Reinventing Wall Street: Rogo’s AI Revolution with Gabriel Stengel

Hey everyone, welcome to Generative Now. I am Michael Magnano. I am a partner at Lightspeed, and this week I'm talking with Gabe Stengel, the founder and CEO of Rogo. Rogo is a platform that uses AI to do what junior bankers usually spend sleepless weeks, days, and nights doing— analyzing earnings, benchmarking companies, and building pitch decks. And it does it all in seconds. Gabe was a former analyst at the investment bank Lazard, and so he saw firsthand just how inefficient finance can be. And so he and his co-founders built Rogo to fix that. Now the world's top banks and hedge funds are using the product, and investors are betting big on their success. Gabe and I talked about his time working in investment banking, how that led him to the idea for Rogo, and where he sees AI fitting into the future of banking moving forward. Let's dive in. Hey, Gabe, how's it going?

Good. How are you? Thanks for having me.

Doing well. Thanks for doing this. Been really looking forward to this and talking to you about Rogo, obviously. Well, I don't know. How long is it?

How long has it been?

When did you guys found the company?

Late 2021. So, so 3 and a half years-ish. So, okay, cool. It's been a little while.

From what I know about it, it's got a pretty cool inception story, especially given your background as an investment banker. Um, tell us a little bit about how you guys got started.

I mean, I guess actually the, the LLC that we rolled into the C corp that is Rogo was started in 2019. Oh, wow. Because John, who's my co-founder, is one of my best friends at, at Princeton, where we both went to school. And we collaborated on our, our senior thesis, which was a sort of research project. Um, I was computer science, John was econ, and we basically jointly created an AI chatbot for financial econometrics. It was not very good. It was using old hat techniques, semantic parsing, context-free grammars, but it was very exciting to us. And it was a tool we could put in front of our less tech-savvy econ friends and have them use it to, you know, do their theses and their work. And so that's when we first had the idea and got excited about this idea of a natural language interface. We tried to commercialize it at the time and couldn't really figure anything out. So we ended up both going into Finance. I was an investment banker at Lazard. John went to Barclays, then JPMorgan. And my experience at Lazard, I got, you know, pretty— I had a pretty unique position. I got to be both on an investment banking team and then later on a data science team, helping find internal tools for the bank to roll out to make junior bankers more productive, senior bankers a little bit smarter. And so just had unique visibility into, you know, both what the work junior bankers were doing were, but also, you know, how were these institutions even thinking about solving it? Um, and we left that job and those jobs and got started with building this company.

Tell us a little bit about, uh, a word you used early on in your story, econometrics. Explain for, you know, the layman viewer, you know, what are econometrics? And also, how is the product— well, it sounds like it was sort of rooted in some early work you did in Princeton where this was really focused on econometrics to sort of where you are now. And, you know, we'll get deep into the history of the product and all that, but, uh, would love to understand, like, There's absolutely no econometrics in the product now.

Back then, our product was really built for postgrads and undergrads doing econometric research, i.e., you know, the simplest version is run a regression of X on Y or run an autoregression or find an instrumental variable. I'm helping economics postgrads do, you know, do their work. I mean, it's, it's not really what we do now. Today, the product is much more helping investors, investment bankers do research on companies, diligence companies, come up with unique investment theses. Uh, we also still analyze structured data, but instead of analyzing economic structured data so much, we analyze company structured data, right? So you might say, uh, run a regression of revenue multiples on margin expansion at big healthcare companies to see, you know, what's really been driving valuation over the past year. Um, or just show me, you know, the share price returns year to date for this basket of companies. Um, but yeah, we, we don't really do that econometrics as much anymore, but we had to put a spin on the, the project to make sure that John could log it as his senior thesis. So we had to do econometrics. Nice.

That's so funny. Talk a little bit about your, your experience at Lazard, and it sounds like your co-founder had some similar experience that sort of like shaped the vision for what you wanted to do with Rogo. I mean, at Lazard, were you doing the same, you know, types of analysis and research that, that you imagine users of Rogo doing now?

Oh, 100%. Yeah, I was a banking analyst. I was putting together you know, models for, for potential buy-side acquisitions. I was putting together PowerPoint decks. I was doing research on industries. I was putting together PIBs, public information books, which if you have any banking listeners on this podcast, they will, you know, understand what that is all too well. Yeah. What's it—

what's a PIB? What is a PIB?

