AI is Revolutionizing Private Equity Due Diligence

Hello, welcome to another episode of Executive Suite. I'm your host, Thomas Braun, and today we have with us Richard Song, the CEO and co-founder of Ravana. Uh Richard, you're uh in a very interesting space at the crossroads of fintech and artificial intelligence. Can you please tell us uh about your company? Be happy to. Thanks for having me. Raana is an AI powered platform that streamlines private equity due diligence. So it makes it faster for investors to discover insights, to find answers um during their diligence process. So, [clears throat] for for those of you who who are kind of new to private equity, what this looks like is due diligence is is a one to two-month period of time where the investors are receiving a lot of information from the company that they're interested in acquiring. And they have to go through that information and and really, you know, dig through it, parse through it, analyze it, u you know, analyze the data, spot any potential risks to the investment. And it's a very intense period of time. Uh and so uh the associates on the team, the VPs will be working, you know, 80 to 90 hours a week during this period. So what we do is we've built an AI that streamlines this process where investors can upload that data room to Raana and we're helping them review the data room. We're helping them analyze documents. We're also streamlining some of the processoriented workflows like diligence tracking for example um that that can be quite tedious and is still done in Excel today. Um and really you know my my background is that I was uh a private equity associate at Bane Capital. So I lived and breathed a lot of these pain points and workflows um day in day out. And so we we took that experience and uh realized that you know AI with how it's developing um really presents a great opportunity to to streamline that process. >> So how does it work? Is you have like a dashboard on your computer and you just start uploading documents or what? >> Yeah. So it's a web- based platform. Uh so you can access it through uh you know your Google Chrome for example and yeah so it syncs with your uh you know your your share drive your your file system and you can upload um files and and folders to Raana. >> Okay. So how does that work when a target company has a data room? >> Yeah. So the target uh company has a data room. they are uh probably sending that through some sort of VDR like a data site or an intral links as an example. Uh so typically what investors will will do and what their current workflow looks like is they're going to bulk download everything into their local computer um into their their company's computer uh system. And the reason for that is because information in the VDR is meant to be ephemeral and uh it can change right uh files can be deleted or added and so it's very important for the investor to maintain their own record of the documents and and download all of that. So once they've downloaded that then they can very easily upload that data to Raana and they can continue to upload new versions of the data room to Raana and we always will be able to maintain the most up-to-date version of the data room on our platform. Um and we also help detect you know new new files that are added files that were deleted really to help keep the investor organized as they're going through this process. So um did I understand you correctly like the uh software is actually going to check the uh data room to make sure that if changes are made that your documents are updated. >> Yeah. So that's that's kind of the the the the long term is it a direct kind of connection with the cellside VDR. Today it's based on the user uploading uh those documents to Raana. >> Right. Okay. So the target company signs like an NDA or something with the PE fund and so PE fund can download all of that documentation. Can you give me an idea of how much documentation we're talking about? Like how would you measure it? Yeah. So the way to think about it is you know the the amount of data in the data room is usually um on the order of like one gigabyte. Sometimes it can be it can be less than that. Uh but our capability to handle um you know large volumes of data is extends beyond that range. So we can we can handle you know up to like 10 gigabytes you know 15 gigabytes of data. It's really kind of unbounded on on our end. um from a technical standpoint and the reason for that is we've built a very um you know highly engineered infrastructure that uh handles the the upload the processing and the parsing of all of that data when it is uploaded to Raana. >> Okay. So are we talking about uh the size of M&A deals that are on the higher end? >> Yeah, exactly. That's right. >> Okay. Okay. So, this would by high end, how would you define that? Like over $500 million. A billion. >> Yeah. Over 500 mill. It could be, you know, tens of billions of dollars in uh enterprise value, right? >> So, your your customer, your ideal customer is a company like Goldman Sachs or some Bane Capital. Can you give me some capital? Yeah. So we >> our customer target base is um on the sell on the buy side. Um so it' be like uh um you know private equity firms, latestage growth equity firms uh that are involved in rigorous due diligence. So, um, you know, this can range from anywhere from like a a small independent sponsor firm to a to a mega fund like a Blackstone, for example, that's evaluating, you know, deals that are, you know, much more complex. Um, but from a technical standpoint, we have the capability to to address, you know, that full spectrum. And so, um, it really is kind of a focus on just the broad private equity space, >> right? Because I think even if you are doing uh like smaller midsize deals, those um smaller institutions or family offices, they don't have the staff. So they need they need the AI just as much as >> Bane Capital or Morgan Stanley or whoever, right? So, um, yeah, I think investment banking could use this, too, because if a company's going to do an IPO, then the underwriter has to do all this due diligence as well. So, you >> absolutely that's that's a that's a yeah, that's a that's another kind of customer segment that I think our product would be uh very well suited for. And even even in a sellside process um you know it's it's quite valuable to be able to anticipate how the buyer