CIOs still have issues to overcome in their approach to AI agents, including trust concerns and not enough focus on outcomes.