customers, value as much as the technology itself.