Create a agent which can list down the ServiceNow users based on the natural language filters

[prasad_parab]

Hi all,

I’m trying to build a ServiceNow AI Agent that can understand a user’s request in either natural language or NQL and directly return a list of matching sys_user records.

The agent should be able to:

1. Accept a user query like “show all active users in HR” or an NQL filter.

2. Parse the input to identify relevant fields and values.

3. Map those fields to the correct sys_user table columns.

4. Run the GlideRecord query internally.

5. Return the actual list of users (name, email, sys_id, etc.) back to the requester.

Has anyone implemented something similar or have pointers on best practices for designing such an AI Agent?

[GlideFather]

Hi @prasad_parab,

when you say AI AGent you refer to Now Assist Agent, right?

I would create an AI Agent, add tool for search records in HR, and other steps. I don't understand the parsing and mapping points, what's the goal with this? 

When you want to receive name, email and sys id of users that are in HR, can you do it just by dot-walking? 

I most probably miss the whole requirement :((

[Simon Hendery]

I agree with @GlideFather. Interesting idea, and fun to build, but I doubt there's a real world business case for it.

If I understand correctly, what you basically want is: a voice-activated version of ServiceNow that a sys admin can use to mess around with records. Or am I wrong?

As I say, cool project idea. Very HAL 9000. But keyboard and mouse probably offers a better ROI.

That said, perhaps if you explain what part of the architectural structure you're stuck on, someone can fill in the gaps.