As agentic AI moves from pilot into real credit decisions, real money and real customers, institutions are under pressure to make those systems feel trustworthy and human rather than purely efficient. This panel looks at how design, governance and accountability choices shape whether a customer feels served by a system or processed by one. The audience leaves with a clearer sense of where human judgement still needs to sit inside an increasingly automated customer journey. The harder problem sits underneath: designing AI experiences that feel human risks quietly removing the humans from the loop altogether.