Research and guides for claims leaders.
Longer reads from the Isomer™ team on claims risk, how the products work, and putting AI to work on claims.
Research
Judgment at scale: what a full-file review catches
Claims professionals and the signal engine reviewed the same 700 claim submissions at a large commercial carrier. What both caught, and what sat deep in the attachments.
Guide
12 questions to ask before you put AI on your claims
What to ask any vendor about accuracy, explainability, security, and who approves what.
Paper
Isomer Graph: every page of every claim, understood
How emails, faxes, calls, and PDFs become one claim graph, with every fact traced to the page it came from and nothing thrown away.
Paper
Isomer Signals: 90+ detectors, cited to the page
How the signal library finds time-limited demands, representation, and severity as each message arrives, and how every finding is cited and reviewed.
Research
Models alone, or models with the claim graph
The same models on the same claims, with and without Isomer Graph: what the graph adds to accuracy, citations, and consistency.
Research
Model performance on claims work, compared
Frontier and open models side by side on real claims tasks, from classification to extraction to matching: accuracy, speed, and cost.
Research
Which AI model for which job?
Five models on six jobs from a live production pipeline, from deciding and reading to judging and writing, scored for accuracy against cost.
Paper
Sovereign deployment: Isomer in your own Azure tenant
What runs where, who holds the keys and the models, and how the managed application is installed, reviewed, and updated.
Get the papers as they’re published.
Choose the papers you want and tell us where to send them. Each one arrives by email when it’s ready.
Prefer a conversation?
Thirty minutes with our team on what Isomer would find on your book.
