AI gathers manager diligence while investment decisions stay human

Transcript

From the transcript How I Invest publishes with the episode.

1. 27:16

   We tried to be early adopters of AI when ChatGPT first came out, but then we realized the confidentiality concerns and we weren't really able to put, you know, manager material in there. You know, you could obviously have it, you know, proofread a memo that doesn't contain any confidential information or all the other things we might use chat GPT for, but we really thought we were going to be able to, you know, adopt it early on.

2. 27:40

   We had a working group we called Skynet, but each time we met, we were like, I, like, we're kind of locked out because the, you know, because of our confidentiality provisions that we agreed to with managers and, put it to the side. And then we finally, you know, got our hands around the corporate for lack of better term, you know, subscription chat GPT. And then we were like, you know, individually using it with, you know, everyone's working on their own different prompts and with more or less success.

3. 28:07

   And, you know, it's actually a lot of work to get, you know, those prompts right and to get get it where you needed to be. And meanwhile, we were working with and looking at software providers. You know, there were quite a few out there, but the, we spent about a year looking at names because each time we looked at a new name, things were moving forward, things were evolving, and each time we had a question or like, well, we're looking for a CRM solution as well, they'd say, oh, give us a month and we'll have that, you know, programmed in and we'll, we'll talk again and sure enough, you know, people are coming back and so it took us a while, we were pretty patient, And then more recently, you know, we've we've adopted a group or, you know, we've we've subscribers signed on, implementing it.

4. 28:48

   And the idea is, I think, you know, maybe to state the obvious that it does not make investment decisions for us. It doesn't make any decisions for us. And I I'm and not yet. Yeah, I used to say the, the interns were cheap AI, now the interns are expensive AI, but, you know, that kind of level of work that you might have an intern or an administrative person do, you have to gather the data, then you, you know, winnow it down somehow with criteria, you collate the rest of it to make sure that you've got all the pieces you need to underwrite in, in one place and you do your best to score it and, you know, maybe it's after lunch or maybe you had a, didn't have a good night's sleep or you're mad because you didn't source the deal and now you gotta work on it anyway and all these things, like, oh, that's a low score, you

5. 29:33

   know, and we've always tried to have objective metrics to look at and point to when we, you know, in our templates that we're underwriting and it's just really hard to be objective and score certain things. But now we can say, you know, take the first path, like get all the data, put it all in one place, make sure you're giving us the good and the bad. You don't just say like, man, you know, you're being simplistic, the prompt should be pretty specific on, please don't give me an answer to a decision, but please highlight what the ownership structure looks like, what percent, you know, of this firm is employee owned and how is it distributed amongst the team.

6. 30:12

   And if it's greater than x or smaller than x or whatever, then that gets a certain score. And when you point the system at a, it's, you know, say a folder or data lake with all your stuff in there, your meeting notes, your manager notes, all the man of the collateral paraphernalia, all that stuff that comes along with doing the due diligence and investigating managers, you can take your criteria, your philosophy, your criteria, and your own framework and your own template and have it be, you know, delivered to you as you want to know ownership, here it is.

7. 30:45

   You want to know fees, here it is. You want to know what terms are in the LPA that you'd set are red flags? Here they are. You can do that more or less, you know, or you can do a lot of the due diligence kind of data gathering and assessment really early upfront if you have all the materials, including legal due diligence, operational due diligence, before really spending the time to do the softer work around culture and, you know, how do they generate, like, do I buy this philosophy?

8. 31:12

   Is this philosophy going to persist? And it's just such a times, you know, it's a, it's a, it's a leverage for the team. We're, it affords us time to think and discuss and actually have everyone read the memos before a team meeting versus, know, you're trying to write your own memos, let alone read everyone else's. And, you know, a good example is, you know, someone has to go through the LPA to or something to look up exactly what the fees and operating expenses are when it really should just be in a table somewhere.

9. 31:42

   It's pretty exciting. It's pretty early days, so we'll you know, we have no real kind of success stories yet other