amab
1
Hello,
I’m currently building my own AI Assistant to manage my emails and calendar. During testing, it accidentally sent emails to 3 of my contacts 
I actually just wanted to reschedule something, but that incident made me realize I need to add a security layer to avoid unintended actions.
I’m looking for the best way to implement a confirmation system where:
- The assistant clearly explains what it’s about to do
- Then asks for my confirmation before taking any action (like sending an email or changing an event)
Has anyone implemented something like this?
Any ideas or best practices to make the confirmation flow smooth and secure?
Thanks in advance 
Hello Amab.
Please, can you explain more about the service of email ?
are you making in google calendar?
Do you use Ai Agent node ?
thanks
Prof. Jose Lins
amab
3
Hello, thank you @N8N_NATAL_BRAZIL
Below is the structure I designed :
ok, thank you for share the structure, in AI agent its necessary put in the prompt informations to use the tools , use markdown languages, like
<tool>
use always the tool "CalendarAgent" to read all calendar informations
use the tool "emailAgent" to read all emails
</tool>
the prompt its the soul in AI Agent.
And dont use gpt miniO use gpt-O
congrats
system
Closed
6
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.