CertPilot watches your certificates inbox, reads each COI request, pulls the real policy data, and fills the ACORD 25 — then leaves a finished draft for a licensed human to approve. Twenty-five minutes of CSR time becomes twenty seconds of software time.
Book a 15-minute demo We'll run it live on one of your own certificate requests.How it works
The AI reads the request and decides what's being asked. The certificate itself is filled by deterministic code from your policy data — never improvised, never retyped. Then it stops and waits for a person.
Gmail or Outlook — the address your requesters already use. Nothing changes for them.
Identifies the insured, the certificate holder, and any wording — additional insured, waiver of subrogation.
Carriers, policy numbers, limits, dates — from your records, exactly as they appear.
Every field, every checkbox, plus a reply email drafted and attached. About 20 seconds.
A licensed human reviews every certificate. Nothing leaves without approval — that's the design, not a caveat.
Guardrails
Every reply is a draft in your own Drafts folder. The send button stays human.
A go-live timestamp means it only reads messages that arrive after you turn it on. Your history stays untouched.
If it can't match the client exactly, it drafts a clarifying question instead of picking the closest name.
It runs against your inbox and your records, under access you grant and can revoke in one click. Every action lands in a timestamped log you keep.
Pricing
A fixed implementation fee plus a flat monthly service, sized to your certificate volume — one mailbox, your book of business, human approval on every send. Founding-client pricing is available for the first agencies we take on.
Common questions
Both are supported natively — Microsoft 365 and Google Workspace. Your requesters keep emailing the same address either way.
Three things change, all in your favor: certificates are built from your policy records instead of retyped by hand, a licensed person approves every send, and every certificate has a timestamped audit trail. Consistency is a defense.
No. The agent works inside accounts you own — your mailbox, your records — under a permission grant you can revoke at any time from your own Google or Microsoft security settings.
They're flagged for a human with a note about what's unclear, and logged. It skips newsletters and non-certificate mail automatically.