Claims and service teams work with names, policy numbers and plates all day. Nymgate tokenizes them before the model sees them.
On 14 October ⟦PERSON:7F31:0001⟧'s vehicle ⟦PLATE:7F31:0003⟧ was hit from behind at a junction. The third party admits fault. Repair estimate £3,180, payable to ⟦IBAN:7F31:0004⟧.
From claim to answer
The handler pastes the file into ChatGPT.
Names, policy numbers, plates and IBANs.
A timeline, a liability note, a reply.
Tokens are restored on their screen only.
Your formats
| Rule | Pattern | Example |
|---|---|---|
| POLICY_NO | P-\d{2}-\d{7} | P-22-0091834 → ⟦POLICY_NO:7F31:0001⟧ |
| CLAIM_ID | CL-\d{4}-\d{4} | CL-2026-0912 → ⟦CLAIM_ID:7F31:0001⟧ |
| PLATE | [A-Z]{2}\d{2} [A-Z]{3} | AB12 CDE → ⟦PLATE:7F31:0001⟧ |
Rules live in your config and reload without a restart. Try a rule in the demo →