Submission Triage: How It Works

You can read more about how and why we built Submission Triage in our launch announcement. In this blog, we’ll take a deeper dive into how Submission Triage works to put the right deals in front of your underwriters, instantly. 

Applying Agentic AI to Submission Intake

As we learned in the process of developing Submission Triage, a big reason that most traditional attempts to apply technology to insurance submission prioritization haven't worked is that they try to evaluate documents and data points in a vacuum. 

But when an underwriter examines a submission to determine if it’s worth pursuing, they don’t just look at a single item and make a decision. Instead, underwriters evaluate the entire context of the submission. They look at emails, SOVs, loss runs, and anything else associated with the submission, and apply their judgement to decide whether or not a submission is a good opportunity.

Submission Triage leverages agentic AI to evaluate submissions the same way underwriters do, with the full context. Submission Triage directs an LLM to identify and extract only the data within a submission necessary to assess eligibility, appetite, and winnability. Because LLMs can look at an entire submission (including documents) and use human-like logic to map and connect data between attachments and the body of an email, they can deliver a complete, highly accurate assessment of each deal in seconds. 

The LLM-based reasoning can also take factors into account like overall portfolio composition and changes in appetite as more risks are signed. This allows Submission Triage to dynamically tailor its recommendations to changes in the portfolio, and continually learn from human underwriters to improve its decisions over time. 

From Inbox to Analysis: Submission Triage in Action

As soon as an email arrives in the underwriter’s inbox, Submission Triage gets to work:

1. First, Submission Triage checks for information in the email and attached documents to determine whether the email is a new submission, or related to an ongoing deal. If the email is unrelated to submissions, the workflow terminates. 
2. Once it’s determined an email is related to a submission, Submission Triage examines all attachments and classifies them by type, such as application, SOV or loss run.
3. Submission Triage then extracts all relevant data from the attachments, with specific processes activated per attachment classification. For example, it would extract policy risks from a document classified as an application.
4. Once all the necessary data is extracted and ingested, Submission Triage determines if the submission is a good fit for your current appetite, based on your rules, guidelines, historical underwriting data, and current portfolio status.
5. If a submission is determined to be a fit, Submission Triage then predicts how likely you are to win the deal. Winnability scores leverage data from past deal outcomes, like whether similar submissions have been quoted, won, or declined.
6. Submission Triage surfaces the deals that are highest-appetite and most likely to bind for your underwriters’ first attention, and automatically declines deals that are clearly a bad fit. 

This entire process happens within just a few seconds, with minimal human intervention required. And speed is important: the faster an underwriter can deliver a quote, the more likely they are to win that submission. 

By triaging inbound deals before they get to an underwriter and ensuring that high appetite deals get attention first, not only can we significantly reduce time to quote and boost underwriter productivity, we also boost portfolio outcomes as a whole. 

When underwriters focus on high-appetite deals that will actually bind, they win more of the best business available to them. 

Learn more about how Submission Triage helps underwriters put the right deals first, or get a live demo today.