Jaeha LeeNew York

Morgan Stanley, part 1 of 1

Making research faster, and what the time was for

Automation is most useful when it gives you more time for judgment, not when it replaces it.

Fall 2025Private Wealth Management, academic-term intern

Context

In fall 2025 I worked at Morgan Stanley in Private Wealth Management, on Team Mondial, as an academic-term intern. It was about 20 hours a week alongside my classes. Like Piermont, the role came from a relationship that started with a freshman-year cold email.

My work was research supporting the team’s portfolios and clients: fixed income and equity research, and diligence that came around every week.

The problem

A lot of the time went to work that repeated. Weekly reporting meant pulling portfolio data from an internal platform by hand. Recurring research meant checking the same kinds of sources, in the same way, for every project.

What I did

  • Automated the weekly report. I built an Excel tool that pulled portfolio data from the internal platform automatically. Weekly reporting went from about two hours to fifteen minutes.
Beforeabout 2 hours
After15 minutes
Fig. 1 Time spent on the weekly report, before and after the Excel tool.
  • Used the firm’s AI tools on recurring research. Morgan Stanley has its own internal AI tools. I used them to make our recurring research workflow faster across more than ten sources, which saved more than 15 hours of analyst work a week.
  • Found something nobody had intended. Looking across more than ten equity portfolios, I found more than $12 million of exposure concentrated in one sector that no one had chosen on purpose. I helped the team work through the rebalancing decision.

What I learned

The tools were good at the repeated part: gathering, summarizing, checking the same sources again. They didn’t decide anything. The most useful thing I found that semester, the sector concentration, came from looking across portfolios and asking why they looked the way they did. No tool was going to ask that.

  1. toolsGather the sources
  2. toolsSummarize and check them
  3. peopleLook across portfolios
  4. peopleAsk why, and decide
Fig. 2 Where the line fell. The tools sped up the repeated steps. Deciding what mattered stayed with people.

Automation is most useful not when it replaces judgment, but when it gives you more time for it.

I had the chance to keep working at Morgan Stanley the following spring and summer. I chose not to. I wanted to learn how a business works as a whole, not one part of it, so I went to an early-stage startup, where there wasn’t a role yet for what I wanted to do. So I proposed one.