Document Processing · Enterprise Knowledge Search
Morgan Stanley needed trustworthy AI that could help advisors search a large internal knowledge base, summarize research and meetings, and support client conversations while meeting strict quality, compliance, and privacy standards.
Reported impact
Morgan Stanley achieved broad advisor adoption, sharply expanded the share of searchable documents, shortened follow-up turnaround from days to hours, and created a reusable eval framework for additional AI products across the firm.
Morgan Stanley paired GPT-4 with an evaluation-driven rollout so advisor-facing AI could be trusted before it reached production. AI @ Morgan Stanley Assistant serves as an internal chatbot for wealth-management teams, using retrieval over the firm’s growing corpus of intellectual capital to answer advisor questions quickly and consistently. The team continuously refined prompts, retrieval methods, and translation support through evals scored by advisors and prompt engineers, expanding coverage from 7,000 questions to effectively any question across 100,000 documents. On top of that foundation, Morgan Stanley launched AI @ Morgan Stanley Debrief, which combines Whisper and GPT-4 to turn consented Zoom calls into client notes, CRM updates, and draft follow-ups. Daily regression testing, expert review, and zero-data-retention controls keep outputs aligned with financial-services requirements.
Advisor questions are routed through AI @ Morgan Stanley Assistant, which retrieves and summarizes content from the firm’s internal document corpus. Client meeting recordings flow through Whisper and GPT-4 in Debrief, then reviewed notes and follow-ups are pushed into CRM systems for advisors to finalize.
"Debrief turns Zoom recordings, with client consent, into actionable outputs like client notes, which are automatically integrated into CRM systems"
"OpenAI’s zero data retention policy also addressed key security concerns"
"Today, over 98% of advisor teams actively use AI @ Morgan Stanley Assistant"
"We went from being able to answer 7,000 questions to a place where we can now effectively answer any question from a corpus of 100,000 documents"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
OpenAI Sample Jul 31, 2026 Univé Univé needed to build organization-wide AI capability across insurance operations so employees could safely redesign their own work instead of waiting for centralized software projects or isolated AI pilots. 85% of licensed users active every week ChatGPT Enterprise · Custom GPTs · Workspace Agents OpenAI Sample May 6, 2026 Singular Bank Before Singularity, analyzing a client portfolio meant working across systems: pulling positions, reconciling data manually, and assembling a coherent view ahead of a meeting. That process could take significant… Singular Bank built an internal assistant that analyzes portfolios, recommends next actions in real time, and… Codex · Singularity OpenAI Sample Feb 18, 2025 Allianz Allianz needed a secure, compliant way for employees to use generative AI with internal company content in daily workflows without exposing data to public models or violating ethical AI, compliance, and privacy… all 158,000 Allianz employees worldwide Allianz Azure Cloud · AllianzGPT · AllianzGPT Premium 