In Great by Choice, Jim Collins describes a ship at sea with a hostile vessel bearing down on it and only a limited supply of gunpowder.
If you load everything into one cannonball and fire, you’ll probably miss. Then you have no gunpowder left.
The better move is to fire a bullet first. See where it lands and adjust your aim. Fire another. Once a bullet hits, load the cannonball.
This is how brokerages should be testing AI
Here’s what I see instead. A firm buys an enterprise AI platform and holds a big rollout meeting. Three months later, four agents are using it and nobody can say whether it helped. That was a cannonball fired blind.
What a bullet looks like
A bullet is small, cheap, and low-risk. For example:
- One agent sets up the 5 p.m. daily check-in prompt from Configure the AI Before You Configure the Deal and runs it for 30 days.
- One team lead uses AI to draft listing descriptions for a month, then compares the edits against the old way.
- One recruiter uses AI to prep for five coffee meetings, then asks whether those meetings went better.
Each test involves one person, one task, one measure, and 30 days.
The honest part
Most AI pilots don’t fail because the tool is bad. They fail because nobody decided at the start what success would look like. “Let’s see how it goes” isn’t a measure.
Before you fire, write down one sentence: In 30 days, this worked if ___.
The other honest part: bullets have to be safe. Keep client personal information and contract details out of any tool until you know how it handles data. A cheap test shouldn’t create an expensive compliance problem.
Then load the cannonball
When a bullet hits, commit fully. Train everyone, build it into onboarding, and put it in your recruiting presentation. Calibrated confidence beats hopeful spending every time.
Try This
- Pick one friction point. Choose the task your agents complain about most.
- Find one volunteer. Look for someone curious, not necessarily your most tech-savvy agent.
- Write your success sentence before you start.
- Review at day 30. Then decide to fire another bullet, load the cannonball, or walk away.
Question
What cannonball have you fired lately without a bullet first? And what would a bullet have cost you instead?
Resources
Great by Choice by Jim Collins and Morten T. Hansen.
Test small. Win big. Win the day.
