How to use AI in an insurance agency: a guide for owners

An insurance agency (the UK word is brokerage) can use AI to draft emails, summarise documents, write up meeting notes, prepare renewals and match statements. "AI" here means software that reads and writes text. It prepares the first version, and a named person checks it before it reaches a client or an insurer.

Key points

AI can draft emails, summarise documents, write up meeting notes and prepare renewals.
A named person checks every draft before it reaches a client or an insurer.
Four jobs need no setup: drafting emails, summarising a document, writing up meeting notes and drafting a procedure.
Read the data guide before anyone pastes a client's details into a chat window.
Never hand AI the advice, the decision to place cover or the promise made to a client.

How can AI be used in an insurance agency?

AI can be used in every area of an agency where people read, write, copy or compare. The table below goes office area by office area. For each task it shows what the AI does and who checks the result. The rule in the last column does not change: a person looks before anything goes out.

Office area	Task	What AI does	Who checks it
Everyday writing	Client emails	Writes a first draft from your notes, in plain language	The person sending it
Everyday writing	Long documents	Summarises a policy wording or a long email thread	The person who needs the answer, against the original
Everyday writing	Meeting and call notes	Turns a recording or rough notes into a list of actions	The person who was in the meeting
Everyday writing	Office procedures	Turns "how I do it" into a written, numbered procedure	The person who does the task
Service	Client requests	Reads the request and lists what is needed to act on it	The service team member handling it
Service	Policy changes	Builds the change request with the details filled in	The person who confirms with the client
Renewals	Renewal preparation	Gathers what changed and compares wording with last year	The account handler
Renewals	Quote comparison	Lines up premium, excess, limits and exclusions	The adviser who recommends
Paperwork	Issued policy checks	Reads the policy against the terms agreed	The account handler
Paperwork	Schedules	Builds one clean schedule from spreadsheets and attachments	The person who owns the account
Claims	Claim tracking	Keeps one record per claim and lists what is outstanding	The claims handler
Finance	Statement matching	Checks commission lines against your own records	Whoever runs the accounts

The excess in a quote is the part of a claim the client pays first (the US word is deductible). The management system that holds your records is a BMS in the UK and an AMS in the US.

What can an agency do with AI tomorrow, with no project?

Tomorrow an agency can use a general AI assistant for four jobs that need no setup: drafting emails, summarising a document, writing up meeting notes and drafting a procedure. A general assistant is a chat tool you type to, such as the ones from Microsoft, Google or Anthropic. All of these need one rule first: read whether ChatGPT is safe for client data before anyone pastes a client's details into a chat window.

Start with a task where no client data is needed. Ask the assistant to turn a rough list of steps into a written procedure for a new starter. Ask it to rewrite a stiff email in plain English for a client who reads English as a second language. Ask it to list the questions a client should answer before a renewal. None of those needs a name, a policy number or an address.

Microsoft, for example, says on its Copilot privacy page that "Prompts, responses, and data accessed through Microsoft Graph aren't used to train foundation LLMs." That statement, read on 5 November 2026, covers Microsoft's business Copilot product. Other makers and other plans say different things, so read the page for the plan you actually use.

What should I type first?

Type an instruction in plain words and say what you want back. That instruction is called a prompt. Here is one to try, using no client data:

Rewrite this renewal reminder in plain English for a reader who knows nothing about insurance. Keep it under 100 words and polite. Then list anything you are unsure about. Reminder: "Your policy is due for renewal. Please review your sum insured and tell us of any material changes before the renewal date, as cover may otherwise be affected."

The assistant should return a shorter version in everyday words, followed by a list of points it was unsure of. Read both against the original before you use anything.

Make the last request a habit. An assistant can state wrong things with confidence, and asking it to say what it is unsure about gives you a list of places to check. It will not catch everything, so the checking stays with you.

What should an agency never hand to AI?

An agency should never hand AI the advice, the decision to place cover or the promise made to a client. Those carry your name and your regulator's attention. The FCA, the UK financial regulator, says on its page on its approach to AI: "We do not plan to introduce extra regulations for AI." The existing rules still apply to the work, whoever or whatever prepared it.

AI also makes mistakes that look confident. It can quote a limit that is not in the document or merge two clients' details. That is why the last column of the table names a person. A check against the original document takes a minute. A wrong figure in front of a client costs far more.

Which area of the office should go first?

Go first where work repeats most often, takes real time and can be checked afterwards. For many agencies that is renewals, statement matching or policy changes. Everyday writing is the easiest starting point because a mistake is caught by the person already reading the draft.

A rough order that suits most small offices:

Use a general assistant for everyday writing with no client data, for a few weeks, so the team learns what it gets right and wrong.
Agree the data rules in writing (who may paste what, and where).
Choose one repeating job and measure how long it takes by hand today.
Run that job with AI preparing the draft and a named person checking it.
Decide whether to keep it, and who owns it from then on.

Steps three to five are the heart of our 90-day AI plan for a small agency.

What do you do next?

Next, pick one task from the table that your team groans about, and write down who does it today and how long it takes. That one page is the start of a plan. If you want help choosing between a general assistant, the AI in your management system, a single-job tool or a built system, the guide to AI tools for insurance agents sets out the questions to ask. The main guide to AI for insurance brokers links all the jobs in the table to their own pages.

If your first pick is a repeating office job, these pages show what it looks like when a system prepares the draft: preparing renewals and matching insurer statements.

Where Navicade fits

Navicade builds and runs AI systems that do work the office does by hand. We start with a free 30-minute call and a free operations map of where your hours go. Then we take on one agreed task at a time. Results arrive in your existing tools. Insurance decisions stay with your team. The system checks its work on every run and alerts us.

A beginner may not need us yet. Everyday writing with a general assistant needs no outside help. If a repeating job is eating hours, you can ask for the free call and map.