February 28 2023 | Read on FFNEWS

Speed and security are often seen as being mutually exclusive in claims processing…Clearspeed CEO and Co-founder Alex Martin learned his lessons on the battlefield to resolve that conflict once and for all

Faster processing is a common goal for insurers because it improves customer experience at key touchpoints in the insurance lifecycle, such as first notice of loss (FNOL).The claims process is always stressful and customers expect a quick resolution, so the better the service at this moment of truth, the happier they are. Faster processing is good for retention and reputation, but speed, especially in high-volume claims situations, can increase the likelihood of fraud. The winning formula behind Clearspeed, a voice analytics company whose AI-enabled technology can accelerate low- risk transactions while flagging others with remarkably high accuracy rates, addresses that dilemma for insurers.

Launched in 2016, Clearspeed is based in San Diego and was set up to solve the challenge of how to objectively screen people, in any language, at scale. Alex Martin, CEO and co-founder, drew on his background in the armed forces, combined with his experience of working with talented voice technology specialists, as the inspiration for Clearspeed’s technology.

“The problem we faced in combat zones, particularly in the special operations field, was how to operate efficiently in environments where trust was imperative, but the data to trust people was incomplete,” says Martin, who served nearly 20 years in the US Marine Corps in active duty and as a reserve. “Operating across different languages can create uncertainty, which undermines trust, and trust is something you must establish quickly in military situations and hostile regions.”

Martin explains that in exploring ways to trust people faster, they focussed on what voice could reveal about a person and whether what they say could be relied on. He cites the example of British and American forces on missions in Africa and the Middle East and their need to liaise with local nationals. How could they form trusted relationships when there was no immediate way to vet a contact and establish confidence in what you were being told? “You might think a lie detector would be the simple answer,” says Martin. “But this isn’t a lie detection problem – this is a question of risk identification; lie detectors don’t scale, plus the technology wasn’t available to us in conflict zones such as Iraq. Lie detectors help you make binary assumptions, either lie or no lie, in one-to-one situations. But we faced risk identification and risk stratification problems and needed to cover more ground at speed. The only way was to come at this problem from a different angle.”

Rather than a lie detector, Martin says that the Clearspeed model is akin to a metal detector: it clears large volumes quickly and efficiently, while producing risk alerts for follow up. It clears the way and eliminates people from the search if there is nothing of interest or concern, he says. Unless there is an alert recommending you take a closer look, you can safely move people ahead. The quest to build trust led to the development of the Clearspeed technology and the ability to apply the metal detector concept across industries – including insurance.

“Our product suite is based on the need for speed with safety,” says Martin. “Instead of trying to find the bad actor, and make a final determination as such, which is what a lie detector does, we wanted a system to quickly clear all the low-risk people, while identifying the few who needed further scrutiny in the form of a risk alert, to be mitigated or escalated. Traditional logic says you should slow things down to tighten security whereas we’re accelerating the vetting process with technology that’s designed not to find the ‘bad’, but to rapidly clear the ‘good’.”So, how does Clearspeed’s technology work?

Martin is keen to underline that it’s not voice recognition and has nothing to do with the growing field of biometrics. Instead, it’s based purely on the assessment of vocal characteristics associated with risk.

“It’s an important distinction,” he says. “We’re analysing the voice, based on a series of automated questions to which people provide ‘yes’ or ‘no’ responses. Their answers provide information that indicates low, potential, or high risk.” It’s a four-step process, beginning with the telephonic questionnaire. The speech is converted into Clearspeed’s unbiased and language-independent data model. The data is then analysed to determine the presence or absence of specific vocal characteristics associated with risk. Next, the risk is scored from low to high and the results are provided via an API or an app.

“Our technology captures detectable and quantifiable characteristics that are universally known to be present in speech,” says Martin. “And that’s not down to the AI; it’s because our system knows exactly what characteristics to look for, and we know exactly what’s noise and what’s a reliable signal. The AI is simply conducting the evaluation of low, medium, or high risk against our extensive global data set.”

Martin adds that the reservations people have about AI, such as creep or bias, are not a concern with Clearspeed because it functions more like a heart monitor, identifying something that is not subjective. In other words, the signal is either there or it isn’t, and if it is, to what extent. And the voice analysis requires no additional data, like race, gender, or location, and is language agnostic as well, says Martin. Depending on the product, and the application, he claims very high accuracy and flagging precision for the technology in any real-world situation where screening is required.

“The questions are tailored for the element of risk that you’re trying to identify,” he says. “ We create simple questions, directly related to each customer, for an experience that’s personalized, accurate and rapid. We’re not looking to complicate a process that traditionally proves stressful for customers. Just like the heart monitor tracks signals and provides alerts when there is a patient issue and additional care is required – which a medical expert then determines – Clearspeed voice analytics alerts to potential high risk to enable claims experts to follow up, but most importantly, has cleared all the low-risk claims with unprecedented speed.

