A company stores customer records in OpenSearch and needs an AI-based solution that turns queries into data requests, exports the results as CSV files, and uploads those files to Amazon S3. Which option provides the most operationally efficient way to implement this workflow?

Creating an AI agent is the most operationally efficient solution because agents are designed to orchestrate complex, multi-step workflows involving various tools and APIs. An agent can interpret the user's query, interact with OpenSearch to retrieve data, process that data into a CSV format, and then upload it to S3, all autonomously. Using a single foundation model with few-shot prompting is insufficient because FMs excel at text generation and understanding but lack the inherent ability to directly interact with external systems like OpenSearch or S3, or to perform file format conversions without additional tooling. Building a traditional software application is feasible but less operationally efficient as it requires significant development, maintenance, and integration effort compared to leveraging an AI agent's built-in orchestration capabilities. Training a decision tree model is inappropriate; decision trees are classification or regression models and cannot perform multi-step data retrieval, formatting, and storage operations.