Directions Brain

Broker Operational Tech Stack

4 min read

Broker Operational Tech Stack

Definition

The concrete software an independent or mid-regional US insurance agency runs day to day, layer by layer, and — more usefully — where each layer stops working. Built from the first two direct broker interviews (2026-08-17), this page exists to answer guy-barkat’s standing action item from 2026-08-09-directions-operational-ai-and-axiom-challenge: which AMS/CRM systems do brokers actually run on, and what are their capability gaps? It is the operator-grounded counterpart to the vendor-side insurance-vendor-diligence-tracker.

Key points

Layer 1 — Agency management system (system of record). vertafore QQ Catalyst at greg-ehly’s three-broker agency; ezlynx (Applied-owned) doubling as CRM at jim-coronado’s mid-regional. The AMS holds customers and policies, generates COIs and ACORDs, and reports new business by period/carrier/line. It is genuinely load-bearing — but only for the half of the book it can see.

Layer 2 — Carrier download (the automation that exists). A daily “download” pull from direct-appointment carriers brings in new policies, renewals, claims, and cancellation notices automatically. This is real, working automation, and it is why brokers do not experience direct-appointment business as painful.

Layer 3 — The manual half. Anything placed through a wholesaler does not download. Customer name, email, DOB, and policy data are re-keyed by hand solely so the AMS can produce COIs and ACORDs. For Greg this is ~50% of the SMB commercial book — precisely the complex risks (manufacturing, pharma, technology). The AMS gap is not a feature gap; it is a coverage gap over the distribution channel.

Layer 4 — Rating. Personal lines are solved: one intake through ezlynx out to ~30 carriers, ~30 quotes back. Commercial has no comparative layer in either agency. Jim’s agency has the capability available and declined it on price-vs-volume grounds; Greg’s simply quotes 2–3 direct carriers by hand. Two independent agencies, same absence.

Layer 5 — Marketing/retention, and it is already an AI product. levitate syncs to Greg’s AMS and runs book segmentation, AI-generated campaigns, the agency website and SEO, inbound lead routing to Gmail, and the full post-bind cadence (3-day welcome, 10-day check-in, ~21-day review ask, monthly tips). Jim’s agency has nothing equivalent — his post-bind cadence is a personal quarterly email. So this layer is neither universally solved nor universally open; it is contested, with a working incumbent.

The seam that no layer covers. Nothing in either stack tells the broker what the prospect currently has. Existing coverage arrives as a texted or emailed dec page, and a broker cannot see a policy written by another broker even at a carrier they themselves sell. This is the operator-side confirmation of policy-ingestion-requires-a-data-rail and the reason the vendor layer (canopy-connect, insurgrid) exists at all — neither interviewed broker uses one.

The other seam: appetite. Carrier portals accept a full submission and decline only on the final screen, so routing runs on tacit knowledge and phone calls to marketing reps. No software layer holds this. See carrier-appetite-knowledge-is-the-broker-bottleneck.

Evidence

Open questions

  • Two agencies is not a sample. Is the wholesaler-doesn’t-download gap universal, or a Vertafore-specific limitation? Check against applied-systems Epic and momentum-amp.
  • Is commercial rating genuinely absent at small-agency scale, or merely priced out? ezlynx commercial pricing is the fastest way to find out.
  • Does levitate’s presence in one agency and absence in the other track agency size, owner preference, or vintage?
  • Neither broker uses a policy-ingestion vendor. Is that a cost decision, an awareness gap, or evidence the tools don’t fit the commercial workflow?

Graph View

Backlinks