will analyze the information in the data room. And oftent times in the due diligence process the buyers will be asking the sell side and the investment bank um specific questions about the information in the data room. And this can be you know very timeconuming to address especially if you're in a an active process involving you know anywhere from like 5 to 10 potential uh you know buy you know buyers u that are all sending you questions right so the ability to use AI to sift through that vast amount of information um to to be able to address those questions quickly absolutely is a is a functionality that would be useful on the cell side as well. >> Right. Okay. So um wow. So do you think um this something that M&A advisors need to know about? >> Yeah. I mean I think there's there's a lot of uh when when we talk about the sell side, right? That encompasses M&A advisors. It it encompasses u you know the more bulge bracket institutions like a Goldman Sachs or JP Morgan um anywhere down to you know like a like a [clears throat] smaller advisory firm. So all of these kinds of you know businesses that are involved in uh M&A transactions and corporate transactions in general um that that's something that that I think you know would would be quite relevant for for Raana. >> Yeah. Well even law firms because a lot of times those other firms will will they'll just turn around and they'll say to the law firm okay go do due diligence on this company and they make the company pay for it. So they go to the most expensive law firm in town. Um now one of the other things I was wondering about is do so does Raana have sort of like preloaded checklists? Um is it making sure that you are checking off all of the boxes that you have all of the different documents and does it warn people if you're missing something? >> Yeah. So that's um speaking directly to one of the features that we have which is kind of a diligence request list feature. So typically in an in an M&A transaction there will be a pretty extensive diligence request tracker um that basically lists out like a set of documents that um you know that the buyer's expecting to receive from the data room. And we we have a standard template version of that in Raana, but we also enable the uh you know the user to upload their own custom tracker. And what Raana does is it uh goes through each of those requests in parallel and sits through the data room to identify okay what are the relevant documents? Is this request completed or incomplete based on um you know what's in the data room today? What are the findings uh in regards to to uh this request item? And so it provides the investor instead of having to you know dig through all those you know hundred requestless items manually right it gives them a instantaneous snapshot of all right what do I have what do I need to to ask for from the sell side and what what's missing and what are the potential risks in in each of these request items. >> Okay. So does the AI generate comments? >> Yeah. Yeah. Exactly. >> Okay. So you can then send over your list of questions that were generated by the AI. >> Yeah, that's that's exactly right. It's it's fully kind of exportable into uh Excel. So um really facilitates easy back and forth between the buy side and the sell side on those request items. >> Yeah. So, what do you think about um if somebody then takes those uh questions and comments and they load it into an AI uh program and ask for the answers and then they get a list of answers and they send that back to you? >> Yeah. Well, I do think that that's a very interesting point. And I think that that is where the industry is moving. Uh in terms of I think you know both the sell side and the buy side are going to have AI and the lawyers and and the other third party advisors. I think it's going to be a a tool just like how everybody has access to Microsoft Excel, right? I think it's going to be one of those uh tools that becomes deeply entrenched within our our uh you know our working environment, right? And into our workflow. Um, but I think, you know, it's it's a good thing because it it increases collaboration, increases efficiency. Um, you know, the the most annoying thing in a sellside process is when you email the the uh you're an investor, you email the the investment bank, the M&A advisor for for some piece of information and their analysts are overworked. So, they're not getting back to you until, you know, maybe 12 hours later or maybe 24 hours later, maybe it slips through the cracks. Um, and so the ability to just quickly kind of get get the 8020 answer from AI, I think will, you know, definitely provide a lot of efficiency and and drive increased speed in in this process. Uh, but I think as long as it's grounded in the the truth of what's in the data and it's verifiable, it's accurate, um, that is that and that's the type of product that we're building. Um I think that that this will be kind of deeply embedded throughout the ecosystem. >> Interesting. So is this going to mean that um the investment banks and and these PE funds will they need less staff overall or will they increase their deal flow? >> Yeah, it's a that's a good question, right? and very relevant as you know the broader uh economy is thinking through what AI means uh for for their workforce. I I what my belief is that it will not lead to a reduction in the the size of the workforce but I think what people will spend time doing and where kind of the bottleneck moves in the process will shift. So it'll shift from some of the, you know, the the low sort of low-level work and and manual grunt work to really, you know, judgmentbased uh work, right? Which is really what what draws people, smart people and hardworking people who graduate college and go into the industry in the first place. Um, and you know, there's there's, you know, this this term called the Jievens paradox. um which is actually that as you make uh a resource cheaper to consume the consumption actually increases. So we could you know see as a result of AI uh an increase in M&A activity an increase in deal volume and at least an increase in the speed at which deals are done. Um I think that could be be quite exciting for the industry. Um but I I do think that the nature of the role will change especially at the junior level. It'll be less about kind of managing the grunt work and uh and more about applying