Full transcript of “The race for excel AI agents: Here's what to know”. Click a timestamp to open the video at that moment. Need another video? Use the YouTube transcript generator.
0:00 Well, now for nearly 40 years, uh, spreadsheet mastery has been a right of passage, uh, in a multitude of entry- levelvel jobs from investment banking to private equity to consulting. And that may be about to change thanks to AI. I welcome this, Leslie. Yeah, cuz you you you were a spreadsheet cruncher. No, but I wasn't very good at it. You were an Excel monkey. I was. Yeah. Yeah. I did it for five years before I joined CNBC. And I always was way slower. Do you know what? Not just than my colleagues, than my late great older brother. He was a machine on it and it just always annoyed me. It's so for some people it literally clicks. When I started in business school, I literally bought a Excel for Dummies book, read it, still I'm terrible at Excel, but just within the
0:45 last month, here's some good news. There's been a slew of AI product launches that aim to take on Excel or Google Sheets. One of them called Shortcut.ai AI went viral when its Excel agent was unveiled in late July. The technology can build a discounted cash flow model from scratch in minutes. The CEO Nico Christie told us that Shortcut is already serving thousands of paying customers at thousands of companies mostly through individual lower level employees. The work that you do tends to be very timeconuming and I wouldn't say menial but you know um repetitive. And what you'd rather be doing is doing what the associates and managers do at these firms, which is mostly reviewing uh the work that takes more time to do. So now that we can have AI that can actually do
1:30 four hours of work in 20 minutes, um it's a no-brainer in terms of ROI and making the switch over. Christy said that ROI unleashed a battlefield in the race for superhuman Excel agents. Shortcut's parent company, Fundamental Labs, has rad in more than $40 million in funding from Process Ventures, Andre Horowitz, and First Spark to name a few. OpenAI just invested 14 million in something similar called Endex. General Catalyst just needed another company called Paradigm, and Google and Microsoft have been quickly rolling out AI features to defend their spreadsheet turf. So far, none of the agents though are 100% perfect. They still require human oversight, especially in scenarios with legal or regulatory or compliance
2:16 implications, but the technology is improving by the day and it's expected to have broad ramifications for Wall Street talent as well as bottom line efficiencies. And guys, this is all really just exploded in the last month. It's super exciting and makes your life better and should make other people's lives better in certain ways. Those that actually have to do the job, it will make your life easier. But there's those who won't get the job because there won't be a job to do, right? And I I mean I asked Christie about that. I said, "Is this going to just completely supplant the entry level workforce in all of these industries?" I mean, we're not just talking about investment banking. We're talking about consulting. We're talking about private equity. We're talking about accounting. I mean, there he said two billion Excel users um that that are proficient with it that use it every day for their jobs.
3:01 But I have a bigger point which is this kind of highlights a broader question of does AI now just arbitrage away kind of all the potential different views on a stock or or in in the market. I mean if AI builds the spreadsheet in the model from which then the whole of the sales side based their kind of forecast aren't we just kind of all going to converge on the same thing? I mean, maybe we already have that now because the biggest stocks are covered by so many people and you get the consensus forecast, but they're going to give you ranges, right? Yeah. And I think it's all going to be a range and someone someone's going to have to decide I'm picking the low end of this range. I'm picking the high end of this range. But I'll step back from I see I still remember in my early days when you had to build your own model. That was like what you were told. And yet you could
3:48 have just picked up any one of the 28 sellside analyst models and used that initially. But what does it actually really come down to? I am going to put in a bigger input on AI sales of this company than most others because I believe in the big themes. So again, it's it's so interesting like the the amount of time and energy that goes into all this and really do we then does it really make the difference that the underlying model versus getting those big themes right? Well, I think yeah, it all comes down to the prompts that you give the AI and and you know, tell it to look for certain assumptions and pull this 10K and compare it to this 10K. I mean, you still as a human, you have to instruct it to do what you want it to do. AI is not and so I tell my kids this all the time. AI is not like a mind readader. The prompt actually does
4:34 matter. If if if you have a view, you tell it the view. If you don't have a view, it will come up with its own view. I agree. If nobody has a view and you just are literally saying to the AI, what do you think? Yes, you will get to that point. If everyone has no agency over themselves whatsoever, you will we will we will all come to the a median in the middle of some sort, some average, but hopefully you have some agency in your life and you're not just going to listen to whatever the the AI gremlins have decided. So, I I basically agree with that. I kind of put it a different way, though. I I think the people who take more agency over the next 5 years and rely on AI less will do better because I think it's just going to speed up arbitrageing
5:19 away the consensus. We'll take a random walk down AI street perhaps and we'll find out the answer. I also think in human development and I think this is really relevant for kids, you develop your intelligence much more by reading and thinking yourself. And again, the people growing up that embrace AI less, I think will just generally grow their intelligence more. So I I I think there's a a counter trade here of embrace AI less, and you might go further. Well, we'll find out. I I'll take the other side of that coin for now. But anyway,