What Do I Do After I Build My First AI Agent?
The test result is back and the build is done. The order I'd go in: feedback with a reason, a review reminder, three mornings of use, then one new tool at a time.
Tell your AI agent why you liked or didn't like the first result, put a review of its instructions on your calendar, and open it first thing three mornings in a row. After that, add tools one at a time, with the approval rule in place before each one.
Your test result is back and the build is done. What comes next isn't obvious.
What do I do with the first result?
Tell the AI agent what you think of it, and tell it why. At 46:26 I went through three cases: a result you love, a result you have mixed feelings about, and a result you don't like. For the mixed one I said to tell it "Hey, I like this, but I don't like that," and to tell it why (46:32).
For a result you don't like at all, I said to tell it why too (46:42). I tied it back to onboarding: the instruction I'd give a really competent new hire is something similar to what I'd want to give my AI agents (46:54). And don't answer with "try again" (46:47). How Do You Test an AI Agent and Give It Feedback? goes through the test step by step.
When do I read what it changed on its own?
If you left the edit own instructions toggle on, your AI agent can rewrite its own instructions between your chats. I keep a 15 minute meeting on my calendar that comes around every three weeks, and I use it to read mine (24:51). Three weeks works for me (25:19).
I asked the group to put a calendar reminder at some frequency to come back and check (25:36). Set yours before you close the build. How Often Should You Review an AI Agent's Instructions? has the cadence.
How do I make it a habit?
The session page has a homework step. Tomorrow morning, open your Chief of Staff AI agent before you open your inbox and ask for the brief. Do that three mornings in a row. The page says an AI agent is only useful if it's the first tab, not the fifth, and it gives you three lines to keep: "Give me today's brief." each morning, "Prep me for my 2 PM." before a call, and a one-sentence note when a draft is off.
Mine run on a schedule. Most of my AI agents run a couple of times a week (25:15). How Do I Schedule an AI Agent to Run on Its Own? shows how I set the days and the time.
Which tool do I add next?
One at a time, and the approval rule goes in first. The file on the session page ends on the same advice.
From the file on the session page. Once you know it well, add those one at a time from your own account, with the same rule: never send, never post, without me saying so in that turn.
The session page names what to add after today: Exa and Firecrawl so the AI agent can research a person or a company before a call, Google Docs so the brief lands in a document, and Slack if you run a team. It says to add the approval setting before any of them. Which Connectors Does an AI Agent Need? covers the connectors, and How Do I Make an AI Agent Wait for My Go-Ahead Before It Sends Anything? covers the approval rule.
When do I build a second one?
The session page's next step is to build your second one with me. It names the kinds of live build I run most weeks, free: meeting prep, job search, donor outreach and more. On the recording I said it this way:
From the recording, 49:52. I mentioned to you at the top of the hour that since the beginning of the summer, I've been doing this for free one to three times a week.
The session page links the upcoming builds. Pick the one that fits your work.
What goes wrong
An AI agent that lives in the fifth tab. The session page says it's only useful if it's the first one you open. The homework is three mornings in a row.
A result you didn't like, answered with "try again." I said to tell the AI agent what it surfaced and why that didn't help (46:47).
Instructions that update on their own with nobody reading them. If you keep the toggle on, I said you want oversight into what your AI agent is updating (25:31), and the calendar reminder is how I get it.
The homework and the order for adding tools come from the session page.
Where this comes from
This article grows from the session I taught live with Women Applying AI on October 7, 2026. The quoted passage from the recording is my own words, taken verbatim, and the other lines from the recording are restated with their timestamps. The passage marked as from the file on the session page comes from the files published with that session. The steps, tables and checklists around them were drafted by my team's AI agents from those words and my published pages, and checked against the recording before publishing.
Build Your First AI Agent in One Hour: the full session with the clips, the prompts and the transcript.
Related reading: Where Do I Start With an AI Agent?, What Should Your First AI Agent Be?, and How Do You Test an AI Agent and Give It Feedback?.
Frequently asked questions
What should I do with my first test result?
Tell the AI agent what you like, what you don't, and why. I said not to answer a result you don't like with "try again" (46:47).
How often should I review my AI agent's instructions?
I use a 15 minute meeting that comes around every three weeks (24:51). If you use yours daily, I said you may want to check every two weeks (25:10).
When do I add another tool?
One at a time, with the approval rule in place before each one, as the file on the session page puts it.
Build something that actually runs your workflow.
A focused, free 60-minute live session with Jacqueline. You build alongside her, on your own real task, and leave with an AI agent that is already running.