Day 2 of Catalyit’s GetLYIT 2 focused on making better AI decisions: protect agency information, evaluate new marketing opportunities, and tie technology spending to measurable results.
Ryan Smith covered safeguards. The ChatGPT Ads session explored a new advertising channel. Dan Mory explained what to ask before buying another tool.
Clear rules give employees confidence to use AI.
Ryan framed the risks around three questions:
Start with an inventory. Include standalone AI tools and features built into existing software. Identify what each can access, who owns it, and whether that access is necessary.
Then document approved uses, data restrictions, review responsibilities, and escalation steps. Train employees on those expectations.
Ryan also recommended testing with fictional data and keeping a backup process for work that depends on AI.
Put it to work: Inventory your tools, review their access, and close gaps in your policies and oversight.
Jason Furst and Bria Sylvester explored how insurance agencies could approach ChatGPT advertising.
Their starting point was the prospect’s situation. Someone asking AI about insurance may describe a new business, a specific concern, and a need for local guidance. Your message should address that need directly.
Explain whom you help, what questions you can answer, and what happens after someone contacts you. Carry that message from the ad to the landing page.
The session also separated paid advertising from organic AI visibility. Buying an ad does not secure a recommendation in an AI-generated answer.
Jason made clear that Catalyit was presenting an opportunity to evaluate, without endorsing the platform.
Put it to work: Choose one audience and one goal. Set a spending limit, define a qualified lead, and establish tracking before launching a test.
Start with the problem your agency needs to solve.
Dan urged agencies to document the workflow before evaluating a purchase. Find the delays, duplicate entry, missing information, and unclear handoffs.
Check existing systems first. Better training, an unused feature, or a simpler process may solve the problem.
Include the employees who do the work in demonstrations and testing. They can identify extra steps, correction work, and disconnected systems that reduce a tool’s value.
Measure the entire task, including setup, verification, corrections, and follow-up. Compare the improvement with the full cost of implementation and ongoing use.
Put it to work: Document one workflow, record a baseline, and define the result that would justify an investment.
Bring three questions to your next team meeting:
Assign an owner and a review date.
Before your next AI investment, complete this sentence:
“We’re considering ______ to improve ______. ______ owns the evaluation, and we’ll measure ______.”