My full submission: one new AI product for your insurance clients, and three high-leverage upgrades to your onboarding system. The full argument in 90 seconds. The full document below it.
Your onboarding system is built on the right idea - AI generates, automation orchestrates, humans refine. Keep 80% of it. The bottleneck isn't the prompts. It's the intake.
The Instant Policy Review Engine: prospects upload their declarations page, AI delivers a branded coverage review in minutes - and the system captures their renewal date, the most valuable data point in insurance marketing.
Structured intake with client confirmation. Analysis-first generation so the three deliverables never contradict each other. Two early human gates plus a learning loop, so the system gets better with every client.
Everything below is the evidence. Every claim about American Safeguard Insurance came from actually researching them - judge the system by its output.
American Safeguard's own homepage says it: 'Send your declarations page and we'll review it with you.' Today that's manual and slow. This makes their best offer instant, scalable, and measurable—while capturing the Renewal Date, the most valuable data point in insurance marketing.
Five steps, zero new tools - GHL, Claude, Make, ClickUp, Claude Design.
Automated reviews today build a massive, perfectly-timed sales pipeline for tomorrow.
Insurance prospects are only truly buyable in the 30-45 days before their policy renews - and most agencies have no idea when that is. Every uploaded dec page tells the system exactly when this prospect becomes winnable. The funnel doesn't just generate a lead today - it schedules a perfectly timed sales conversation months from now, automatically. The database compounds.
AI drafts, a licensed agent approves - nothing reaches a prospect without human sign-off. Documents stay in the client's own GHL sub-account. Every report carries clear 'review, not advice' language. E&O-safe by design.
Only three parts are client-specific: brand kit, rules library, funnel skin - a GHL snapshot deployable in under a day. Setup fee plus monthly license, per insurance client. DCM improves it once, everyone benefits.
What looks solid: your dwelling and liability limits are in line with homes like yours in Faulkner County.
Three things worth a conversation: your deductible may be costing premium without real benefit — we don't see flood coverage listed, and parts of Conway sit in zones standard policies won't respond to — and your personal property limit hasn't changed since 2021.
Scattered notes, delayed research, slow drafting, expensive revisions.
Structured data, instant AI synthesis, human strategic review, immediate delivery.
Every sales call transcript, policy detail, risk profile, and Voice of Customer insight is structured into a central JSON record. AI doesn't guess—it queries the Brain.
The red boxes are the only two places a human touches the process before delivery - one cheap gate early beats one expensive gate late.
Fix the data capture first. Record the sales call, merge everything into one Client Brain record, and have the client confirm a 'What We Heard' summary in 5 minutes - before any research runs.
A staged pipeline, not a mega-prompt. Gemini collects, Opus synthesizes, Sonnet drafts. Structured JSON between every stage.
Review mining for VOC, FB Ads Library for competitor reality, Search Atlas for keyword gaps. Every claim carries a source.
The market analysis generates first and becomes the single source of truth. The blueprint and creative brief derive from it - three documents that can never contradict each other.
Make.com orchestrates end to end. Exactly two human handoffs: a 10-minute research review before generation, and approve/annotate after drafting. No silent failures.
Source-or-flag rule, schema validation, fact-table check, and a fast brand-lint pass.
Compute scales trivially - review hours and Make ops budget are what break. Approve/annotate review, an industry template library, cost-per-onboarding measured from day one, and the learning loop below.
Reviewer edits are logged and folded back into the prompts monthly - client #40's first draft is better than client #4's, so review time falls exactly as volume rises.
I'd verify how far its API goes inside a Make scenario. Fallback: Canva templates populated by structured output - proven, less elegant.
I'd run three pilot onboardings and time every stage before committing publicly. An estimate isn't a baseline.
Video belongs in fulfillment, not the signature-to-strategy window. Forcing every tool into every workflow is how stacks get slow.
Depends on how many insurance and adjacent clients sit in your portfolio - I'd want that count before projecting revenue.
Everything in this submission was designed for that window - the intake confirmation builds trust on day zero, the early gate protects speed, and the learning loop means the system gets better every month you run it.
Marjohn Robillo - GHL Systems Architect and AI Automation Specialist. 300+ GHL sub-accounts built. Automation and revenue operations background.
Download the full proposal (PDF)This page was designed, written, and shipped on GoHighLevel in one working session - the same platform your clients are served on.
© 2026 Marjohn Robillo · marjohnrobillo.me/davecreekmedia