A travel-planning assistant built in Foundry must answer questions about tonight's weather at a customer's destination city. The internal knowledge base has no weather data, and weather changes hour to hour. Which approach fits this requirement?
Select an answer to reveal the explanation.
Short Explanation
Weather changes by the hour, so you can't rely on old records or canned examples. You need the assistant to call a live web or search grounding tool at query time to pull current conditions. That's the only way it can answer tonight's weather accurately.
Full Explanation
Configuring the agent to call a real-time web or search grounding tool that retrieves current weather data at query time is correct because it lets the assistant retrieve current weather conditions at the moment the question is asked, matching the requirement for hour-to-hour accuracy that no static internal source can provide. Fine-tuning the model on historical weather records for major cities is incorrect because it teaches the model patterns from the past, such as typical seasonal conditions, but cannot report what the weather actually is tonight, since weather is not a fixed fact the model can memorize. Building a retrieval-augmented generation index over last year's weather almanac is incorrect because it is also a fixed, historical snapshot; it can describe average conditions for a date but has no way to reflect what is happening right now. Adding a few-shot example showing typical weather phrasing is incorrect because it only demonstrates a response format or tone, supplies no actual data, and leaves the model guessing at real values rather than reporting them. Only a tool that fetches live data at query time satisfies a requirement that changes hourly.