You are designing an agentic workflow in Gemini Enterprise for a media organization. The agent will enable natural language search across thousands of hours of archived video training content. End users must be able to search the video library with text queries and receive precise video segment matches. Your solution must process video files stored in Cloud Storage, map visual content and user text queries into a shared vector space, and perform real-time similarity retrieval at scale.
What should you do?
Choose two

• A. Store video files with embeddings in BigQuery, create a vector index, and execute pattern matching queries against the index.
• B. Use the Text embeddings API with the text-embedding-005 model to generate vectors from Cloud Storage video object path strings and file names.
• C. Index the generated embedding vectors in Vector Search to perform approximate nearest neighbor (ANN) retrieval for user query embeddings.
• D. Generate vector representations for video segments using the Multimodal embeddings API in Agent Platform.
• E. Convert video files into Base64-encoded strings, and store them in the vector store. Send query requests to the Gemini API.

You are the lead architect at an organization where multiple engineering teams are building conversational agents. The teams frequently need their agents to query operational telemetry in BigQuery and inspect Cloud Logging logs. Currently, teams duplicate effort by writing custom API wrapper scripts and separate credential management logic for each agent tool.
You need to establish an enterprise-wide tool integration architecture that provides ready-to-use cloud capability connectors and a shared governance directory for team capabilities, without writing bespoke wrapper code.
What should you do?

• A. Configure Agent Platform Feature Store to store key-value client profile attributes, and execute a custom feature retrieval API call during agent prompt construction to inject historical facts.
• B. Deploy serverless Cloud Run functions for each cloud service API, and configure direct HTTP webhooks inside each agent prompt.
• C. Configure Dialogflow conversational fulfillment webhooks to route all agent capability requests through a central proxy service.
• D. Connect the agents to Google Cloud MCP servers, and register shared tools in Agent Registry.

You are developing an autonomous financial research agent using Agent Development Kit (ADK). The agent has access to a registry containing more than 50 distinct financial analysis tools. When you bind all 50 tools to the agent at once, the system prompt consumes too many tokens, causing high latency, increased costs, and frequent LLM hallucinations in which the agent selects the wrong tool schema. You need to improve the agent's tool-calling accuracy and reduce token consumption while ensuring all 50 tools remain available for the workflow.
What should you do?
Choose two

• A. Dynamically bind only a retrieved subset of tools to the agent's execution context before initiating the reasoning loop.
• B. Implement an ADK semantic tool retriever to query a vector index and only fetch the top-K most relevant tools based on the user's current prompt.
• C. Increase the model's temperature parameter to allow the agent to find the correct tool schema from memory.
• D. Combine a subset of all 50 tools into a single ADK ToolChain object to compress the tool schemas into a single system prompt description.
• E. Configure an ADK asynchronous execution hook to run tools in parallel and return only the valid data payloads to the agent.

You are troubleshooting a production customer device support agent built with Customer Experience Agent Studio (CX Agent Studio). Users report that the agent occasionally provides inaccurate warranty extension terms and device repair procedures that do not exist in the connected enterprise Knowledge Base datastore or tool outputs. You have inspected the affected sessions and raw JSON responses and verified the system instructions, variable states, and tool outputs.
You need to identify the root cause of the hallucination errors and enforce strict grounding while minimizing conversational latency.
What should you do?

• A. Configure Cloud Monitoring log-based metrics and alerting policies to track HTTP 500 status code errors across downstream Knowledge Base endpoints, and increase the API execution timeout threshold within the CX Agent Studio global connector settings.
• B. Export the conversation logs to BigQuery, run SQL queries to calculate overall session latency and token consumption, and decrease the Gemini model's temperature setting in CX Agent Studio to 0.0.
• C. Implement a custom before_model_callback Python function to intercept the conversation call, programmatically analyze the current agent behavior and conversational state to test for missing logic pieces, and dynamically inject hardcoded corrective instructions into the session variables before each model turn.
• D. Refine the system instructions using XML tags to explicitly prohibit the agent from answering using its own internal parametric memory, and run evaluation test cases with the "Find issues with AI" option enabled to generate a loss_report.