Tell us that. A PIB is basically a senior banker says, hey, we're going to meet with, let's call it, you know, DataDog. Give me a PIB on the company, i.e., put together all the resources I would need to get smart on this company over 24 hours. So you pull— it's a— each MD has their own formula, but maybe the last 2 10-Ks, last 10-Q, some initiating broker research coverage, some recent broker research coverage that's good, any sort of interesting news articles that have come out recently into one, you know, 200-page condensed PDF that you email to them, and then they can leaf through and, and see if there's anything interesting or, or write you a follow-up. Um, so So it's just an information packet to get started, which now is a little bit dated, right? Like you don't need this 200-page research packet when you can just go into a tool like Roga or ChatGPT and say, hey, tell me about Datadog and what's been going on. But yeah, I was, I was doing the, the, the very basic banking work. Um, and I loved it. I mean, I, I actually, I think there's this, uh, this perception that no one likes investment banking. I don't think that's true at all. I mean, I think you get access to some of the most interesting executives, companies, see how they think. see how they think about M&A capital allocation. And I really enjoyed it.

Putting together a PIB, especially without AI a couple of years ago, I mean, it sounds like a ton of work, especially if you're telling me these things need to come together in like 24 hours and this is a 200-page PDF. I mean, was the initial thinking like, hey, this is a lot of manual work? AI is obviously helping to sort of give people leverage on work and units of intelligence. Let's make this like a PIB creator. Is it— was that like sort of the basic idea?

Yeah, I would say it's funny. I mean, we— the basic idea was, Yeah, there's all these types of work like benchmarking revenue multiples, putting together PIBs that are so basic and not interesting, and there should be an AI tool that can do it. But I would say the thesis then, and the thesis still now, was always more so how do we empower people to do things they can't do? And that goes back to our original senior thesis, which was giving our peers the power of writing Python code through a natural language interface chatbot to do their econometrics work. And even now, I want to make sure we're doing all the mundane work that junior bankers and junior investors do. But I also want to empower them to do things they've never done before. Right?

Right.

I want to help them do all sorts of new types of analysis, new types of insight-generating work, uh, that wasn't possible before. And so even at the time, I think we had loftier aspirations. It took us a while to, to even get to the mundane automation. Um, but it was always focused on, hey, how can I, you know, give you, Mike, a tool that allows you to do the types of analysis that before you couldn't really do if you didn't have a team of data scientists or, or a team of juniors underneath you.

Got it. So it almost seems like the, the thing that really becomes accessible with a tool like Rogo, at least in the early days, was sort of almost like the data science function and like the real deep analytical work, not necessarily the stuff that, you know, as you said, you can do a lot of this stuff in ChatGPT now with, with, through like a deep research query.

Yeah.

Deep research, I'm sure, isn't doing a lot of the data science work, but it's probably doing some of like the basic, just like pulling of, of, of existing data and records. to keep somebody like just generally more high-level informed?

I mean, look, Brogo is in many ways like deep research except plugged into financial data sources and financial tools. And so instead of only having access to the web and a Python code interpreter environment, our deep research has access to the bank's internal data, their CRM, their SharePoint, their precedent transactions analyses. And then it also has access to tools like Excel, like filings, like FactSet, CapIQ, PitchBook, Uh, all the financial data sources that, that they use. And so it's, it's basically just Deep Research that's been trained to be, you know, a Wall Street analyst as opposed to just anyone's research analyst. Yeah. Yeah.

So, so, all right, let's, let's talk about that a little bit. So you mentioned things like PitchBook. These are, you know, these are data sources that are out there on the web. You have to pay for them, but they're out there. And then you mentioned sort of like the company's data. Is, is, is that, is that really like sort of the secret sauce there and that it takes your internal data that's proprietary, and then it links it up with all this stuff that's on the web, but maybe hard to get and something like ChatGPT can't just go and grab.

I would say there's 3 things that make Rogo special and are our IP and make it a valuable product. One, yes, is the content, the data that you have access to, right? We work with providers like S&P Global CapIQ to make sure that we have, you know, if they, if we have mutual customers, they can query the CapIQ data through Rogo. That is a business data licensing problem. The second is the actual tools that the model has access to, right? Even if you have this ingenious LLM, if it doesn't have access to a tool like Excel, can it really, you know, build a full 3-statement model? It can in Python, but that's not auditable and it's not really built for that. Can it make a PowerPoint output? Maybe, but probably not the one that's up to your, your output specification. The final thing that's different is our reasoning and the actual model quality. And the same way that, you know, if you look at Claude 4 or Gemini 2.5 Pro or, you know, GPT-04 or 03, they're ingenious, super smart models, but they don't know how to use the tools, the datasets, and do the workflows that bankers and investors do. And if you had, you know, Albert Einstein start at Lightspeed on Monday, just because he's smarter than everyone at Lightspeed, which may or may not be true, doesn't mean he'll actually be a valuable—

Maybe.

