Skip to content


Claims handlers can be overwhelmed by decisions. This is due, in part, to the explosion of data — everything from photos to handwritten notes — that now accompanies claims. It takes a long time to evaluate this data, but not all of the accompanying information is relevant.

It’s hard to ask claims handlers how to separate good data from bad, but many insurers don’t have any other way to do it. Instead, the industry uses decades-old rules-based systems in an attempt to deliver repeatable outcomes. These processes often miss important subtleties, such as:

  • How should the claim process be personalized based on claims detail? 
  • Is this a claim that requires different support based on complexity? 
  • Does the detail on a document match the data related to a claim? 
  • Are there opportunities to optimize total recovery based on 3rd party evaluations?


AI delivers the power to evaluate claims data based on context. This ability has the potential to streamline the claims process in a fast, scalable, and repeatable manner.

What does context mean for claims?
Not every claim is the same. Rules-based processes attempt to deal with this fact by sorting claims into different categories, but some – perhaps most – defy categorization. Without AI, specifically AI that considers the context of claims information, the only way to process these claims is via human intervention.

How does context work in the claims environment?

Consider a property claim where the claimant includes an invoice for water mitigation services related to a flood in the basement. A rules-based system might not include document recognition, so the claim gets escalated to a human claims adjuster. The employee, pressed for time, sees that the description of the claim matches the service charges on the invoice, so they approve the claim and move on.

AI, on the other hand, might notice that line items on the invoice are repairs unrelated to the water loss, and that information is different from what’s described in the claim submission. In other words, this invoice is out of context and deserves further investigation. Here, contextual automation helps defray loss costs.

Conversely, imagine an OCR-based system that incorrectly interprets the wording on a handwritten form. OCR has no built-in intelligence, so the error results in an escalation to a claims adjuster. Even the best OCR might only be 99% accurate, which means that it will encounter 18 errors in 1800 characters. This results in many claims getting flagged for human review.

Meanwhile, an AI might look at that same erroneous interpretation and then understand, using the context of the surrounding form, what the word was supposed to be. This means that the form no longer needs to be escalated, and the claim can be processed without human interaction.

Contextual automation can be applied throughout the claims process. It can help with claim intake, determining whether a policyholder is eligible for compensation, and it can help determine whether a claim should be paid. Along the way, it can help answer questions like:

  • What data is missing or requires review? 
  • What detail may be relevant to the unique circumstances of each claim?  
  • What information can be given the most weight when considering the next best action?


Context prompts handlers with the complete picture that carriers need to scale operational improvements and deliver outcomes that reduce total loss costs. 

Where does context provide the greatest impact for insurers? 
At Shift, we see focus areas insurers can attack to deliver immediate value. 

Our Claims Intake Decisions product lets insurers leverage decisions to increase satisfaction and reduce costs. By identifying whether a claim contains information that’s in-context or out-of-context, the solution can flag a claim for straight-through processing or escalate to claims handlers as needed. If escalation is necessary, then Intake Decisions will be able to alert on the specific information that’s out of context and then identify the next best step, which will help handlers speed time to resolution.

Meanwhile, Claims Document Decisions is an artificial intelligence specifically trained on insurance paperwork. It’s designed to seamlessly ingest claims forms, police reports, estimates, repair bills, and more. Here, Document Decisions notices what’s missing – but equally important, it can use its contextual powers to ingest forms that other solutions might reject. This means that it can automatically extract information from a claim even if a form is misspelled or mislabeled.

By using the power of context Shift AI solutions can help reduce workloads for claims handlers, while at the same time increasing speed and accuracy. The improvement opportunity for insurers is tremendous, providing needed solutions that will help them create capacity for growth in the future. 

For more information, Shift Technology has sponsored a report from Aite-Novarica entitled, “Intelligent Decisioning for P/C Insurance: How AI Is Automating Insurance Business Processes.” Download it today!