We're reclaiming inactive PDIs to keep them available for active builders. Learn what's changing, who's affected, and how to protect your work. Read More

How to perform semantic similarity matching between Incident text and Knowledge Articles in an AI Ag

DesiS
Tera Contributor

Hi Community,

I am building an AI Agent in AI Agent Studio to automatically recommend Knowledge Articles for resolved incidents.

I created a custom tool script that:
1. Retrieves incident details (short description, description, work notes, close notes).
2. Retrieves published articles from kb_knowledge.
3. Returns incident_text and kb_candidates to the AI Agent.

I expected the AI Agent to use these inputs and generate:
- Most similar Knowledge Article
- Similarity score
- Recommendation type
- Reasoning

However, the AI Agent only displays the returned data and does not perform semantic comparison or ranking automatically.

I explored the following Script Includes:
- AISearchRetrieveSearchResults
- AISearchRetrieveSearchResultsWrapper
- AISearchSemanticSearchUtils

I can retrieve search results, but I cannot find a supported API that returns semantic similarity scores or allows programmatic comparison between arbitrary incident text and Knowledge Articles.

My questions are:
1. Is there a ServiceNow-supported API for semantic similarity matching between incident text and KB articles?
2. Can AI Search embeddings or reranking capabilities be invoked programmatically from a custom AI Agent tool script?
3. If not, what is the recommended architecture for implementing Incident-to-Knowledge recommendations using AI Agents?

Any guidance, examples, or best practices would be greatly appreciated.

Thank you!

0 REPLIES 0