Well, maybe, uh, I don't think he would be valuable on day one. He needs to learn how you guys think about a company, how you think about investments, how you use tools, what your formatting is. Right. And that's actually learning. Uh, and so we teach the model how to do that by, by using a technique called post-training or reinforcement fine-tuning to learn how to use the tools that finance professionals use.

So, so these are, these are your own models. Uh, you're not using, um, you know, GPT?

Well, we build on top of those. So it's like, okay, a lot of these platforms now offer fine-tuning on top of their models. So you can fine-tune 4o, you can fine-tune 2.5 Pro. You can also distil these models into open source and then run reinforcement learning. So we're not doing any pre-training, but we are then doing post-training on these models.

Yeah. And you mentioned that you're fine-tuning them to work with the tools that people in this industry are working with. Talk a little bit about that. Do you mean actually sort of agentically using these tools?

Oh, yeah. If you were an analyst at Goldman and you started on Monday and someone said, hey, find the precedent transactions for the Microsoft Activision deal, you would probably go to EDGAR, which is the SEC's, you know, list of filings, and start looking. Microsoft Activision, look for the DEF 14-A. Okay, I found it. Now I'm going to Ctrl+F through the DEF 14-A for the precedent transactions. Okay, great. There they are. That's not a tool that an agent has access to.

Right.

I think we more and more see agents that have browser capabilities and so could do that hypothetically. That's a lot less efficient than just giving an agent access to a filing search tool where it can put in the ticker of a company, filter by a filing type and a date range, and then Ctrl+F through those documents. And so we both build the tools and then we teach the model how to use it. And so for each different type of tool, we actually can have specialized models that query those documents more effectively, right? They do the right keywords or a specialized tool for, you know, formatting Excels. And so we have to teach the models how to use these in a much lower latency, much higher accurate accuracy sort of way.

Yeah, I'm sure you're asked this all the time. Uh, you know, common question for, I feel like, any AI startup, you know, how do you think about the role of Rogo in using all these tools and accessing all these data sources that we talked about versus the role of an OpenAI, uh, an Anthropic, you know, an xAI that, you know, probably want access to many of the same data sources and, you know, maybe over time have the capability to do you know, perform many of the same agentic tasks.

The better the next generation of models is, the better Roga will be, right? We can very quickly take those models and take that intelligence and, and have it build into our system almost immediately based on the way that we architect. So it's great for us. And the difference will be we are spending all of our time on financial tools and financial workflows. And so when GPT-5 comes out and we already have access to all this tool data and how they're used most effectively, And evals over how to do these financial workflows, we're going to be able to more quickly distill that into an end-to-end system that can you know generate a sim, generate a PIB, analyze an S one than they will. And they'll have to collect those evals and create those tools over time. And you know finance is a big enough market where they will likely start to focus on it. But there's been more focus on sort of generic consumer applications competing with the Microsoft Office suite coding. And I think finance is a domain where. You know, there's a lot more intricacy from the product layer and the workflow layer than people expect. And so I think we're going to have, you know, a distinct, distinct advantage for a while. And ideally, by the point where they're turning their attention on finance, we have a lot of, a lot of actual user data and evals and a sense of what the workflows are such that, you know, we'll always be a step ahead.

Yeah, that seems to be the big debate, right? You know, where obviously the word moat, you know, we talk about so much in AI and it's, you know, it's constantly this, this question of, oh, you know, do you want to be the model provider? Do you want to be at the application layer? And I think, you know, the, the logic that, that I hear often, and I frankly, I agree with, and it sounds like it's similar to yours. It's like, well, you need to specialize in something and you need to run really, really far ahead. Right. And, and, and as you said, unlock the proprietary data and, you know, lock in the, the, the different workflows that are going to be really, really hard for someone else who's very horizontal to specialize in. Is that, is that kind of how you guys think about it?

Yeah, 100%. I mean, I actually think more and more so the value is in the application layer. Even if I look at my own usage of ChatGPT versus Gemini versus Claude at this point, because I have a history of threads on ChatGPT, it responds in a more personalized way that's really helpful. And I don't need to go in and explain to Gemini, hey, I'm Gabe Stengel, I have a company called Rogo.