“Clearspeed is focused on helping our clients make informed decisions, not making decisions for them. This is all about making bold decisions that benefit the customer with profound confidence,” says Martin.There are three Clearspeed insurance products. Clearspeed Express is designed to accelerate FNOL. The automated question-and-answer process enables insurers to triage claims at scale, boosting straight-through processing so that low-risk claims can be paid without any unnecessary delays, while high-risk claims can be flagged for a closer look.

“Traditional logic says you should slow things down to tighten security whereas we’re accelerating the vetting process with technology that’s designed not to find the ‘bad’, but to rapidly clear the ‘good’”

“Clearspeed Express fast-tracks decision-making,” says Martin, “reducing the impact on your operations and giving adjusters more time to focus on high-priority claims. Our cloud architecture makes it quick and easy to integrate Clearspeed alerts into any claims process via any customer touchpoint.” For straight-through processing (STP) to work efficiently, it needs to start with claims triage at FNOL, because the vast majority of claims are legitimate and don’t need to be questioned; good actors should not suffer because of unnecessary friction. However, many insurers have yet to implement STP at this crucial stage, despite the growing use of techniques such as natural language processing (NLP) to better understand customers. The problem with NLP systems, though, is that they require large amounts of data.

In contrast, Clearspeed’s automated questionnaires require very little data to draw accurate conclusions. “The questions we pose at FNOL are very simple,” says Martin. “For instance, ‘were you on a mobile phone when the incident occurred?’ or ‘were there any passengers in your vehicle when the incident occurred?’. Based on the responses, the risk is assessed, and an overall risk score is provided. If the score is satisfactory, the claim can move forward.” Clearspeed Precision is a tool that enables insurers to gain greater accuracy when assessing claims beyond FNOL.

It uses the same voice questionnaire approach and light touch to achieve a positive customer experience. “Clearspeed Precision goes a little deeper and more granular,” says Martin. “So, instead of the top-of-funnel triage, signalling clear or alert, this is now saying, ‘what aspect of this needs further investigation?’. It builds on the ‘yes’ or ‘no’ format with supplementary questions to either eliminate false positives or escalate to the fraud team.”

Clearspeed Surge is the newest addition to the product suite. Launched at the end of 2022, it enables insurers to assess claims risk quickly and accurately for catastrophic, weather-based events. Operational teams can instantly determine next steps and move customers straight through to payment at critical times if there are no issues. By fast-tracking low-risk customers, Clearspeed Surge maintains Clearspeed’s objective of gaining trust, faster.

“During a ‘surge’ event, meaning anything that’s not routine, call volumes can increase more than 50 times,” says Martin. “And with catastrophic events, the risk of opportunistic claims fraud is far greater. For instance, when hurricane Ian devastated many areas of Florida, there was a big increase in attempted fraud and the US Department of Justice had to issue a fraud warning. “If an insurer gets flooded with tens of thousands of calls a week and doesn’t have the capacity to meet demand, Clearspeed Surge can step in and help triage the influx of new claims. It sits in the cloud and easily integrates with existing systems and offers round-the-clock support. When insurers face this increased demand and struggle with effective fraud controls, trust – and ultimately the customer experience – suffer. The challenge is confidently enabling straight-through processing at first notice of loss, which is where we provide the solution.”

While Martin emphasises that Clearspeed should be seen to enhance customer experience and operational efficiency through straight-through processing, and therefore is not a dedicated anti-fraud platform, it does indeed result in better fraud prevention. With speed comes security. That might seem counterintuitive but it’s a logical outcome of Clearspeed’s approach. Furthermore, the questionnaire format helps to deter potential fraudsters in the first place.“Fraud is not only on the rise with disaster-related or weather-related events,” says Martin. ”It’s a growing problem across all types of claims, and fraud strategies are continually evolving.”

The impact that Clearspeed is having is evident from customer metrics. Martin says early adopters are seeing big gains in operational efficiency and customer experience, while fraud rates are dropping.“One of our first customers, an auto insurer in the Americas, has significantly accelerated its claims payments,” says Martin. “In some cases, it’s reduced claims processing from a month to just one day. In addition to gains in efficiency and customer experience, this insurer also identified nearly $1million in savings on fraud losses.“We’ve also seen some good results in the UK, with one insurer reducing claims processing from 22 days to five days, while another succeeded in pushing injury claims withdrawals at the special investigative unit stage from zero to 30 per cent”.

