Build Your First AI Agent
Create a meeting-follow-up assistant you can use after your next client call.
Your meeting ends, and you still need to sort the notes, pull out what you promised, and write the follow-up. That work can follow you through the rest of the day.
In this live workshop, you'll build an AI agent to help with it. Join Denise Klein to turn meeting notes into a recap, clear next steps, and an email draft you review before sending.
You'll build along, test the result, and learn how to use the same process to plan other agents for your business.
Bring a laptop and an active Claude Pro or ChatGPT Plus subscription. You'll need only one. Both entry plans are currently US$20 per month when billed monthly, separate from your workshop ticket; applicable taxes may apply.
Presented by Castro Klein. Existing Castro Klein and Sproutware clients can register free with their client code.
Give AI a job you already know needs doing
If you've used AI a few times, you may already know it can write a decent email. Getting it to follow your process takes a little more thought.
You need to explain what information it should use, what the finished work should contain, and where it needs to stop for your review.
That's what we'll work through together. We'll start with meeting follow-up because you can check the result against the notes and see whether it did the job.
Build it, test it, and make it useful to you
You'll use a reusable set of instructions called a skill to guide your meeting-follow-up agent. You'll adapt those instructions, run the agent on sample notes, and review what comes back.
By the end, you'll have worked through how to:
- 1
Give your agent a clear job.
Define what it should do with the notes and what belongs in the finished draft.
- 2
Keep yourself in the loop.
Set the points where you check decisions, names, dates, and promises before anything is used or sent.
- 3
Improve the instructions.
Spot what the agent missed, explain the correction, and test it again.
- 4
Understand connections.
See how agents can use information from other tools, what access that involves, and when supplying the notes yourself is enough.
- 5
Plan your next agent.
Break another recurring task into its inputs, steps, decisions, and result so you can decide where AI would help.
You'll save your meeting-follow-up setup so you can return to it after your next call.
You're still the person making the decisions
Suppose your notes say a client will consider your proposal. Your agent should draft a follow-up that reflects that conversation. It shouldn't thank them for buying.
“Client wants the smaller package. I said I'd send a revised proposal Thursday. They need to check the budget before deciding. Check back Tuesday.”
“Thanks for talking through the smaller package with me.
I'll send the revised proposal on Thursday so you can review it against your budget.
I'll check back on Tuesday to see what questions you have.”
Checking that distinction is part of the build. You'll learn how to give the agent boundaries and improve its instructions when the result isn't right.
We'll use sample meeting notes while learning. Your first agent will prepare work for your review, and you'll decide what to send.
Illustrative example, not a client record or an actual AI output.
Learn a process you can use again
Once you've built the meeting-follow-up agent, we'll look at how to apply the steps to other work.
You might want help preparing for meetings, putting together client updates, or reviewing what needs your attention each week. You'll learn to examine the process before building: what starts it, what information it needs, and which decisions should stay with you.
You'll leave with your first tested setup and a way to think through the next one.
Your $27 ticket includes
The live, 90-minute build-along workshop with Denise, including time to test your agent and ask questions.
The meeting-follow-up skill and setup instructions to keep using and adjusting afterward.
The AI Daily Practice Handbook for exploring a morning briefing as another useful routine.
Seven days of replay access so you can revisit the build at your own pace.
Your Claude or ChatGPT subscription is separate. Bring your laptop with your account ready to use.
Learn with someone who builds business systems
Denise Klein is the co-founder of Castro Klein and your teacher for this workshop. Hilary Castro and Denise bring about 40 years of combined business experience and 20 years building systems.
Denise will explain the decisions as you build, including how to spot an unclear instruction and what to change when the first result misses the mark.
You can take another AI class, but you'll never get another teacher as patient as Denise, who can hop on your screen and walk you through it.
A few practical details
What do you mean by an AI agent?
Do I need coding experience?
Which AI account do I need?
Do I have to connect my email or share client information?
Will my agent send emails for me?
What if I need more time?
What if I can't attend live?
Will you offer more help afterward?
I'm already a client. Do I need to buy a ticket?
Build something you can use after your next meeting
Bring your laptop and join Denise on October 29. You'll build a meeting-follow-up agent, learn to check its work, and take away a process for planning your next one.
Thursday, October 29, 2026 · 11 AM Eastern / 9 AM Mountain
90 minutes on Zoom · $27 · Seven-day replay included
Claude Pro or ChatGPT Plus required for the build-along. Subscription sold separately.