How to Build a Two-Step AI Agent Flow for Keyword Extraction and Knowledge Article Search?

[MayankGupta77]

Hi Community,

I’m trying to build a two-step AI Agent flow using Now Assist in ServiceNow, where:

1. The first AI Agent takes a user’s natural language prompt and intelligently extracts meaningful keywords or phrases (ignoring filler words like "how", "can", "I", etc.).

2. The second AI Agent uses those extracted keywords to:

   • Search the Knowledge Base

   • Find the most relevant article(s)

   • Return a summarized response with the article’s title, number, and direct link

I’d like to understand:

✅ Step-by-step:

• How to set up two collaborating AI Agents or skills in Now Assist or Agent Builder

• How to pass data (i.e., extracted keywords) from one agent or skill to another

• How to structure the search logic to query kb_knowledge effectively

• How to ensure the response includes a clean summary + article links

Bonus:

• Any tips on using NLU or scripting to improve keyword extraction

• Ways to format the response cleanly in Virtual Agent or Now Assist

If anyone has a sample use case, flow, or configuration for something similar, I’d really appreciate it.

Thanks in advance!