Summaries
The first draft, while the kettle is still on.
Upload the records. Oathdesk Summaries reads every page, builds a chronological treatment timeline, and drafts a demand-ready summary with every line cited to a page in the file. You review. You ship.
- Median time to first draft
- 12 min
- Page records, ingested clean
- 2,400+
- Quote traceable to source
- Every
- Off-platform training
- 0
Cited drafting
Every quote knows where it came from.
Oathdesk doesn't paraphrase a treatment note and walk away. It pulls the exact passage, attaches a page-and-paragraph citation, and shows you the source side-by-side when you hover. If the model can't cite it, the model won't write it.
Timeline
Treatment, in order, no matter how the records came in.
Two providers, three referral letters, one set of imaging, and a stack of bills out of order. Summaries reads all of it, dates it, and renders one clean chronology — the spine for your demand and your mediation brief.
How it works
Records in. Demand draft out. Nothing in between you didn't ask for.
Ingest anything
PDFs, faxed paper, handwritten provider notes, EOBs. Summaries OCRs everything and indexes by date, provider, and treatment.
Page-level citations
Hover any sentence in the draft. The page it came from opens beside it. Reviewers stay grounded.
Demand templates
Choose your firm's house template. Summaries fills the facts, the timeline, and the damages section.
Discrepancy flags
When two records disagree — pain levels, dates of injury — Summaries surfaces the conflict rather than picking a side.
Per-firm tenancy
Your matters never train a public model. S3 originals stay in your tenant. Audit logs on every prompt.
Bring your own model
Default to Oathdesk's tuned drafting model, or wire in an enterprise OpenAI / Anthropic / Bedrock endpoint your firm already buys.
Run Summaries on one of your actual matters.
Pick a closed file. Send us the records. We'll show you what a 12-minute draft looks like on something you already know cold.