Right.

This is what we do. Use that context to answer the question. ChatGPT does a very good job of, of personalizing responses, recording what I'm doing, you know, having a good sense of, of the types of answers I like. And now it's kind of a pain for me to switch.

Totally.

Um, and that even if the model is slightly worse, I mean, I think that's a, that's an advantage of application and product layer. At the model layer, I mean, I'm not, I'm not an expert on these models at all, but my sense is, you know, no one's ever gonna be more than 6 months ahead of, of anyone else. Um, and so the, the distinct— distinct advantage you'll have as a model provider is just, you know, how, how low are your costs? How reliable are your, your services? How low latency is that kind of thing?

Let's go back a little bit. So we talked about, you know, sort of your role as an investment banker. We talked about some of the early theses. You know, we talked about when you got started. Take us back to that moment. You know, it sounds like 2021, 2022, obviously ChatGPT has sort of like just burst onto the scene sort of at the beginning of 2023. What was it like then starting this company and, and maybe even like, you know, talking to investors and your first customers about what you were trying to do at a time where people had far less understanding about AI and, you know, the, the impending implications than, than they do now?

So we started the company before ChatGPT came out. GPT-3 was out. And so for those paying attention, you know, it was obvious that this was a, was a profound leap in, in technology. But, uh, all the banks and hedge funds and private equity firms we spoke to, you know, I think they thought this guy Gabe is full of snake oil, not this is going to be a— this is going to change the way we work. It was, it was a long time until the market was receptive to what we were building. And it was frankly also a long time until we sort of figured out what exactly the market wanted too, right? It's easy to say, great, let's make people smarter and more efficient with AI. How does that manifest itself in a product? What are the use cases like? How are people going to use it? How important Are, you know, these use cases versus those use cases? How do you pitch it properly? Who do you sell it to? Those were all the things that we slowly had to learn over our first, let's call it 18, 24 months where, you know, the market was not that receptive to us.

What were you promising them? Like, what would you, what would you say to your, your early customers?

I mean, there weren't that many early customers. Yeah, we didn't have real customers until 2023, late 2023. And even then, like, you know, you as a founder, you're trying to get get someone to use what you're building at all costs, right? And so you'll, you'll do a whole song and dance just to get someone to be hands on keyboard in your platform just for the product to break after, you know, 25 minutes. The initial pitch was, hey, what if you could ask questions over your data and get answers in 5 seconds as opposed to using a clunky interface like one of these old tools like Bloomberg or, you know, the EDGAR SEC filings database or, you know, pinging an analyst on your team. And so that was the pitch and it was pretty simple. Yeah, but it was, it was hard to build and then hard to actually expect what folks were going to ask. Right. When you're pitching a generic search or chatbot, it means, you know, you have to build everything in on day one. Otherwise people are going to try, you know, the one thing you're expecting and then the 5 things you aren't and the product is going to, you know, not suffice. Right.

What about the early investor pitches? You know, what were you telling them and what resonated and what didn't?

Kevin Ryan, who led our pre-seed round from a firm called Ally Corp, um, based in New York. But, but Kevin himself is sort of a, you know, self-proclaimed godfather of New York City tech, you know, started MongoDB and Gilt and Business Insider and was the CEO of DoubleClick. Um, he kind of got it immediately, but his function, his routing function, I don't think was AI is going to change the world. I think his routing function was Gabe and John are 2 guys who are smart and understand technology and intimately understand investment banking. And, you know, being from New York, I know investment banking is a is a domain that is, is ripe for disruption.

Did you know him from Mongo?

I knew him actually from, from growing up in New York. So I had actually met his, his children a long time when I was going to school here. And then I met him again at Mongo. Yeah. So I interned at Mongo and he came to speak and I would send him, you know, a few one-off messages. And then he actually got in the routine of, of, you know, and to his credit as an investor, every 6 months he would send me an email being like, hey, what are you thinking about? Do you still like Lazard? You know, what are you, what are you doing next? And so when I had this idea, Well, at Lazada, he was the first person I messaged, and it was a very natural, natural start. That's awesome.

Yeah, yeah, that's, that's really cool. You talked early on about how early on it was, it was tough to get customers to sort of take this thing seriously. You said you didn't really have any. What was sort of the moment in the product that sort of flipped to, you know, product market fit and getting your first early sticky customers?

