The Five Hurdles Between Your Brokerage and AI That Works

At the NIBA Convention on the Gold Coast this week, I ran a MasterClass on getting AI working inside a brokerage. Most of the room agreed AI is here. Far fewer had it doing real work in their business. In our experience across more than 90 brokerages, the gap is rarely the technology. It's five practical hurdles that show up in almost the same order every time.

1. Where do I start?

Don't start with the AI. Start with last week's to-do list and look for tasks that pass three tests, which we call the 3 Rs. The task repeats: it happens often and the same way each time. It has a recipe: you could explain it on one page to a new starter. It's reviewable: a broker can check a draft faster than writing it from scratch.

2. Where does my clients' data go?

"Australian data" isn't one tick box, it's three. Where is the data stored? Where is it processed? Is it used to train the model? Some tools store data locally but process it overseas, so ask every vendor all three questions and get the answers in the contract, not the brochure.

3. How does it know our stuff?

AI is a brilliant new hire who has never been inside your office. It doesn't know your insurers, your wordings or your process until you hand it the binder. Connect it to a knowledge base, and make someone responsible for keeping that knowledge base current.

4. The senior and the young gun

A two-page "perfect prompt" doesn't scale. Six months later there are twelve versions in twelve inboxes. Package the expertise once, as a skill, a GPT, an agent or, in Brokernote, a routine, so the whole team presses the same button and gets the same quality.

5. Does it sound like us?

"Let's delve into your exciting renewal journey" is not how your brokerage talks. To fix the tone, show the AI your best client letters, because examples work better than adjectives. To fix the look, put the output in your own template, not a wall of chat text.

Your Monday list

You don't need a budget to start:

1. List last week's tasks and pick one that passes the 3 Rs.
2. Write its recipe on one page.
3. Ask your AI vendor the three data questions.

Get the full playbook

We've put the whole session into a free playbook: the 3 Rs worksheet, the vendor question sheet, a one-page recipe template, and a cheat sheet comparing Claude, ChatGPT, Gemini, Copilot and Brokernote. Read the takeaways and download the playbook.