“When insurers can’t cope with the demand and struggle with effective fraud controls, customer experience and trust suffer”

Martin says that Clearspeed has no direct competitors. While there are many providers who offer risk tools, some of which may be complementary, no one is covering the voice assessment angle in the same way. Given the interest and successful uptake from insurers so far, he believes there is plenty of scope to develop voice technology in other areas.

“I think there is incredible potential for growth,” he says. “First, by developing end-to-end technology to help with underwriting through to claims, which will strengthen customer relationships and retention. The trust factor is very important, and if customers have consistently good scores across all touchpoints, it may help to lower premiums and further boost customer experience.“Secondly, voice and automated interactions can lift customer support to another level and provide a real-time service when customers need it most. For example, for repairs following a car accident. Above all, Clearspeed is showing that prioritising customer experience also has benefits for fraud prevention.”Although Clearspeed’s focus has been claims, Martin says the voice technology could also be used at the quote stage to accelerate onboarding and help determine premiums. As for adding visual technology to support voice, he says that’s not part of the Clearspeed strategy.

“We know what we’re good at,” he says, “and that’s providing voice technology and developing new use cases for it. We can partner with others and integrate our services to create an even stronger insurance proposition, but voice is our core business and selling point.”

Social responsibility is very much part of Clearspeed’s business ethos, says Martin, and it’s closely involved in climate change and poverty alleviation. This is partly borne of Martin’s experience working as a team leader for Nuru International, a social venture that helps rural communities in Africa, but also because one of Clearspeed’s board members and first investor Bob King is a co-founder of King Philanthropies. It draws on a team of highly strategic social impact investors who are committed to projects that alleviate poverty and fight climate change. Clearspeed aligns with the organisation’s philosophy and values, especially through products such as Clearspeed Surge, which helps to mitigate the impact of extreme events and supports people in times of greatest need.

“Clearspeed is proud to be developing socially responsible technology with the support of impact funding,” says Martin. “We have a group of incredible investors who want to make a meaningful difference to people’s lives, and that gels perfectly with our values as a technology innovator. It’s about creating products that establish direct value and impact, while retaining their accessibility and quality in use.”

“Our products are helping people every day in insurance and other fields, and we are committed to finding new and socially responsible uses for voice analytics where the risk-speed paradigm must be flipped to favour building bridges not walls.”

A ‘clear’ case for adoption in Guatemala

Seguros El Roble is a multi-lines insurer commanding more than 20 per cent market share in Guatemala.It’s constantly reviewing and trialling digital solutions – both customer-facing and in the back office – and it was this open-minded attitude to innovation that led it to a tech show in Las Vegas in late 2019 where Clearspeed, then only three years old, was exhibiting its voice-based analytics technology. More than 100 companies piqued the insurer’s interest at the show, but only Clearspeed progressed to production, so impressed was it by the simplicity of the solution and ease of integration into its existing fraud detection stack.

Jorge Luis Linares, Seguros El Roble reinsurance project manager, said a solution that could help us confidently assess the risk of fraud in a claim simply by analysing the answers to a handful of yes/no questions ‘was to us like magic’. “It was easy to implement and the impact was very high. For me, it’s still kind of magic!” he said during a webinar early last year.At the time, the insurer had recently implemented the Dutch trust automation platform FRISS and was looking to further enhance the impressive affect that had already had on straight-through processing of vehicle claims.

The technologies are complementary to one other. Every claim is run through FRISS and 90 per cent of them are settled without intervention; for the remaining 10 per cent that receive a ‘red’ or ‘orange’ FRISS score or are a total loss or theft claim, the company activates Clearspeed.

“We are not denying claims based on either the FRISS score or a call with Clearspeed. They’re only a tool to say, ‘we should look into this’,” said Linares. Most customers – more than 90 per cent – sail through. But applying Clearspeed to the remainder delivered a 31-times return on its investment in the technology during the first year. The results were so impressive that in 2022 the insurer began looking at applying both FRISS and Clearspeed to its health insurance line, too, where the ability to analyse responses to questions such as ‘do you smoke?’ are obvious. And now it will be moving Clearspeed to the top of the funnel, to apply to all claims.

Clearspeed has not only helped identify claims that legitimately need referring to the insurer’s special investigation unit but has also saved it from paying out claims in instances not covered by the policy wording, such as a driver under a certain age being at the wheel of a car or where a driver’s licence has expired.

“In all these cases where we used to pay before we can now say to the customer ‘there is no coverage for that’,” said Linares. “One of the measurements for success for any tech is if you keep using it – and our intention is to keep using both [Clearspeed and FRISS] because they do not replace but complement each other.

Learn More About Clearspeed Insurance Solutions