your judgment really you know your your differentiated perspective on the world and uh I think that that's quite exciting right to be able to engage at the heart of investing at an earlier stage than than than you typically would be able to do before before you have access to these tools. So is uh is your uh program dedicated to just uh this document management feature or um can users ask it questions like uh chat GPT? >> Yeah. So uh to to fully clarify this is a fully AI native platform. So, it enables users to ask questions just like chat GBT except, you know, we're able to provide answers over a a much larger base of of of information, right? Um, a common issue with chat GPT is you upload, you know, 10 files to it and it just breaks down. It starts to lose accuracy and and the the citations are not quite as good. With Raana, we're providing the ability to to chat across, you know, the whole data room, gigabytes of data, uh, with pinpoint accuracy, citations that that bring you back to the primary source on on every fact and detail that we're generating. We also have, you know, prompts that are uh, autogenerated uh, so that, you know, it provides people with a, you know, highle overview before they have to even like prompt with the AI. I think that's also part of where the future of AI is is headed is, you know, the ability to present the user with information rather than having the user to ask for it. >> Interesting. So, you've got what are really sort of enterprise level um features there. I was now looking at some of the um monthly costs for things like um enterprise level um chat GPT or or Grock. I think they're around $400 a month. Does that sound right? >> $400 a month for for Grock at the enterprise level. Yeah. >> Is that like per user or per company? I can't remember. >> But yeah, I know like Claude, for example, I believe they have like a $200 per user per month um offering that's kind of near the high end. Uh same I think similar with with Chat GBT. So that that sounds sounds right in the ballpark. >> Yeah. So, and these these lower price ones that are like 10 or I mean 20 or $30 a month, they have that problem you were alluding to where you upload a certain number of documents and you ask >> by the time you get to your 10th question, it starts to run out of steam. Basically, it starts to hallucinate and >> just can't answer the question or whatever. You got to start a new chat, upload your document again, and and go back to um to what you were doing. and then it doesn't benefit from all the last 10 answers that were already generated, >> right? >> So, so you're saying that with yours you you're um you can just keep going and going and going that that won't happen. >> Yeah. So I think you know the the main benefit of Raana is you're able to have accurate responses over a vast amount of of data where you don't even know as the as the human like which of these documents contains the answer that I'm looking for. And that's a massive value that that we're bringing to the table with Raana is like you can just dump the whole data room into Raana and Raana sifts through it like a like a human would and find you know what are the right folders what are the right so I think that that absolutely is is is critical and to your to your second point about kind of remembering the context of of conversations I would say that's kind of a broader point about can AI handle scale does it does it scale well right if you ask it to analyze 10 documents. Is it actually reading through 10 documents or is it just reading through the first document and maybe summarizing the the the rest of the nine documents? Right? So with Raana, we've built our platform to be able to uh work at scale accurately across you know all 10 documents, all 20 documents, all 100 documents. Um and there's, you know, specific techniques that we're using and and kind of workflows and and user interface that we're using to to uh enable that. Um, but that brings to the table a much deeper analytical capability than what you'd have from chat GPT. >> Yeah. So, would this allow you like in a normal circumstance, you might just be looking at their audited financial statements, you know, let's say it's a 20-year-old company and they have audited financial statements for the last 20 years. with AI, do you think due diligence will then go further and people say, "Well, now I can look at your bookkeeping records for the last 20 years, >> hand over hand over your QuickBooks for the last 20 years." >> And then you can sit there and say to AI, "So, is there anything uh I should be concerned about like uh are there any, you know, anything suspicious?" Right. And the AI can presumably start looking through all of this stuff. Do you think that'll happen? Yeah, I think M&A will continue to get more sophisticated, right? Um, in terms of the the amount of information that's being processed and um, like you said, like some of that that might be in the form of more detailed financials, that might also be in the form of um, diligence that is not in the data room and maybe it's uh, you know, doing a few extra expert calls to really learn about the industry from an outside perspective. Um right it might be kind of thinking through questions about the industry's growth. A lot of a lot of topics that you know really deal teams today are are very time constrained. they're human capital constrained, right? And so any efficiency they can gain in analyzing the information that that is presented enables them to then go deeper, right? And and that's where really like alpha comes from and I think that will become you know the standard sort of way of investing across across the industry. >> Okay. So uh your AI is also going out into the world and able to read all all the documents that are available to >> Yeah. So we are integrating with >> yeah that's that's something that that we're currently developing u at the moment and uh the ability to integrate with you know third parties you know trusted verified sources of data like expert call libraries private company information from from sources like pitchbook or crunch base financial information which um is public which we already have in our system from SEC filings earnings transcripts and and of course the the web uh really presenting that kind of holistic view uh that blends the the proprietary confidential data room information that is