Morgan Stanley built an evaluation framework before AI reached 98% of advisor teams

What happened

Morgan Stanley rolled out an AI Assistant for financial advisors, but the more interesting part isn't the tool itself - it's that the firm built a formal evaluation framework testing every use case before deployment, covering accuracy, coherence, retrieval, summarization, and translation.

According to Morgan Stanley, more than 98% of advisor teams now use the AI Assistant. Document access reportedly rose from around 20% to 80%, with AI cutting the time spent searching for information.

The business problem

Scaling an AI tool inside a heavily regulated industry without a structured way to evaluate whether each use case was actually safe and accurate.

Why it worked

• Governance can be what drives adoption, not what blocks it
• The chain is governance → trust → adoption → ROI, not governance → less AI
• Evaluate every use case against defined criteria before rollout, not after
• A regulated industry needs this evaluation layer more, not less

Where this points