There were 2 core things. One was changing our focus from just structured data to structured and unstructured data. So from— because the original product was like, was econometrics on structured data, that's what we pitched, right? It was, hey, make a chart of revenue multiples, you know, create a scatter plot of this versus that. And we weren't taking advantage of what LLMs are best at and most magical with, which is unstructured data.

Yeah.

And we were too stubborn before, you know, opening the aperture. The second thing was We, I just changed the way we were pitching the product. I mean, I distinctly remember being on a call with a large private equity firm and it going, you know, the product demo going better than I'd seen a product demo go in a while, in part because we'd added this unstructured data lens. And they asked for pricing and I was like, well, firms this big never ask for pricing. They normally like race to get off the call. Right. Of course. And so I have no sense of what to charge. And so I took, you know, 10 seconds in my head and thought, okay, you know, I'm going to, let's, let's push the full-court advantage here, $2 million a year, right? Just pull a number out of thin air. And they go, okay, makes sense. And I— and as soon as I got off that call, I called my co-founder and I said, hey, you know, this firm didn't laugh at that number. This— we might be on to something. You know, if someone hears $2 million a year and goes, oh, that, you know, that could make sense, it means you're pitching something that they think is very valuable. Um, and so it was those 2 moments in changing how we were pitching and what we were building that really started to unlock customer traction.

The unstructured data. I mean, I think that makes sense. I mean, you're seeing this in a lot of products right now where not just the use case of it and the value it provides to be able to just like throw a bunch of stuff into the model and have it come back at you with something structured, but also just like the time to, you spend building an architecture and a database structure and like the old way of doing it, I would say like the amount of time and cost that must save your startup. Is insane. And we see this a lot of startups.

Yeah. No, I mean, to query structured data well with an LLM, you still need a, you know, a beautiful database that's documented well with data dictionaries. To query unstructured data well, I mean, you need some Elasticsearch indices and like a, you know, smart keyword search prompts.

And a big context window and like, you know, exactly. Just, just like throw it in there.

Yeah, exactly.

When you're talking to this big firm and you're like $2 million, Clearly, that's a big contract, especially given your stage.

Didn't end up being $2 million, I will say. But I'm glad it was, it was tethered, anchored to that as a starting point. For sure. For sure.

You know, I'm guessing what, you know, what they start to think and customers at their scale must think is, okay, obviously, you know, this product has value such that we're considering paying for it, but it also better have things like rock-solid security, encryption, all this stuff that you associate with, you know, big, robust financial tools. And at the time, you're, you're a small startup, you know, you're investing investing in features and moving quickly, how much are you thinking about that enterprise-grade stuff at the same time?

So we, I mean, we were forced to think about it early on by the nature of our customer set, right? Like we were forced to think about what it meant to be multi-cloud, what it meant to deploy on-prem, what it meant to actually have these enterprise-grade, you know, permissioning systems and RBAC systems and governance systems. And so we, I mean, we had to do that much earlier than a startup would in a typical lifecycle. goal. And it slows you down, right? Like it takes attention away from feature velocity and into enterprise stuff. But it's a prerequisite for us. And I think it was, it was also a mindset thing, right? Like we didn't think of security as a side dish and the main dish being the product. We thought of security as one of the main value propositions about what we were building. And so if you treat it like that, you know, you treat it like the product, right? And it's not, oh, we have to build out this enterprise permissioning system. It's, oh, this is a feature. Like this is something our clients want and it's just as part just as, you know, much a part of our product market fit as, you know, the way that we respond to user questions is.

Yeah, got it. Makes sense. Did you have that capability on your team? Like, how do you go about solving for that? I mean, are you— you hire a CISO and— yeah, tell us about it.

You figure it out. I mean, okay, it's like all of these things. You— yeah, trial by fire. Can you execute? Bring in the expertise where necessary. Otherwise, you know, like a lot of engineering problems, you can think about them from first principles and there's a lot of resources and you can, you know, work with your partners and stakeholders to solve them.

But it's interesting, like you said, it slows you down. And the name of the game, it feels like right now for all AI startups is speed. I mean, we even talked about it early on. In order for a company like Rogo to succeed, you've got to race far enough ahead of the model providers before they decide that finance is actually a vertical they care about. How do you balance that? How do you, how do you move fast In a world where the cost of moving fast and making a mistake in your industry is, is really, really high.

I mean, we move slower than if we were a consumer company, for sure. From a feature, from a feature perspective, like you just have to. I mean, it's a sort of a, you know, threading the needle constantly and thinking about tradeoffs constantly and allocating as much bandwidth as you can to all your problems at the same time and just staying disciplined. There's no, You know, there's no silver bullet for execution. It's, it's, it's execution.

