Written for Chief underwriting officers, MGA principals, and the sponsors who have to fund a human-decided underwriting workbench around a process that is already running.
What's in this playbook
This playbook is the operational document for a human-decided underwriting workbench at an MGA. It is complementary to the Insurance AI Playbook already in this library, and is not a replacement for it. That playbook remains the industry map. When that mid-size insurer is in view, its first-case reading still stands: inventory of claims pilots first, and underwriting later, as decision-support for a named underwriter. This playbook will not reprint that map, and will not give a different reading of that insurer. It will not write the separate underwriting-transparency, MGA-submission, or pipelines-for-MGAs topics that sit later in the library plan.
What it is, instead, is a service playbook for a workbench at a mid-size MGA introduced in this document: six families of work, fourteen design tests, and a first-case path that starts with an inventory of running underwriting assists. If the tests pass, appetite checks and document prep for a named authorised underwriter are in-scope. Bind, decline and price are the authorised underwriter's decision. A recommended reading is advice, with an override recorded. An unsupervised bind, decline or price, a hit-ratio used as a baseline, a fairness threshold this playbook will publish, or a twelve-week framework, is not a first case this playbook will recommend. This playbook will not invent a hit-ratio or a loss-ratio, and will not pre-clear an unsupervised bind. A write into the policy-administration system the MGA already runs is later; we will not name a vendor architecture, and will not claim a certified connector.
How the tests and the method are meant to be used
The financial method is The Complete Guide to AI Automation ROI. This playbook will not invent an underwriting-workbench-sector return, and it will not promise a twelve-week go-live. The 3.8x average first-year ROI across client engagements is an average, not a forecast or a commitment for any individual project. ROI is measured across client engagements against a baseline agreed before work starts. It is an average, not a forecast for any individual project, and is not used as a sector benchmark. Built to SOC 2 standards is the exact wording we use; it is not a certification of the reader's system.
The illustrative firm is a new hypothetical mid-size MGA, introduced in this playbook. It is not the mid-size insurer already used in the Insurance AI Playbook. Overall 18, Emerging, Data Foundation the constraint at 2, Governance and Strategic Alignment relative strengths at 3 — those strengths are not a reason to auto-bind — with a next-step target of 24, Developing. The honest first work is to put every running underwriting assist on an inventory, with an owner, a sampling plan and a stop, and, if the tests pass, fund a bounded workbench: appetite checks and document prep for a named authorised underwriter. Bind, decline and price remain that underwriter's decision. The firm is illustrative, not a sample. The target is a next level, not a timeline, and not a promise. A companion workbook, asymilink-underwriting-workbench-playbook.xlsx, holds a family register, the fourteen tests, questions for counsel, a baseline sheet, and the same illustrative MGA. Do not wait to be Leading.
Six families of workbench work
each with an honest first case, a case that is too wide, and a gate. Each item below is the shape of the map; the full wording, honest first cases and “too wide” warnings are in the PDF.
F1
Inventory of running underwriting systems (first)
F2
Appetite checks and document prep, for a named authorised underwriter (in-scope)
F3
A recommended reading as advice, with an override recorded
F4
Bind, decline and price always the authorised underwriter’s decision, and not pre-cleared unsupervised
F5
A write into the policy-administration system the MGA already runs (later)
F6
Hosting, decided before the case is funded
The PDF gives, for each family, an honest first case, a case that is too wide, and a defined gate. The summaries above are the shape of the map, not a substitute for that wording.
Fourteen design tests (excerpt)
Fourteen design tests are applied to every case. The answers will differ. The standard does not. The full wording is in the PDF and the companion workbook.
T1–early
a named inventory; a person on every bind, decline and price; counsel asked without a date taken from this document.
Mid-tests
a refusal to recommend an unsupervised bind, or a hit-ratio used as a baseline, as a first case; a trail and a stop.
Later tests
a signed baseline on the ROI method; a write into the policy-administration system as a later design, without a named vendor architecture.
A completed test sheet is not a certificate, not a compliance score, and not a finding that a system is safe or aligned with a named regime.
Get the guide and workbook
Enter your name, work email and company. The files appear on this page immediately. A backup download link is also emailed when email delivery is available.