73 Strings' Yann Magnan - unlocking valuation intelligence in private markets

00:36:21 · Yann Magnan, Michael SidgmorePlay from here

[00:36:21] S0: I think it will be massive. I think it will be massive in terms of the ability to operate, the ability to do things faster or better, the ability to extract, again, insights and intelligence out of all this type of information. Yeah. I think it will be massive. [00:36:43] S1: The other interesting thing, and my mind goes back to when you said, seventy three Strings started in 2020. That was quite an interesting time to start because you started before AI really took off. Right? It kind of hit consumer adoption in really in earnest in '23. Claude Cowork and Opus 4.6 came out end of twenty five, and then you've seen a real step function change in some of the things that AI is able to do, and it feels like it's now on this somewhat exponential growth curve, not without its issues, but getting much, much better, much, much faster. What was it like starting when there were some green shoots of AI and you knew AI probably existed, but it was still very early days versus today, AI is probably integrated and embedded in so many processes and workflows that you have to think about it as you build the business. What was it like then and what is it like today?

00:36:43 · Michael Sidgmore, Yann MagnanPlay from here

[00:36:43] S1: The other interesting thing, and my mind goes back to when you said, seventy three Strings started in 2020. That was quite an interesting time to start because you started before AI really took off. Right? It kind of hit consumer adoption in really in earnest in '23. Claude Cowork and Opus 4.6 came out end of twenty five, and then you've seen a real step function change in some of the things that AI is able to do, and it feels like it's now on this somewhat exponential growth curve, not without its issues, but getting much, much better, much, much faster. What was it like starting when there were some green shoots of AI and you knew AI probably existed, but it was still very early days versus today, AI is probably integrated and embedded in so many processes and workflows that you have to think about it as you build the business. What was it like then and what is it like today? [00:37:40] S0: It's very different, no doubt. At seventy three Strings, we actually have been using AI since day one, mid twenty twenty since we started the business. But AI at that point in time, as you rightly pointed out, was a different type of AI. We're using machine learning and natural language processing, still algos. So what we've seen is that these algos have improved dramatically in the past few years, and they will continue to improve dramatically. So it's a question of continuing to embed this technology and basically continuing to add efficiencies into these platforms that you can basically push to your clients.

00:37:40 · Yann Magnan, Michael SidgmorePlay from here

[00:37:40] S0: It's very different, no doubt. At seventy three Strings, we actually have been using AI since day one, mid twenty twenty since we started the business. But AI at that point in time, as you rightly pointed out, was a different type of AI. We're using machine learning and natural language processing, still algos. So what we've seen is that these algos have improved dramatically in the past few years, and they will continue to improve dramatically. So it's a question of continuing to embed this technology and basically continuing to add efficiencies into these platforms that you can basically push to your clients. [00:38:20] S1: Where do you think AI will have the biggest impact on valuation, particularly as you think about things like the identification of certain industries and workflows, things like co work, etcetera?

00:38:20 · Michael Sidgmore, Yann MagnanPlay from here

[00:38:20] S1: Where do you think AI will have the biggest impact on valuation, particularly as you think about things like the identification of certain industries and workflows, things like co work, etcetera? [00:38:32] S0: Moving faster across the evaluation process and also creating some kind of valuation intelligence. I'm very much certain that it's not going to happen anytime soon that an agent is fully going to be replacing a human to perform a valuation because the short story is that you will always need a regulator to be comfortable with the process. You will always need an auditor to be comfortable with the process. You will need the owners of the four zero one ks to be comfortable with the process. You need to have someone that you can basically turn to and ask, okay, how did you do this and how did you do that? Which at this point in time, as you know, AI is probabilistic. AI does not know how things are being done. So that is not going to happen anytime soon. But the other way around, can agents accelerate the process dramatically? Yes. Agents can work when you're sleeping as simple as that. So you can imagine that tomorrow you get into the office and most of the evaluation, most of the rollover of your valuation of your portfolio has been done overnight. And then you have to check what really matters. So all of that is going to massively accelerate the process, which means that the valuations are going to be churned faster with the same, if not higher level of quality. And that's going to inform, again, LPs and retail investors faster and better.

00:38:32 · Yann Magnan, Michael SidgmorePlay from here

[00:38:32] S0: Moving faster across the evaluation process and also creating some kind of valuation intelligence. I'm very much certain that it's not going to happen anytime soon that an agent is fully going to be replacing a human to perform a valuation because the short story is that you will always need a regulator to be comfortable with the process. You will always need an auditor to be comfortable with the process. You will need the owners of the four zero one ks to be comfortable with the process. You need to have someone that you can basically turn to and ask, okay, how did you do this and how did you do that? Which at this point in time, as you know, AI is probabilistic. AI does not know how things are being done. So that is not going to happen anytime soon. But the other way around, can agents accelerate the process dramatically? Yes. Agents can work when you're sleeping as simple as that. So you can imagine that tomorrow you get into the office and most of the evaluation, most of the rollover of your valuation of your portfolio has been done overnight. And then you have to check what really matters. So all of that is going to massively accelerate the process, which means that the valuations are going to be churned faster with the same, if not higher level of quality. And that's going to inform, again, LPs and retail investors faster and better. [00:39:57] S1: Is that the type of thing where it's going to just naturally mean that valuations are gonna occur much more frequently? Because with AI, it can? Yeah.

00:39:57 · Michael Sidgmore, Yann MagnanPlay from here

[00:39:57] S1: Is that the type of thing where it's going to just naturally mean that valuations are gonna occur much more frequently? Because with AI, it can? Yeah. [00:40:08] S0: I mean, definitely, I think so. That being said, we need to define exactly valuation. I mean, valuation requires... When you perform valuation from A to Z on the portfolio company, you need two things. You need the most recent set of information from your portfolio company, and then you need information on transactions for similar businesses, public companies that are similar. You need to be able to calculate discount rates over that. You can basically deliver on a daily basis. Portfolio company data is not going to be delivered on a daily basis. So what you will be able to do is basically refresh everything that's market and then take a call, take a judgment on where you think the underlying KPIs of the portfolio company could be at any given point in time. [00:40:57] S1: You think eventually AI will get smart enough or better than humans at doing that?