OpenAI automates junior investment banking tasks

SPEAKER_1 (0:01)

Parenting unlocks some pretty impressive skills, like carrying grocery bags, a soccer ball, and a sleepy kid all at once. Good thing the newly designed Chrysler Pacifica has a few tricks of its own. Hands-free power sliding doors and lift gate, rear seat entertainment, and stow-and-go seating and storage that adapts faster than your family's weekend plans. Visit chrysler.com to learn more. Chrysler is a registered trademark of FCA US LLC.

SPEAKER_2 (0:31)

This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up require compatibility and availability varies 18 plus.

SPEAKER_3 (0:56)

College football is back, so Hilton called on me, the superstition concierge to make your fan rituals a reality. Need a room to match your lucky number? We got you. Want to make sure our team doesn't wash your lucky jersey? Oh, that smells lucky.

Hilton's unmatched hospitality can keep up with any superstition. Even a marching band wake up call at 555 and 55 seconds. Hit it. When you need a team that will do whatever it takes on game day, it matters where you stay. Hilton, for this day.

SPEAKER_4 (1:26)

OpenAI launched ChatGPT for Financial Services, developed directly alongside Morgan Stanley and Evercore as their primary design partners.

SPEAKER_5 (1:34)

Yeah, which basically means they are pointing their technology straight at the core, routine tasks that currently consume the lives of junior Wall Street bankers. We are looking at a system built explicitly to automate the manual research, the formatting and the financial modeling that really define the early years of a career in that industry.

SPEAKER_4 (1:55)

Right. If a machine is doing the grunt work now, how does the next generation of juniors actually learn to become senior bankers?

SPEAKER_5 (2:03)

That is the structural problem hanging over all of this. But before you can analyze the human element, you have to look at the mechanics of what they actually built and why it functions differently than the models we have been using for the past few years.

SPEAKER_4 (2:15)

Because this specific deployment is running on the GPT-6 Astra reasoning model. It is not just answering standalone questions or, you know, summarizing a block of text anymore.

SPEAKER_5 (2:25)

No, not at all. The demo they provided showed something highly integrated.

SPEAKER_4 (2:29)

Yeah, I saw that. The system was asked to analyze a potential mergers and acquisitions target.

And in a single continuous flow, the AI selected relevant peer companies for comparison, pulled the current stock prices for that group into a spreadsheet, cross-checked the resulting financial charts against the underlying raw data to ensure accuracy.

SPEAKER_5 (2:50)

Which is a huge step right there.

SPEAKER_4 (2:52)

Exactly. And then it provided a written explanation for a recent market sell-off and the subsequent rebound in the stock.

And after all that, it generated a fully formatted PowerPoint presentation in the exact style and template of the specific bank.

SPEAKER_5 (3:08)

The time compression is really the variable you have to focus on there because an OpenAI employee who ran that demo pointed out that this specific sequence of tasks is exactly what he used to hand off to human analysts when he worked in the industry.

SPEAKER_4 (3:20)

Right.

SPEAKER_5 (3:20)

When you hand that workflow to a junior banker, you generally expect a three to six hour turnaround.

SPEAKER_4 (3:24)

And a lot of that time is just pure friction, honestly. Like the human analyst has to log in to multiple different databases just to find the peer companies in the first place.

SPEAKER_5 (3:33)

Yeah, finding the comps is half the battle.

SPEAKER_4 (3:36)

They have to pull the financials, export them to Excel, format the sheet so it is actually readable, build the charts manually, write the commentary, and then spend, I don't know, an hour making sure every logo, font size and color hex code in the slide deck aligns perfectly with the firm's strict brand guidelines.

SPEAKER_5 (3:54)

Right. Checking RGB values at two in the morning. So shrinking three to six hour process down to seconds fundamentally alters the physics of time on Wall Street.

You are moving from a state where analysis is constrained by human endurance basically, to a state where analysis is only constrained by the number of queries a senior banker can think to ask.

SPEAKER_4 (4:15)

Think about how that alters the velocity of decision making for a second. Like when a process takes six hours, a managing director is going to be very selective about what they ask an analyst to model. You only ask for the comparisons you really need because you know the human being on the other end has a physical limit. They need to sleep eventually.