How Morgan Stanley turned ~100,000 research documents into an advisor's instant answer
Morgan Stanley's edge is decades of research most advisors could never fully search. Working with OpenAI, it built a GPT-4 assistant that puts all of it one question away — and got nearly every advisor team to actually use it.
A library of intellectual capital almost no one could fully use
Morgan Stanley's competitive advantage in wealth management rests on a vast body of research — on the order of 100,000 reports and documents. The problem was access: any individual advisor could realistically tap only a sliver of it, however good it was.
Seeing generative AI as a genuinely disruptive moment it “didn't want to get left behind” on, the firm partnered early with OpenAI and became the first major Wall Street player to deploy a bespoke GPT-4 solution to its own employees.
An assistant that reads everything, instantly
The “AI @ Morgan Stanley Assistant,” rolled out to advisors in September 2023, lets them ask questions in plain language and get answers drawn from the firm's entire research corpus in seconds — turning a fragmented, hours-long search into a single query.
It put the firm's best collective thinking within reach of every advisor at once, rather than leaving it locked in documents most people never opened.
From answers to admin: the meeting “Debrief”
In 2024 the firm added “AI @ Morgan Stanley Debrief.” With the client's consent, it takes notes during a meeting, summarizes the key points, drafts a follow-up email for the advisor to review, and saves a note to Salesforce. AI moved from a question-and-answer box to something embedded in the advisor's actual day.
By public accounts, adoption reached more than 98% of advisor teams, and the share of the research library advisors could practically reach rose sharply. The tools earned their place by removing real, daily friction — finding the right research and killing repetitive admin.
Why nearly every team actually uses it
The striking thing isn't that Morgan Stanley built an assistant — plenty of firms have. It's that nearly all of its advisor teams use it. That happened because the tools solved problems advisors felt every day, not because anyone was ordered to log in.
The firm also made adoption someone's explicit job: in 2024 it named Jeff McMillan its first Head of Firmwide AI, elevating the effort from a wealth-management initiative to an enterprise mandate with a clear owner.
From one assistant to an AI portfolio — and now, agents
The wealth assistant was the beachhead, not the destination. AskResearchGPT (October 2024) brought the same pattern to the institutional side — staff ask roughly three times more questions than they did of the previous tool, and salespeople answer client queries in about a tenth of the time. DevGen.AI (January 2025), built in-house, translated nine million lines of legacy code into plain-English specs in its first five months, saving an estimated 280,000 developer hours in 2025 by the firm's own count. Adoption held up too: as of late 2025, the firm still reported 98% of advisor teams actively using the Assistant.
Then in June 2026 came the most telling move: Morgan Stanley announced it would open its stock-plan platforms to external AI agents via the Model Context Protocol — reported by CNBC as roughly $1.2 trillion in assets across ~3,400 corporate clients — making it the first major Wall Street bank to let clients' AI agents connect directly. CEO Ted Pick's framing on the Q1 2026 earnings call: AI is evolving 'from pure efficiency exercises… to something that over time becomes a productivity phenomenon.'
One owner, many use cases, careful evaluation
Morgan Stanley's pattern — prove it in one place, then scale it firm-wide under a dedicated leader, expanding carefully with systematic evaluations — is becoming the enterprise playbook. The direction is AI across the full advisor workflow: retrieval, meeting capture, CRM updates, follow-ups — and increasingly, agents acting on the client's side of the table.
And the lesson Fautons would underline: the adoption number (98%+) is the achievement, and it came from solving real work, not from a mandate.
Common questions
What is the AI @ Morgan Stanley Assistant?
A GPT-4-powered assistant, built with OpenAI and rolled out to advisors in September 2023, that answers plain-language questions from Morgan Stanley's proprietary research corpus — around 350,000 indexed documents — in seconds instead of a 30-minute manual search.
How widely adopted is Morgan Stanley's AI?
Over 98% of financial-advisor teams use the Assistant — a figure the firm first reported in mid-2024 and reaffirmed in late 2025. CEO Ted Pick has estimated AI tools could save advisors 10-15 hours a week.
What is AI @ Morgan Stanley Debrief?
A meeting tool launched firm-wide in June 2024. With client consent it joins the call, takes notes, summarises key points, drafts the follow-up email, and files a note to Salesforce — moving AI from a Q&A box into the advisor's actual workflow.
What AI tools does Morgan Stanley use in 2026?
Four public ones: the advisor Assistant and Debrief in wealth management, AskResearchGPT for institutional securities, and DevGen.AI for modernising legacy code. In June 2026 it also became the first major Wall Street bank to open client platforms to external AI agents via the Model Context Protocol.
The shift
- ~100k research docs, mostly unsearched
- AI treated as an experiment on the edge
- Advisors losing hours to admin and search
- No single owner for AI
- Adoption left to individual enthusiasm
- The full research corpus one question away
- First bespoke GPT-4 tool deployed to staff
- Meeting notes, summaries & follow-ups automated
- A Head of Firmwide AI owns the mandate
- 98%+ of advisor teams actually using it
I've never seen anything like this in my career, and I've been doing artificial intelligence for 20 years. We saw a window of opportunity that was just completely disruptive, and I think as an organization, we didn't want to get left behind.