AI for insurance agency workflows means using AI in the office tools an agency already has to prepare renewals, certificate requests, carrier submissions and coverage explanations for licensed staff to review. Most articles about AI for insurance agency workflows are written by companies that sell an agency management system or an add-on to one. This one starts from what an agency or broker with 50 to 300 people already has, Microsoft 365 or Google Workspace next to its AMS, and looks at four workflows where AI can prepare work for licensed staff. It then sorts out which NAIC documents address agencies and which do not.
This article is not legal advice. State insurance laws differ, so check your state’s version of any model law mentioned here.
Where agency hours go
Account managers and CSRs spend much of the week on work that is mostly reading and writing: renewal reminders and questionnaires, certificate requests, gathering client data for carrier submissions, and explaining coverage in emails. The decisions are made by licensed staff; the preparation around them is where time goes.
Most of this preparation happens outside the AMS, in email, spreadsheets and PDF documents from clients and carriers. That is why the office suite the agency already pays for is a reasonable place to start. The AMS stays the system of record; the drafts and extracted data are prepared next to it and entered by staff. It also keeps the trial reversible: if a workflow does not save time, the agency stops using the prompt and nothing in the AMS has changed.
Using AI for insurance agency workflows: four examples
Four agency workflows fit the tools an agency already has: renewal reminders, certificate requests, carrier submission prep and plain-language coverage explanations.
Renewal reminders and pre-renewal questionnaires
From the renewal list exported from your AMS, AI in Outlook or Gmail can draft a reminder for each client with the right policy names and dates, and a short questionnaire on what changed in the past year: locations, payroll, vehicles, contracts. A CSR reviews and sends.
Draft a renewal reminder for a commercial client whose general liability and property policies renew on [date]. Ask them to confirm any changes in locations, revenue, employees or vehicles since last year. Keep it under 150 words and do not mention premiums.
Certificate of insurance requests
Requests arrive by email in many formats. AI can sort them, extract the holder name, address and requirements, and prepare a draft for the person who issues certificates. Issuing stays in the AMS and with the staff authorized to do it, and any requirement that looks unusual goes to an account manager.
Submission prep for carriers
Client questionnaires and documents rarely match the format a carrier wants. AI can read the client’s answers and supporting documents and lay the data out in the order of the carrier’s application, with gaps marked. The closest example in our own work is a customs broker, where a system reads supplier invoices and fills the declaration and a declarant checks every field next to the original document before filing. The same pattern of extraction plus human check applies here.
Explaining coverage in plain language
Clients often ask what a policy covers. AI can draft a plain-language summary from the policy documents for a licensed person to review and correct before it goes out. The draft should quote the policy for anything that matters and never fill gaps on its own.
What the NAIC AI Model Bulletin covers, and who it addresses
The NAIC AI Model Bulletin covers insurers’ use of AI systems, and it is addressed to insurers only.
Addressed to insurers
The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted on December 4, 2023, is addressed to “All Insurers Licensed to Do Business In (Insert Name of Jurisdiction).” It expects insurers “to develop, implement, and maintain a written program (an “AIS Program”) for the responsible use of AI Systems,” covering areas such as “product development and design, marketing, use, underwriting, rating and pricing, case management, claim administration and payment, and fraud detection.” The bulletin does not use the word producer.
Where it has been adopted
The NAIC adoption map, with status as of November 3, 2026, lists 29 jurisdictions that have adopted the bulletin: Alaska, Arkansas, Connecticut, Delaware, the District of Columbia, Hawaii, Illinois, Iowa, Kentucky, Maryland, Massachusetts, Michigan, Mississippi, Nebraska, Nevada, New Hampshire, New Jersey, North Carolina, North Dakota, Oklahoma, Pennsylvania, Puerto Rico, Rhode Island, South Carolina, Vermont, Virginia, Washington, West Virginia and Wisconsin. California (Bulletin 2022-5), Colorado (3 CCR 702-10), New York (Circular Letter No. 7, July 11, 2024) and Texas (Bulletin B-0036-20) have their own rules.
Why carriers may ask agencies questions anyway
The bulletin defines a third party as “an organization other than the Insurer that provides services, data, or other resources related to AI,” and section 4.2 asks insurers, where appropriate, to include audit rights in contracts with third parties and an obligation to “cooperate with the Insurer with regard to regulatory inquiries and investigations.” Our reading is that an agency using AI in work it performs for a carrier may receive questions from that carrier as part of the carrier’s own program.
