Agentic AI for Oracle AI Database Professional (1Z0-1159-26) - Practice Exam 1

1. A business team needs a no-code environment for building, testing, and deploying governed agents over enterprise data. Which Oracle offering is the closest fit?
2. A changed paragraph is embedded again, but its old chunk remains searchable. What ingestion property is missing?
3. A chunking pipeline drops headings and document IDs. Answers contain relevant facts but reviewers cannot locate their origin. What should the pipeline retain?
4. A company wants several MCP-compatible assistants to call the same approved database tools. Which architecture minimizes custom per-assistant integrations?
5. A COSINE vector index exists, but a query explicitly orders by EUCLIDEAN distance. Why should the team inspect its execution plan rather than assume this index is used?
6. A developer adds a vector index hint because the optimizer chose another access path. What should happen before retaining the hint?
7. A developer exposes a database function as an MCP tool. Which design best limits its potential impact?
8. A developer needs an Oracle in-memory graph index for approximate semantic search. Which organization should the vector index use?
9. A document contains text instructing an assistant to ignore its system policy and call a privileged tool. What should the design do?
10. A large corpus favors a partition-based approximate index rather than a neighbor graph. Which Oracle index organization matches?
11. A legal search must return the mathematically exact nearest ten embeddings. What should the design prioritize?
12. A policy assistant keeps confidently answering from outdated retrieved procedures. Which fix addresses the knowledge freshness problem most directly?
13. A production query returns approximate neighbors quickly, but users need a different business relevance order. What is the best next step?
14. A question asks for an exact invoice number embedded in a long document. Semantic retrieval misses it despite related content. Which enhancement is most appropriate?
15. A RAG assistant retrieves confidential HR documents for non-HR users. Where should the primary authorization control apply?
16. A RAG prototype answers questions even when no retrieved passage supports the answer. Which evaluation case is essential?
17. A retailer wants to find descriptions with similar meaning even when they share no product keywords. Which representation best supports this requirement?
18. A search schema mixes embeddings of several dimensions. A developer proposes one fixed-dimension vector column for all values. What should be checked first?
19. A selective department filter changes a vector query's execution plan. What is the best interpretation?
20. A shared departmental Agent Factory deployment is planned after a local experiment. Which installation approach should be evaluated for the shared environment?
21. A support agent loses important conversation context too early. What should its designer review first?
22. A support portal needs exact product-code matching and semantic matching of issue descriptions. Which index approach should the team evaluate?
23. A team adds a reranker after vector candidate retrieval. What should the reranker primarily improve?
24. A team assumes MCP support automatically makes every connected tool safe. What should an architect clarify?
25. A team benchmarks an approximate query but has no reference answers. Which comparison is most useful for measuring retrieval recall?
26. A team doubles top-K, but answer quality declines as irrelevant context consumes the prompt window. What should it evaluate next?
27. A team embeds entire 300-page manuals as single vectors and misses short troubleshooting sections. Which change most directly addresses retrieval granularity?
28. A team needs collaboration across specialist agents with a unified final response. Which concept is the closest fit?
29. A team selects a prebuilt Agent Factory agent for a pilot. Which production preparation is still necessary?
30. A team selects IVF for a very large corpus. Which characteristic explains its candidate-pruning mechanism?
31. A team wants similarity ranking to emphasize vector direction rather than magnitude. Which distance is the most direct fit?
32. A team wants to distinguish retrieval failures from generation failures. Which review data is most useful?
33. A workflow must query sales, draft a summary, and send a notification. Which Select AI Agent building blocks best organize the work?
34. An agent team produces an unexpected notification. Which records best support investigation of the execution path?
35. An analyst wants to review generated SQL before executing it against sensitive reporting tables. Which Select AI action best supports this workflow?
36. An application needs an agent to invoke a specialized PL/SQL business operation. What should the author provide?
37. An application needs to poll an asynchronous team run rather than immediately start another duplicate run. Which capability should it use?
38. An application switches its embedding model but keeps old document vectors. Newly embedded queries return irrelevant matches although dimensions match. What is the best correction?
39. An assistant should search support cases but must not update them. What should the database MCP design expose?
40. An embedding service is unreachable from database-side vector processing. Which two areas should be checked first? Select TWO.
41. An engineer raises EFCONSTRUCTION for a new HNSW index. Which consequence should be evaluated?
42. An external agent needs to discover database agent teams and delegate work at the agent level. Which integration is intended for that purpose?
43. An HNSW build needs a denser graph to improve candidate connectivity. Which parameter directly controls the maximum neighbor count?
44. An IVF index performed well before a major shift in newly inserted data. Recall has since declined. What should the team investigate?
45. An MCP client can reach the database endpoint but cannot invoke a tool. Which checks are appropriate? Select TWO.
46. An organization wants to evaluate an agent update before replacing the version used by employees. Which practice is most defensible?
47. HNSW creation fails because the database lacks the required vector memory configuration. Which area should the DBA investigate?
48. Increasing approximate-search target accuracy improves recall but raises latency. Which explanation best fits?
49. Index creation omits a distance metric. Which metric should the team account for when reviewing the definition?
50. Security reviewers want proof that a database operation came through an approved AI integration. Which capability is most relevant?

Comments — 0