Wall Street usually loves a buzzword. Every few years, there is a "new electricity" or a "game-changer" that everyone talks about but nobody actually knows how to use. When OpenAI’s Sam Altman started taking meetings with big banks, most of them just tinkered. Morgan Stanley didn’t just tinker.
Jeff McMillan Morgan Stanley's Head of Firmwide AI, is the reason why.
He’s not some Silicon Valley transplant brought in to "disrupt" the culture with a hoodie and a roadmap. Honestly, he's a 15-year veteran of the firm who previously ran data and analytics for Wealth Management. He knows where the bodies are buried—or at least where the messy data is hidden.
Back in March 2024, the bank made a big move by naming him their first-ever Head of Firmwide Artificial Intelligence. It wasn't just a title change. It was a signal. While other banks were still arguing about whether employees should be allowed to use ChatGPT on their personal phones, McMillan was already deep into a partnership with OpenAI that would change how 80,000 people work.
The Ferrari in the Parking Lot
McMillan likes to use a specific analogy when he talks about the current state of AI. He says the technology is like a Ferrari, but most companies are just driving it in circles in a parking lot.
The car is fast. The engine is perfect. But the driver—the human element—doesn't know how to handle the gears yet.
"The technology has so far advanced beyond our own capacity to leverage it," he noted in a late 2025 interview. It’s a refreshingly blunt take. Most executives want to talk about "synergy" and "efficiency." McMillan talks about the fact that we basically don’t know what we’re doing yet because our brains haven’t caught up to the software.
To fix this, he’s been pushing a "human-in-the-loop" philosophy. It sounds like a technical term, but it’s pretty simple: AI shouldn't make the final call. Ever. At Morgan Stanley, the AI is a "copilot" or a "knowledge assistant." It finds the research, summarizes the meeting, and drafts the email, but the human financial advisor is the one who hits "send."
Why the OpenAI Partnership Mattered
Most people don’t realize how early Morgan Stanley was to the GPT-4 party. They were a launch partner. While the rest of the world was playing with the free version of ChatGPT (which uses older data and can "hallucinate" fake facts), McMillan’s team was building a bespoke, walled-garden version.
🔗 Read more: 46 Billion Yuan to USD: Why This Massive Sum Matters Right Now
- The AI @ Morgan Stanley Assistant: This tool allows advisors to search through 100,000+ internal research documents in seconds.
- The Debrief Tool: Launched in mid-2024, this thing actually listens to client meetings (with permission) and writes the follow-up notes.
- AskResearchGPT: A tool for the institutional side—the bankers and traders—to pull complex data without sifting through PDFs for three hours.
He’s obsessed with measurement. If an advisor uses the AI and it gives a slightly "wacky" answer, there’s a reason code that links back to the original document. It’s about transparency. You’ve gotta be able to see where the machine got its idea.
Training People, Not Just Models
Here is the weirdest part of the Jeff McMillan Morgan Stanley strategy: he thinks you should spend 90% of your AI budget on people, not the tech.
Think about that.
Usually, IT budgets are 95% software and 5% "here's a manual, good luck." McMillan argues that in a highly regulated world like finance, the risk isn't the AI being wrong—it's the human not knowing how to check if it’s wrong.
He’s been rolling out training modules for different roles across the firm. It’s not about teaching bankers how to code. It’s about teaching them how to prompt, how to verify, and how to stay creative when the "boring" parts of the job are automated.
What Happens in 2026?
As we move through 2026, the buzz has shifted toward "Agentic AI."
This is the next level. Instead of just answering a question, these AI "agents" can actually perform tasks—like rebalancing a portfolio or filing a compliance report—with minimal supervision. McMillan is watching this closely, but he’s cautious.
He recently mentioned that seemingly overnight, every firm in Silicon Valley became "agentic." But in a bank, you can't just let an agent run wild. You need governance. You need a framework.
💡 You might also like: Why 225 Liberty Street NY NY is the Real Power Center of Brookfield Place
What You Can Learn from the McMillan Approach
If you’re trying to figure out your own AI strategy, whether you're at a startup or a giant corporation, his playbook is actually pretty easy to copy:
- Clean Your Data First: AI is a mirror. If your internal documents are a mess, the AI’s answers will be a mess.
- Start Small: Morgan Stanley didn't build a "Global Wealth Robot." They built a tool that helps advisors find research papers.
- Keep the Human Accountable: Never let the machine be the face of the brand. The machine is the backstage crew; the human is the performer.
- Prioritize Culture Over Code: If your employees are scared the AI is coming for their jobs, they won't use it. They need to see it as a tool that gets them home for dinner an hour earlier.
The big takeaway from the Jeff McMillan Morgan Stanley era so far is that the most successful "AI guy" in finance isn't actually a tech guy—he's a change management guy. He understands that the code is easy, but the people are hard.
As the "Ferrari" gets faster in 2026 and beyond, the win won't go to the bank with the best algorithm. It’ll go to the one with the best drivers.
Actionable Next Steps:
- Audit your internal knowledge base: Before buying an LLM license, ensure your company’s "intellectual capital" is digitized and organized.
- Implement a "Human-in-the-Loop" policy: Create a formal rule that no AI-generated content leaves the building without a human signature.
- Shift your budget: Allocate more funds toward "AI literacy" workshops for your staff rather than just paying for API tokens.