Day 1 of Catalyit’s GetLYIT 2 gave independent insurance agencies a path from AI experimentation to practical results: choose one problem, give AI the right context, and build a process your team can repeat.
Jason Furst set the focus. Angela Ford explained the foundation. Matt Smith demonstrated the applications.
Another tool. Another announcement. Another idea saved for later.
Jason opened with a question: What has all that activity improved?
Start with a task your team wants to fix. What takes too long? What does everyone dread? What keeps getting pushed to next week?
At Catalyit, the answer was partner reporting.
Preparing reports for more than 100 partners meant pulling information from multiple systems. The team tested approaches, compared results, and built a reusable skill.
The first report was 10 pages long. Generating it was progress, but partners needed something they could scan. The team shortened it, applied Catalyit’s design system, and brought the key findings forward.
That win led to another application: building email communications with the same design system.
Put it to work: Choose one recurring task. Assign an owner. Define what improvement would look like before testing.
Your agency’s knowledge shapes the quality of its AI results.
Angela explained that useful output depends on four layers of context:
“Write a renewal email” leaves AI with gaps to fill. Explain the client’s situation, the review your team completed, and what the message should accomplish.
Then make that background reusable.
Angela’s framework was simple: tell it, save it, and connect it. Include context in your requests, save useful instructions and references, and connect relevant sources through approved tools.
To capture knowledge that lives in an experienced employee’s head, let AI interview them about a familiar process. Have the employee review and correct the resulting guide.
Put it to work: Write one page of context for a recurring task. Compare the output with and without it.
Matt demonstrated how agency context and clear instructions can support five practical jobs in Claude:
He also clarified the role of projects and skills: a project holds reference materials and context; a skill defines the steps for a task.
Human review remains part of the process. Verify facts, calculations, coverage language, and commitments before using the output. Matt used fictional information in his demonstrations to show how agencies can begin testing.
Put it to work: Pick a recurring job, gather the reference materials, document the steps, and name the reviewer.
Bring the three sessions together in one focused test:
At your next team meeting, compare the result with your starting point. Did the task take less effort? Did quality improve? What still needs attention?
Complete this sentence and schedule the test:
“This week, we’ll use AI to help with ______."