Understand. Detect. Act.

Three products, one chain of evidence. Isomer™ Graph understands every communication, Isomer Signals detects the risk in it, and Isomer Actions sets out and runs the next best actions. Start with Signals.

Claim graph connecting messages, documents, and parties

Every step cites the one before it, back to the page.

One claim, from the message to the work. Graph finds the facts, Signals finds the risk, and Actions sets out what to do next.

Slip and fall at a grocery store · General liability1 email · 2 attachments · 1 task
  1. The messageGraph

    Email from the store managerIncident report and vendor agreement attached
    Incident report · p. 2“Entrance mat curled at the edge.”
    Service agreement · p. 6“Store named as additional insured.”
  2. The factsGraph

    Mat vendor, a party to the claim
    Vendor contract, with the store as additional insured

    Each fact cites its page.

  3. The signalSignals

    HighVendor contract names the store as additional insured

    Cites both facts.

  4. The action planActions

    Tender the claim to the vendor’s carrierR. Patel · due Oct 22

    Cites the signal and the Contractual Risk Transfer Guidelines.

  5. The recordActions

    Task and diary date written to your claim system

    Approved by R. Patel, with an audit trail.

Read-only in. Structured results out.

  • Reads from Microsoft 365 or Google Workspace by scoped OAuth, mail forwarding, API, or a sovereign install in your tenant. Past mail there builds the claim graph from day one; adjuster inboxes and the claim system keep it current after the claim is set up.
  • Writes to the handler, a daily report for claims leadership, a webhook, or notes and tasks in your claim system where that’s configured.
  • Builds on what you already run. No claim-system project or schema work to start, and a short briefing for handlers.
  • Not a claim system. It writes back to yours.
  • Not another model on your fields. It reads the documents the fields leave out, as they arrive.
  • Not an assistant you have to ask. It reads every document on every claim, without anyone asking.
  • Not a black box. Every signal cites its page.
  • Not autonomous. Your rules set what runs on its own; everything else goes to a person.

Start with a workshop, on claims like yours.

Your team learns where risk hides in the claim file and how AI reads it, builds synthetic claims modeled on your book, and sees Isomer find the risk in them. You leave with a business case for your book. No data access needed.