In personal injury law, medical ledgers are the record of everything that matters. They document treatment histories, establish injury severity, and quantify the financial damages at the center of every case.
Those medical ledgers arrive from health insurance and automobile insurance carriers as multi-page PDFs, each formatted differently depending on the insurer. In some workers' compensation cases, they still arrive as scanned legacy printouts before they can enter the workflow at all. Someone on the legal team has to open every document, read every row, and manually transfer that data into a spreadsheet before it can be entered into the firm's case management system.
At firms carrying hundreds of active cases, that cycle repeats constantly. Legal assistants spend significant hours on document transcription.
A law firm came to us with exactly this problem. We built them an agent that reads incoming PDF ledgers and converts them into clean, structured spreadsheets automatically.
Why Subrogation Data Entry Consumes Legal Assistant Capacity
Subrogation is the process by which an insurer or medical provider seeks reimbursement from a settlement or judgment. In personal injury cases, that means incoming medical ledgers that document treatment charges, dates, procedure codes, and balances owed.
The data exists in a format that legal case management systems can't use directly.
Legal case management platforms, built specifically for law firms, usually do not ingest unstructured PDF data automatically, meaning that every medical ledger's information needs to be uploaded manually.
The manual process looked like this: a legal assistant opens a multi-page PDF, reads each row, and types the data into Excel column by column. Then they upload that Excel file into the case management platform.
Every insurer formats their ledgers differently. Column headers shift. Some include ICD codes, some include CPT codes, some include both. Some span across multiple pages in ways that break the visual flow of the data.
The data entry work is necessary, but it was also deeply inefficient and time-consuming.
How the Subrogation Data Extractor Agent Works
Before building the agent for our client, we used SoftSnow's AI Opportunity Matrix™ to assess which workflows in the law firm were the strongest candidates for automation. The AI Opportunity Matrix™ is a structured framework that maps a firm's operations against two dimensions: the potential impact of automating a given workflow, and the feasibility of building the solution. This workflow scored at the top of both: high volume and significant time cost, with a process structured enough to automate reliably. Estimated time savings once automated: 100+ hours per month across the full caseload.
SoftSnow built the Subrogation Data Extractor Agent on Cassidy for exactly this problem.
The agent does one job with precision:
- Input: A PDF subrogation ledger, scanned or digital, uploaded via drag-and-drop or browsed from the Cassidy knowledge base
- Process: The agent reads the full document, handling multi-page data as a single continuous dataset. Each insurer's records are organized into their own tab.
- Output: A clean, structured spreadsheet with all entries properly organized.
Two practical requirements determine whether the agent works correctly. First, the PDF needs to be machine-readable. To test it, a legal assistant just needs to check if they can highlight text in the document; then the agent will process it accurately. Second, large files need to be split into smaller chunks and processed using Cassidy's Bulk Run feature, which handles the platform's context window limits without affecting output quality.
The legal assistant reviews the output for accuracy and uploads it to the legal case management system. Before that workflow can operate at scale, though, there is a prior question worth answering: what happens to the documents once they enter the system?
Data Security for AI Tools in Legal Document Processing
Law firms handle sensitive client information. The documents flowing through subrogation workflows include protected health information and case-specific financial data. Any automation tool touching that data needs to meet a higher standard than a general-purpose AI platform.
The Cassidy platform was selected for our client for this reason. Data processed through Cassidy stays siloed within the firm's workspace and is never used to train external AI models. That's a meaningful distinction when the documents in question contain protected health information and case-specific client data.
That evaluation process matters more than most firms realize at the outset. AI platforms vary significantly in how they handle data retention, model training, and access controls.
When assessing any AI tool for document-heavy workflows, these are the questions worth asking:
- Does the platform retain your data after processing?
- Can your documents be used to train external models?
- Where is data stored, and who within the vendor's organization can access it?
- Does the platform offer a dedicated workspace that keeps your firm's data separate from other customers?
For legal teams evaluating AI tools, that question belongs at the start of the conversation, before any document touches a system. Once the right platform is in place, the operational case for automation becomes straightforward.
How Personal Injury Firms Reclaim Legal Assistant Capacity
When document processing moves to the agent, the math shifts quickly. A workflow that was consuming up to 25 hours per case now takes minutes. Across a caseload requiring ledger updates every 30 to 45 days over a six-to-nine month period, that recovered time accumulates into something the firm can actually redirect.
Legal assistants can redirect that capacity toward case preparation and client communication, contributing more directly to how each case develops. Case work that depends on professional judgment and direct client relationships gets the attention it deserves.
The agent and the legal team each do what they are best suited for. The agent processes every ledger consistently, regardless of how each insurer formats their documents. The legal assistant reviews the output for accuracy, then applies their knowledge of the case to determine what the data means and what to do next.
That division is what makes the workflow worth building at scale. Firms that run their ledger workflow this way recover something more useful than processing speed: professional capacity directed at the work that needs it most.
Automate Subrogation Data Entry for Your Personal Injury Practice
Subrogation data entry is one of the most consistent operational challenges we see across personal injury and high-volume case law firms. It's also one of the most straightforward workflows to automate once the right infrastructure is in place.
The opportunity isn't always visible until you look closely at where legal assistant hours are going each month. How many active cases require ledger updates right now? How many insurers are sending documents in different formats? How much of that work is being handled manually?
Those are the questions worth asking. And they're exactly where we start.
If your firm manages a steady flow of subrogation documents, let's look at whether this approach fits your workflow. We'll review your current ledger workflow and assess what your existing systems already support, then show you what automated extraction could look like in practice.
Let's talk about what this could look like for your firm.
