Your deployed legal assistant shows great performance but occasionally repeats incorrect legal terms.
Which tuning method best improves factual reliability?

• A. Add fact-checking steps using external tools during generation
• B. Increase output randomness to improve exploration
• C. Use more verbose prompts to reinforce correct definitions
• D. Replace retrieval with static hard-coded text snippets

When evaluating coordination failures in a multi-agent system managing distributed manufacturing workflows, which analysis approach best identifies state management and planning synchronization issues?

• A. Monitor agent outputs individually to confirm local correctness and examine results of specific workflow steps.
• B. Track workflow throughput and task completions to measure performance trends and highlight workflow outcomes.
• C. Assess synchronization methods during design reviews and use simulations to evaluate coordination across representative workflow scenarios.
• D. Deploy distributed state tracing across agents, analyze transition timing, study communication overhead, and verify synchronization accuracy.

You are building a customer-support chatbot that fetches user account data from an external billing API.
During testing, the API sometimes returns timeouts or 500 errors. You want the agent to be resilient-retrying when appropriate but failing gracefully if the service is down.
Which strategy best handles intermittent failures in API calls while still ensuring a good user experience?

• A. Implement exponential-backoff retries with a circuit breaker, and return a clear message to the user if all retries fail.
• B. Schedule retries using a fixed delay for all failure types, maintaining predictable timing and user notifications after each attempt.
• C. Retry requests with a consistent short delay after each failure and notify the user as each retry takes place.
• D. Return a standard fallback message on failures to maintain conversation flow and reduce the risk of service interruptions for the user.

A logistics company is implementing an agentic AI system for supply chain optimization that manages inventory levels, predicts demand, and automatically reorders supplies across multiple warehouses. Supply chain managers need to monitor AI decisions, understand the reasoning behind inventory recommendations, and intervene when business conditions change rapidly. The system must present complex data analytics in an intuitive way that enables quick decision-making while providing detailed insights when needed. Managers have varying levels of technical expertise and need interfaces that support both high-level oversight and detailed analysis.
Which user interface design approach would BEST support effective human oversight of this complex multi- agent supply chain system?

• A. Create a layered interface featuring intuitive summaries, drill-down capabilities for detailed analysis, contextual explanations of AI decisions, and clear intervention controls with impact visualization and decision support tools.
• B. Create separate specialized interfaces tailored to specific user roles, allowing managers to view AI- driven recommendations with drill-down options for role-specific details, but without a unified interface for cross-role collaboration.
• C. Develop a comprehensive dashboard with AI decision summaries, drill-down access to underlying data sets, and segmented performance metrics to enable targeted analysis of supply chain operations.
• D. Create a streamlined interface presenting only high-level AI decisions and simplified recommendations, with drill-down views limited to basic historical trends for quick reference.