Drawing the Line: Agentic Autonomy When the Data Is Personal

Session Summary

In this session from MoEngage’s Agentic Marketing Tour, Suresh Teckchandani (VP – Product & Engineering, Ancestry) and Angela Fung (Executive Director Marketing, Morgan Stanley Wealth Management) explore how to define agentic AI autonomy when handling sensitive personal data. Managing highly confidential records like family histories and financial lives, both organizations navigate the tension between scaling AI quickly and establishing essential guardrails. The panel covers how to bound AI workflows based on risk rather than technical capability, implement firm-wide governance, and utilize confidence scoring to safely automate compliance and customer interactions.

Top Takeaways

• Aligning Autonomy with Risk: Agentic autonomy should be dictated by the potential risk and consequences of an action, rather than technical capability alone.

• Establishing Guardrails Before Scaling: Building bounded workflows, strict permissions, and continuous evaluation frameworks ensures agents remain safe as they scale.

• Managing Distributed AI Adoption: Scaling AI across teams requires centralized governance to prevent fragmented workflows and maintain compliance standards.

• Utilizing Confidence Scores for Governance: Implementing AI agents that provide transparent reasoning and confidence metrics allows human teams to review exceptions efficiently.

• Defining Clear Operational Boundaries: Establishing explicit lines for where agent autonomy ends and human judgment begins prevents costly missteps in high-trust industries.