Brazil AI Agent: Knowledge Graph Test Schema PASS, Agent FAIL
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
7m ago
Hi All,
Has anyone seen different Knowledge Graph behavior between Test Schema and AI Agent execution on Brazil?
We have an AI Agent using KG to retrieve HR case and related profile data. Current comparison:
PROD / Zurich / AI Agent â KG: PASS
DEV / Zurich / AI Agent â KG: PASS
Training / Brazil / KG Test Schema: PASS
Training / Brazil / AI Agent â KG: FAIL for some HR case types
The failing Agent call returns:
Requested information could not be retrieved
The same records and NLQ work directly in KG Test Schema, while other HR case types continue to work through the same Agent/KG on Brazil.
Agent logs also show group_by_table=true. We are not assuming this is the cause, but we're interested in whether it affects extended/inherited tables.
Has anyone seen this on Brazil or know of a related PRB?
Specifically:
Do Test Schema and AI Agent KG use the same NLQ/query-generation path?
What does group_by_table=true control, and is it system-managed?
Are there known Brazil issues with AI Agent â KG across extended tables?
Is there a way to compare the generated KG query/query plan between Test Schema and Agent execution?
We already have a support case open, but wanted to check whether anyone else has encountered the same pattern.
Thanks.