Ask any litigation paralegal where the hours go and the answer is records. Building a medical chronology by hand commonly takes 15 to 80 hours depending on how thick the file is. A firm carrying around 40 active cases can spend roughly $17,600 a month on manual records review, and outside services that do chronologies for you charge on the order of $300 to $800 per package. The work is essential and almost entirely mechanical, which is exactly the kind of work AI is good at drafting.
Most of the attention on this has been plaintiff-side, around demand packages. The defense side does the same volume of records work, chronologies, deposition summaries, paid-versus-incurred medical-bill analyses, with far fewer tools pointed at it. This is for the team that already has the documents and just wants the first draft to stop costing a week.
The real bottleneck is re-uploading, not the AI
Teams testing Claude on a set of medical records are usually impressed by the output. The friction is everything before it: the records live in Clio, and getting them in front of Claude means downloading each file and re-uploading it, one at a time, every time. For a single matter that is annoying. Across a caseload it is the bottleneck. The goal is not "Claude can summarize a PDF," which it can. The goal is to stop moving files by hand between the system that holds them and the tool that reads them.
What a connector does, and what it does not
This is where it pays to be precise, because it is easy to oversell. An open-source Clio connector lets Claude reach into a matter and find the records: list the documents, filter to the right ones, pull the metadata. That removes the manual hunting and the re-uploading.
What the connector does not do on its own is turn a scanned medical record into a finished chronology. Records are often image PDFs; they have to be fetched, run through text extraction, fed to Claude with a consistent prompt, and the result written back into the matter. That sequence, document to clean text to structured chronology to Clio, is a pipeline that gets built for the firm. The connector surfaces the records; the chronology workflow is the engineering on top. Anyone who tells you it is a single button has not built one.
The three workflows that share one shape
Once the pipeline exists, the same shape covers the three jobs that eat the most time:
- Medical chronology. Pull the records for a matter, produce a dated, sourced chronology with treatment, providers, and gaps flagged.
- Deposition summary. Take a long transcript and produce a structured summary with citations back to page and line.
- Paid-versus-incurred. Read the bills and produce the comparison spreadsheet, instead of a paralegal keying it in.
Each is a fixed workflow over documents the firm already has. Build the pipeline once and it runs across all three.
Keeping medical records and work product safe
Records and the analysis built from them are sensitive, so the data path is not optional:
- Zero data retention. Run the workflow on a configuration where the model provider retains nothing after the response. See our breakdown of how that differs across Claude plans.
- An audit trail. Record every document the workflow reads and every result it writes back.
- A mandatory attorney review. The output is a first draft. A person checks it before it is used, which is both good practice and what professional-responsibility guidance expects.
One nuance for defense firms: whether HIPAA applies depends on who your client is. If you represent an insurer or insured that is a covered entity, your firm is usually a HIPAA business associate, so a business associate agreement and safeguarded handling are in scope; if you represent a defendant that is not a covered entity and simply received records in discovery, a BAA may not be required. Confirm with counsel either way. The zero-data-retention posture is the right call regardless. The point is to be accurate about why the setup is safe, not to wave the word compliant around.
Where Oktopeak fits
We build legal document workflows for regulated and litigation firms, on top of the Clio API. If you want to run the connector yourself, it is open source and free. If you want the chronology pipeline built, parsing, a tuned prompt, write-back to Clio, the review step, and the right data path, our Guided Setup starts at $1,700 and a full custom build runs $17,000 to $53,000, scoped per firm after a free discovery call. See how the Clio integration works.
