Ghostwriter AI
Trained on your agency's own voice, Ghostwriter drafts renewal communications, coverage explainers, and proposal narratives that clients actually read.
[ capabilities ]
- —Voice profile built from your existing correspondence
- —Renewal, proposal, and claims-update templates
- —Human review queue before anything sends
[ core systems & integration architecture ]
Pulls policy and account context for the renewal, drafts the communication, and parks it in a human review queue inside your existing system.
Applied Epic
Native REST APIPolicy and contact reads; draft saved as an Epic attachment
Vertafore AMS360
Native REST APIOrange Partner APIs for policy detail and note write-back
Guidewire PolicyCenter
Native REST APIRenewal job reads for coverage and premium movement
Microsoft 365
Embedded SidecarDraft lands in the producer's Outlook drafts folder
Sync model
Async — drafts generated ahead of the renewal window
Standards spoken
Time to connect
Days
Enterprise guardrails
- —No customer data used to train models — zero-retention inference only
- —OAuth 2.0 / mTLS against your existing identity provider
- —Field-level audit trail on every value written back
- —Nothing sends without a named human approving the draft
[ mcp microservice interface ]
This module ships as an MCP microservice — ghostwriter-mcp. It consumes governed tools from the core-system adapters and publishes its own, so your assistant, orchestrator, or another module can call it directly. No credentials ever reach the model runtime, and every write lands in the human gate.
Consumes from core systems
- epic_get_policy_schedule
- ams_get_policy_detail
- gw_get_policy_period
Provides to other agents
draft_renewal_letterreadDraft a renewal communication from live policy movement
draft_coverage_explainerreadExplain a coverage change in client-ready language
park_draft_for_reviewwrite · human gateSave the draft to the producer's review queue
[ implementation playbook ]
How an agency actually stands Ghostwriter AI up — read-first, human-gated, and measured from day one.
Setup, in order
- 1
Gather 10 examples of your best renewal letters and client emails — the module learns your voice from real work, not templates.
- 2
Define the three communication types to start with: renewal cover letters, rate-increase explanations, and cross-sell introductions.
- 3
Connect read access to client and policy data so drafts pull real names, dates, premiums, and coverage details.
- 4
Set the review rule: every draft goes out under a human's name after a human reads it. No exceptions in the first 90 days.
- 5
Pilot with one account manager for two weeks, editing drafts freely — the edits teach the module your house style.
- 6
Roll out to the service team with a shared prompt library for the three communication types.
Agency use cases
Renewal letters at scale
Ninety days before renewal, the module drafts a personalized letter per account referencing the client's actual policies, claims history, and any rate movement — the account manager reviews and sends.
The hard rate-increase conversation
For accounts facing a 15%+ increase, the module drafts an explanation that pairs the increase with the client's loss history and market context, so the producer starts from a defensible position.
Cross-sell introductions that don't feel canned
When the book shows a commercial client with no EPLI or cyber, the module drafts a short, specific introduction referencing their industry and headcount — not a generic blast.
The first win to aim for
Renewal-letter prep time per account drops from ~45 minutes to under 10, with zero client-facing quality complaints.
Guidance by agency role
Producers
Add the relationship context only you know, especially around rate conversations and cross-sell timing, then approve language before it reaches the client.
Account managers
Treat drafts as a strong first pass: verify names, dates, premiums, and coverage details, and feed useful edits back into the shared prompt library.
Agency administrators
Maintain approved examples, disclosure language, and review rules; sample sent communications monthly for accuracy, tone, and measurable time saved.
[ built for ]
[ where this sits in the vendor market ]
We recommend buying first wherever the market already solves a problem well. Here is what you can buy instead of building this — and where this module earns its place beside what you buy.
No product in the directory covers this yet
That is why Ghostwriter AI exists — the market has not packaged this work, so the module is the practical way to get it.