The data security rule that does apply to agencies
The NAIC Insurance Data Security Model Law (#668) applies to every “Licensee,” defined as “any Person licensed, authorized to operate, or registered, or required to be licensed, authorized, or registered pursuant to the insurance laws of this State.” Producers are licensed under state insurance laws, so on our reading of the definition agencies and brokers fall within it where a state has adopted the law.
The model law requires each licensee to “develop, implement, and maintain a comprehensive written Information Security Program” and says “A Licensee shall exercise due diligence in selecting its Third-Party Service Provider.” A “Licensee with fewer than ten employees, including any independent contractors, is exempt from Section 4,” the section on the security program. An agency of 50 or more people is above that threshold.
An AI tool that processes client information is a third-party service provider in the ordinary sense, so it belongs in the same due diligence as any other vendor. Check whether and how your state has adopted the model law.
Running a first workflow for four weeks
Pick the workflow that repeats most often in your agency. For many teams that is either renewals or certificate requests, but your own volume decides. Then keep the trial small and written down.
- Write the current steps of the workflow on one page, with who does each step and in which system.
- Mark the steps AI will prepare and the person who reviews each draft before it goes out or into the AMS.
- Collect five to ten past examples, with client details removed, to test the prompts before using them on live work.
- Run the workflow with two or three people for four weeks, keeping a short log of corrections the reviewers had to make.
- At the end, compare the time per item with your notes from before, read the correction log, and decide whether to extend, adjust or stop.
The correction log matters more than the time saved. If reviewers keep fixing the same thing, such as a missing additional insured requirement or a wrong policy form name, the prompt or the source data needs to change before the workflow scales.
Questions to ask about any AI tool you add
Since the tool is a third-party service provider under the model law’s logic, the due diligence questions are the ones you would ask any vendor that touches client data: where the data is stored, whether it is used to train models, who at the vendor can access it, how long it is kept, and what happens when the contract ends. Microsoft and Google publish answers to some of these for their AI features; read the current pages for the plan you use and keep a copy with your security program.
What stays with licensed staff
Licensed staff remain responsible for what clients and carriers receive, whatever tool prepared the draft. In practice that means these steps stay with them:
- advice on coverage and limits
- binding, issuing and endorsing in the AMS
- any statement to a client about what a policy covers
- decisions on claims communications
AI prepares drafts and data for these steps. A licensed person reads, corrects and sends. To see whether one of these workflows is worth it in your agency, start with the one that repeats most often, as described in Choosing the first process to review. A review takes about three hours of your team’s time, as explained in How much time a review takes.
Frequently asked questions
How are independent insurance agencies using AI?
Common starting points are renewal reminders and questionnaires, sorting certificate requests, preparing carrier submissions from client data, and drafting plain-language coverage summaries, with licensed staff reviewing everything before it is sent or issued.
Does the NAIC AI bulletin apply to insurance agents?
The NAIC Model Bulletin is addressed to insurers and does not mention producers. It does ask insurers to manage third parties that provide AI-related services, so carriers may ask agencies questions.
Which states have adopted the NAIC AI Model Bulletin?
As of November 3, 2026 the NAIC lists 29 jurisdictions, including 27 states, the District of Columbia and Puerto Rico. California, Colorado, New York and Texas have their own rules.
Can AI issue certificates of insurance?
AI can sort requests and prepare drafts. Issuing stays in the agency management system and with the staff authorized to do it.
Do insurance agencies need a written information security program?
Under the NAIC Insurance Data Security Model Law, as adopted by a state, each licensee must maintain a comprehensive written information security program, with an exemption from that section for licensees with fewer than ten employees. Check your state’s version.
Can we use ChatGPT with client policy information?
Treat any AI tool as a third-party service provider: check its terms, include it in your vendor due diligence and your written security program, and keep client information out of tools that have not passed that review.
Sources
- NAIC Model Bulletin, Use of Artificial Intelligence Systems by Insurers, adopted December 4, 2023 (PDF)
- NAIC, Implementation of the NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, status as of November 3, 2026 (PDF)
- NAIC Insurance Data Security Model Law (#668) (PDF)
This article is general information. It is not legal or tax advice. For your own situation, ask your counsel, your tax adviser or the professional body that governs your practice.