What Should My Boss Know Before I Connect an AI Agent to Work Email?
What I showed and said on October 7, 2026 about letting an AI agent into email. The page lists what I showed, not what any particular boss will accept.
I showed or said five things on October 7, 2026 that a manager can check: what the AI agent can read, what it can send, the guardrails I give it, how access is removed, and how often I review its instructions.
If the person you report to has to say yes before an AI agent touches work email, these are the five things I showed or said on October 7, 2026, in the order the sections below take them. Each one has its timestamp, so you can open the recording and show your manager the moment.
- 01
What it can read (43:33).
- 02
What it can send (44:14).
- 03
The guardrails I give it (41:26).
- 04
How access is removed (45:04).
- 05
How often I review its instructions (24:51).
Where can your manager see what the AI agent can read?
The screens in this session are in Gumloop, where Gmail is one of the connectors. On my relationship to it, I said this:
From the recording, 33:22. I don't work for Gumloop. I found this tool over the summer through Lenny's podcast, and I started using it and have since made myself an ambassador for them. I've created a relationship with Gumloop. I am not an employee.
More on that is in Do you work for Gumloop? (the answer from the session)
As shown October 7, 2026, I clicked into Gmail under the connectors section, and a new screen opened for it (43:33). Under tool management I selected read only tools. There are six options there, and I can restrict or give permission on what I wanted my AI agent to read (43:59). Of that screen I said:
From the recording, 43:44. This allows me to have extra clarity and control over what my agent does.
Checked on 2026-11-05 against Gumloop's agent documentation, which describes a Tool Management panel in a connector's detail view, with the tools grouped into Read-only tools and Write/delete tools.
Where can your manager see what it can send?
Beneath the read-only group there is a second group, called write/delete tools (44:14). That is the group that changes something.
From the recording, 44:19. I can then give permission or restrict access to sending emails, updating, creating drafts, forwarding emails, et cetera.
What guardrails do I give it?
I give my AI agents guardrails: I tell them what they can do and what they can't do (41:26).
For the account itself, someone asked if they could practice with a dummy Gmail account. I said yes, and that it is a good way to build your comfort level (19:32). I suggested starting with a dummy account (20:06), and I called a separate account that only the AI agent uses an extra step:
From the recording, 19:58. It's a great way to be comfortable with the level of access that you're giving to a system.
How is access removed?
As shown October 7, 2026, the removal sits on the same connector row, behind the three dots on the far right.
From the recording, 45:04. If I click on the three dots, it allows me to switch an account, but it also gives me access to easily remove access.
I removed Gmail that way, and I re-added it for the demo (45:22).
How often do I review the instructions?
There is a toggle labeled edit own instructions, and I said it lets my AI agent learn from how I interact with it and update itself (24:18). My review is a standing calendar item.
From the recording, 24:51. So as a best practice, I have a 15 minute meeting that reoccurs every three weeks on my calendar where I jump in and I review my agent instructions so that as it's auto updating, I am still fully aware of what's happening.
I said that if you use your AI agent daily, you may want to check yours every two weeks (25:10). A full page on the cadence is How Often to Review AI Agent Instructions.
What goes wrong
I named two situations. The first is leaving the AI agent short: I said that if I leave out critical information and restrict access, that new hire isn't going to achieve as much (6:05). The second is self-updating instructions I stop watching.
From the recording, 25:23. But I just want to make you aware that if you keep the edit on own instructions on, you want to be able to manage and have oversight into what the agent is updating. So please put a calendar reminder at some frequency for you to come back and check that.
What the recording does not show: the names of the six read-only options, which tools I left on or off for Gmail, or what any employer's policy allows. Whether a particular boss accepts what I showed is a separate question.
Where this comes from
This article grows from the session I taught live with Women Applying AI on October 7, 2026. The quoted passages are my own words, taken verbatim from the recording. The steps, lists and headings around them were drafted by my team's AI agents from those words and my published pages, and checked against the recording before publishing. Platform details that have a public source carry a dated Checked on line.
Build Your First AI Agent in One Hour: the full session with the clips, the prompts and the transcript.
Related reading: How Enterprise Legal Teams Start Vetting an AI Agent Platform, How to Onboard an AI Agent Like a New Hire, and How to Control What an AI Agent Can Do in Gmail.
Frequently asked questions
How do I show my manager that access can be removed?
Through the three dots on the connector row. I removed Gmail that way and re-added it for the demo (45:04).
Where are the read and send settings?
On the Gmail row under connectors, in a tool management section with a read only group of six options and a write/delete group, as shown October 7, 2026.
How often do I review what my AI agent writes into its own instructions?
I have a 15 minute meeting on my calendar every three weeks, and I said that if you use your AI agent daily you may want to check yours every two weeks (24:51).
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.