
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-04-2024 01:36 PM
Any idea why Now Assist is unable to identify the affected users related to the Security Incident in the screenshot, even though they are clearly visible in the record?
I've seen a demo of this same scenario where Now Assist identifies the affected users, so I'm curious to know why it's not working in this case.
Solved! Go to Solution.

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-07-2024 11:06 PM - edited 10-07-2024 11:08 PM
Completely Agree ..
Good to See Now Its Working ; 🙂
Please Close the thread .
Regards
RP

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-06-2024 10:26 PM
Did u tried Playing with PROMPT with few Variations?
Regards
RP

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-07-2024 05:16 PM
Thanks @Rahul Priyadars. I haven't tried any alternative prompts but a) I'm using the exact same prompt that's used in the official ServiceNow demo, which returns the expected result:
and b) I'm struggling to think of an alternative prompt that asks for the same information as succinctly as 'Who are the affected users?'

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-07-2024 08:40 PM
From Above 2 Screenshot i only see one difference. Since Context is Set during the who Chat exchange.
First SS i dont see the name of Users in PANEL on your previous Prompt
while in Latest SS i see Name of Users coming in PANEL on the Previous Prompt before you asked - Who are the affected User.
REGARDS
RP

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-07-2024 09:49 PM
You are correct, @Rahul Priyadars. The context of the first image was that I initially requested a summary of the incident, which didn't identify the users.
However I've just tried the same exercise again, and this time it worked as expected and DID identify the users:
So same inputs, but different results. I guess that's just a function of the probabilistic nature of gen AI models in general - they don't always give the same answer.