Target's competitive edge comes not from AI models themselves, but from the robust architecture built around them, according to SVP Siobhán Mc Feeney. She emphasizes that not every problem requires an AI agent, and autonomy is earned over time rather than granted by default. Agents are strategically deployed to solve problems that deliver the most value for Target's customers. These agents are becoming integral to Target's operations, connecting various systems across the supply chain. Mc Feeney believes AI agents are key to fulfilling retail's promise of having the right product, in the right place, at the right time. Target's approach involves rigorous upfront questions about the problem needing a solution and the type of agent required. Agents must be registered, certified, and their triggers, tracking, and lineage meticulously documented. Autonomy levels are carefully managed, with agents starting with basic autonomy and demonstrating their capability to earn more. Continuous monitoring and evaluation ensure agents remain effective and their performance is transparent. This holistic approach, encompassing architecture, taxonomy, data governance, and observability, is crucial for scaling AI effectively. Ultimately, successful agent implementation depends on a confluence of factors, including architecture, taxonomy, autonomy levels, security, and observability, alongside a cultural shift in workforce skills to manage humans and AI together.