How Elysian uses Reducto to Review Insurance Claims 16x Faster
“Document processing was one of the the foundational problems we had to solve and build upon” - Dylan Hanson, Founding Engineer at Elysian
The challenge: drowning in documents
Insurance adjusters, particularly those handling commercial claims, are drowning in work. Dylan Hanson, Founding Engineer at Elysian, describes their reality:
"The average commercial insurance adjuster is overseeing 150-200 claims at a time, and that number is ticking up across the industry. Each claim can contain hundreds to thousands of attachments and they grow daily. And if they miss anything, the consequences are in the six-seven figures."
Every insurance claim is a puzzle of scattered evidence—policies, incident reports, medical records, photos, repair estimates, legal correspondence, expert evaluations—arriving at different times and in inconsistent formats. The average claim has over 5400 pages of dense information to go through.
Insurance adjusters must synthesize all this to determine coverage, assess liability, spot fraud, and negotiate settlements. But with hundreds of active files, each with tens of thousands of pages, the constant context switching inevitably overwhelms even the best. Critical details get missed and investigations drag on for months.
The outcome is a broken system where claimants, customers, and adjusters are all frustrated. As Dylan puts it: “It’s terrible all the way down the stack.”
Enter Elysian: Results-as-a-Service (RaaS) for complex claims
Founded by a team with deep insurance and technical expertise, Elysian saw an opportunity to build a better solution with AI and a modern technology stack. As an AI-Native Third-Party Administrator (TPA) for commercial claims, they built a platform for claims handling with agentic automation, industry-leading analytics, and compliance management.
“We are a service provider that builds technology to expedite our own claims handling, specifically the high-complexity, high-dollar claims, typically in commercial liability.”
Rather than building another point solution in a fragmented market, Elysian’s engineers built their platform from the ground up working alongside their team of veteran claims adjusters. This human plus AI solution was the first step in breaking through a conservative industry that has historically been hostile to new technology.
The document intelligence challenge
As Dylan explains, document processing was "the foundational thing we built upon"—not just a feature, but the baseline challenge to be solved. Insurance documents present unique complexities: multi-hundred-page policies with wildly varying structures, entire hard-drives of poorly scanned documents, and decades-old faxes that may be referenced as evidence on a daily basis.
The regulatory environment adds another layer of complexity, demanding complete auditability where every AI decision must be traceable back to its exact source. The system needs to process massive volumes while maintaining perfect accuracy and transparency.
Examples of a real claim file Reducto parses, with scans, forms, and embedded images
Finding the right vendor
The Elysian team evaluated every major document processing solution. Microsoft Azure's Document Intelligence emerged as the best of the traditional cloud providers, but still fell short for production needs.
They experimented with other products as vision models evolved, but ultimately found them unreliable for their requirements. Reducto came through a Y Combinator connection who mentioned that many startups in their batch had converged on them for document processing.
After testing Reducto extensively, the team’s decision ultimately hinged on two critical production requirements: coherence and verifiability. Beyond just accurate OCR, Reducto delivered LLM-friendly structural interpretation paired with reliable bounding boxes that Elysian could use as grounding provenance for their citation system.
Innovation from internal needs: Claims intelligence engine
Aside from their claims handling platform, Elysian’s next breakthrough product emerged from solving their own operational challenges. As a results-as-a-service (RaaS) company, they needed to rigorously track and analyze the performance of their own adjusters:
“We needed to know, at any given time, exactly how our adjusters were doing—whether they were following best practices, and how that was impacting claim outcomes. Every week, every day, overnight, we wanted to know immediately.”
To make this possible, Elysian built AI agents on top of their core claims platform. These agents monitor adjuster activity and performance in real time, extracting bespoke data points from every claim and tracking adherence to best practices as claims are handled. When the team demonstrated the analytics platform to prospective clients, the reaction was immediate:
"The customers started being like, wait, we want that. I want that on my claims, even if I don't outsource claims handling."
This customer feedback transformed Elysian’s internal analytics engine into a standalone offering around qualitative claim review. Now, the platform enables real-time analytics across entire claim portfolios—providing both granular insights for claims handling and structured, grounded data that can be overaged by claim adjusters and underwriting teams. This closes a longstanding gap between risk pricing and real-world risk actualization, especially in high-complexity claims, by delivering actionable feedback to adjusters as claims unfold.
Elysian’s approach enables claims to be analyzed and “audited” up to 16x faster than traditional methods, allowing customers to rapidly identify trends, improve adjuster performance, and make more informed underwriting decisions. The result is a level of operational transparency and feedback that has never before been available in commercial claims.
The road ahead: unifying claims and underwriting with AI
With their recent seed funding, Elysian is moving well beyond claims automation. The next step: integrating their claims analytics with underwriting modeling to close the loop between claims, risk, and pricing.
This end-to-end approach—connecting unstructured qualitative claim evidence directly to underwriting decisions—hasn’t been possible until now, largely due to the complexity of messy, unstructured data. Elysian’s technical foundation in AI-driven claims analysis, powered by Reducto’s ground-level mastery of the messiest documents, makes it possible to bridge this gap for the first time.
Elysian's story goes beyond insurance. When companies stop fighting their documents and start building alongside them with AI, they don't just solve problems—they transform industries.
If you’re curious how Reducto could process documents for your business, get started for free or book a demo to see how we’d do on your actual documents.