highly specific to the target with the broader space in which the target operates. I think that that will be the future of due diligence and that's where our platform will evolve. >> Yeah. Because each company is in a different sector potentially. So unless you're just doing the same sector over and over again, um, which case you just use the same checklist over and over again, but if you have all these different sectors, all of a sudden it's like, oh, now I have to become an expert in the airline industry, or oh, look, it's a hydroelectric dam. And you you want to know, okay, well, what are the things that are specific to that specific industry? But better yet, um, I would want to be able to score the deal. And you could ask the AI to develop some sort of a score for the, you know, the quality of the due diligence, how reliable it is, uh, you know, analyze these things like the the the TAM and the SAM and >> and decide whether or not you want to go through with it because they would have to make a compare comparison with other deals that are disclosed in Crunch Base or these other databases you're talking Absolutely. And look, I think I think there's there's a a pretty exciting world that that is going to be created by the use of these tools in in due diligence. Um, I think, you know, there's there's what we're working on already is like being able to mine, you know, all the data that a firm has has accumulated on past, you know, deals that may or may not have transacted, but it's it's valuable information that serves as a as a benchmark um on on future deals, right? And and portfolio companies. So all of the rich kind of institutional knowledge that a firm generates over time is a source of competitive advantage and u is something that AI really helps to to to really um you know bring bring forward in an accessible way. Um and I think look there's there's a there's a spectrum in terms of where we are today with how people are using AI especially in the professional sense. So I think you know on one end of the spectrum it's very human in the loop. It's, you know, the human has a specific workflow that they're, you know, currently doing and it's how can how can AI streamline that specific workflow today and then I think over time it's, you know, trusting the AI and as AI gets better to do increasingly open-ended uh tasks, right? and being able to just say, "Hey, these are the qu here here are the risks here. Here's your, you know, your deal thesis, but here's what I found from the data room that doesn't quite line up or here's the confirmatory evidence that I found. Here's what you should look into and almost kind of serve as a a orchestrator for the deal." Um, I think, you know, maybe we're not there yet today, but I I think that that that is a a a future state that that we are advancing towards, >> right? And you want to just get to the right price as well. It's not necessarily to kill the deal to look for things that are wrong. You just want to make sure that you're you are aware of what the value is. So to get to that, you know, price discovery process, I suppose doing this due diligence is part of the process, I think. Um, now what about some of these big companies like Goldman Sachs, uh, Morgan Stanley or whatever? Are they are they working on their own AI? Like are they are they so big that they make their own in-house or are they still going to be looking to um, you know, just like license this kind of software? >> Yeah, so that's that's a great question. I think undeniably these these massive institutions have a lot of proprietary data that is valuable and um you know very very useful to be able to unlock that through AI. Uh I think some of these big firms have have attempted init have made initial attempts to to develop their own AI. um you know without going too too deep into the specifics right a lot of these were were just kind of um isolated instances of chat GPT in the early days and they were and from an enduser perspective they performed uh less good than than just your regular chat GPT. So I think at this point in time a lot of those those firms have you know probably realized that look like in order to actually build a a a good product that involves you know a level of focus on you know building the right features on the user experience but also on you know the the back end of actually you know the infrastructure of how these the AI works and the technology is just advancing super quickly. So, you know, for firms that are their bread and butter is is, you know, as it whether it's investing or or banking or just, you know, being a financial institution, it can be very challenging to to keep up with what's the latest and actually develop the product that um is bestin-class. So, I think what you're seeing now is those those institutions realizing, okay, it probably makes a lot more sense for us to to to kind of use a a um sort of a AI native, you know, best-in-class solution um just like they they do with, you know, every other other uh you know, software that they currently use, whether that's Bloomberg, whether that's Cap IQ, right? uh Goldman Sachs doesn't necessarily ha you know they they do uh develop some software internally but um for a lot of you know those those core pieces of of data and workflow that that their their workforce uses it is heavily reliant on kind of external data and I think AI will be the same. >> Interesting. Okay. Well, this is all very important stuff for uh people to know and you know, we've got a series on M&A uh here uh on the Executive Suite podcast. So, you'll be added to that um on the playlist uh for M&A. I think uh this is probably not a bad place for us to wrap things up. Do you have any um closing thoughts that you wanted to share with us before we say goodbye? >> Yeah. I know Thomas, it was it was great chatting with you and um love love [clears throat] having the conversation. Uh yeah, would love to uh you know chat with folks who are in private equity who um want to streamline their due diligence. Right. Our our website is ravana.ai and uh we're always happy to to have conversations with folks. >> Okay. Well, that sounds great. Well, everyone who's been watching, thank you very much for being with us. And uh if you like this kind of content, uh please consider liking and subscribing this video. And we will see you later. Bye for now. [music] [music] >> [music]