GARNET · A RED SUMMIT PRODUCT

AI assistant inside PolicyCenter.

Garnet reads the policy on screen, searches your documents, and answers with the source referenced.

Agents and underwriters get answers in the moment, keeping customer and broker conversations moving without holds or unnecessary referrals.

Built for mid-sized P&C carriers and mutuals.
Live in under two months.

Faster underwriting cycles

Underwriters have real-time information at hand to move through their workflows faster.

Consistent underwriting

Give every underwriter access to the same trusted guidance, regardless of experience level.

Faster staff onboarding

New staff ramp up with an AI assistant guiding them through PolicyCenter as they work.

See Garnet in action

Example: with screen context on, you ask Garnet about the standard deductible for a private passenger vehicle under $80,000 and it answers $500, notes that the 2022 Toyota RAV4 on screen falls in this band, and cites the Auto Rate Manual. With screen context off, Garnet cannot see the screen, so it answers from your approved documents only and asks you to paste the vehicle details instead. The library icon lists the documents loaded into Garnet: underwriting guidelines, rate manuals, bulletins, and procedures.

  • Reads the policy open on your screen
  • Answers with a citation to the exact section
  • Click any citation to verify it in the source document
  • New manuals join the library, regression tested

Chrome-based extension

Garnet overlays PolicyCenter in the browser. No PolicyCenter configuration is required or changed.

Your documents are secure

Your underwriters' questions and your documents are never used to train models, and PII stays protected.

Production-ready solution

Built with the safeguards, monitoring, and alerting an enterprise environment expects, and the scalability to serve your whole underwriting team.

HOW TO START

Three steps to production.

20-minute demo

2-week pilot

2-months to production

0120 MINUTES

Demo

Live on your screen. Garnet answers real underwriting questions against a sample library, every answer cited. We close by discussing how to get you started.

02TWO WEEKS · YOUR DOCUMENTS

Pilot

Evaluate Garnet on your own documents before any commitment. Your subject matter experts spend two weeks asking the questions they already know the answers to.

The output of this step is not a purchase. It is trust.

03 · 2 MONTHS

Implementation

We align on the documents Garnet needs for your use case, then stand it up with the security review, safeguards, and monitoring a production rollout requires. Your team tests until they trust every answer.

A Canadian mutual insurer has walked this exact path and is now proceeding to full implementation. Case study coming soon.

Osman Minhas
WHO YOU WILL WORK WITH

Osman Minhas

P&C insurance technology · Guidewire program leadership

Close to ten years in P&C insurance technology transformation. First at EY in their Guidewire delivery practice, then on the carrier side, running these programs from within: alongside the policy, claims, billing, and technology teams who live with the outcome, and the CIOs who own the decision. Then I built Garnet.

An AI solution should solve real problems, cost less, and reach production fast. I make sure that shows up where you can measure it: the results, the invoice, and the timeline.

WAYS TO WORK TOGETHER

Technical Program Delivery

Full accountability to execute your technology initiative.

  • Guidewire Cloud implementations and upgrades
  • Legacy policy system migration
  • M&A technology integration

CIO Advisory

Counsel for P&C technology leaders making decisions they only get to make once.

  • Cloud upgrade strategy and sequencing
  • SI selection and contract structuring
  • Program health checks and course correction
  • Board-ready business cases and spend defence

Garnet

An AI assistant that answers underwriting and system questions inside PolicyCenter.

  • Embedded in the PolicyCenter UI
  • Pilot to production in weeks
  • Fully managed, no internal AI team required
More about my background →