Yeah, totally. It's, it's tough. I mean, it's, it's crazy the pace of, of innovation and shipping you're seeing from startups right now. And it's, it's, it seems like it's, it's just by necessity, right?

Yeah. Well, and it's also thinking about, look, if I was a horizontal company, like let's take a Glean, you have different tradeoffs and you say, hey, am I willing to build an on-prem version or a cloud-prem version or build in this security stuff? Probably not, because I'm racing more against ChatGPT, and I'm going to be slower to these other verticals. Whereas if you're a finance company and the race is to get into the largest financial institutions, well, focusing on security is actually the thing that allows you to move faster and get into customers' hands rather than the thing that's slowing you down. And so it depends on what sort of company you are. And for us, it's all about the trade-off of like, you know, is this functionality and enterprise-grade security actually a speed-up in the near term and the long term, or is it something that, that's going to slow us down? And I think for a company like OpenAI, like a lot of the security stuff is a slowdown, right? Because they're not competing with us selling into large investment banks. They're competing with, you know, Perplexity and Copilot and Anthropic to get to all consumers. Yeah, totally.

So, okay, so now you've got a bunch of customers, you're, you're post-PMF. Walk us through maybe the day-to-day for some of your customers in terms of some use cases, like the people that sit in Rogo all day, like what are they doing today?

So I'll give you a great example, which is when I was an investment banking analyst and we were, you know, showing a bunch of great potential acquisitions to a large healthcare company, you would have a list of 10 company profiles in slides and you would say, hey, look at, you know, this company, look at this biotech, look at this research company. And you would say, this is their CEO, this is what they focus on. This is why it's strategic to you. This is the, you know, what their snapshot financials look like. And that would take hours and hours and hours. One, because, you know, if I'm— let's take some random biotech name. Maybe I'm looking at a company like Regeneron or Vertex. I don't know what they do. I'm an investment banking analyst. I'm not, you know, an oncology expert. And so it takes a while for you to read, synthesize, figure out what's in their R&D pipeline and write it up. In Rogo, you can say, hey, take these 10 companies, make 10 PowerPoint slide pages, and you can get an output in 30 seconds. And so that saves a huge amount of time. On the other end of the spectrum, not just junior bankers, I mean, there's senior investors and bankers who go in and instead of, you know, instead of having to ping an analyst on their team and say, hey, Gabe, I need a PIB on Datadog, they can just go into Roga and say, hey, news run on Datadog. Okay, cool. What are they doing in this market? Why are they down in that market? What has their product focus been? What's their M&A strategy been? And just have a thread where they get answers that they can work with immediately and dive deeper instead of asking for some deliverables just to help them prepare for a meeting.

Yeah, that makes a lot of sense. The thing that people have been talking about across every industry, but also sort of finance and capital allocation is, you know, is AI going to replace X? What do you think of sort of the role of a product like Rogo in ultimately displacing sort of human-powered work that it augments today?

It's going to replace workflows. I don't think it's going to replace people. Yeah. In the same way that You know, that, that workflow I describe of creating a PIB for my managing director, 10 years before me, you know, you would create the PIB and then you would go and print it out and bind it and go deliver it to your managing director's, you know, apartment at 11 PM so that they could read it in the morning. You're not doing the printout work. You're not doing the binding. You're not delivering it.

Right.

You're still working pretty hard. You've filled up your time with other things, right? Like, these are institutions that work hard. They want to you know, compete with each other. They're not trying to rest on their, you know, rest on just healthier margins. There's going to be more and more work. That said, it's going to be different work, right? Like you might not spend as much time data cubing and preparing presentations and like, you know, in the thick of an Excel and you might spend more time thinking about a company or a problem. You might spend more time meeting with stakeholders and founders and executives, uh, forming those relationships.

One of the industries I see parallels and similarities to, I wonder if you do as well, is, is, is legal work and law firms. You know, one of the things we see And we hear about law firms, big law firms, like the really big ones, is, you know, they're, they're taking all of their data, you know, their years and years of contracts and sort of variables from within those contracts that they have sort of unique proprietary insight into. And they're, they're building their own models, right? They're building their own tools. They don't want to give that stuff up to a, to an OpenAI or an Anthropic to build tools that'll ultimately displace them. Are you seeing the same thing happening? from big investment banks, the Lazard's of the world? Are they building their own tools? Do you view them as competitors?

I think a lot of professional services organizations know what they're good at and know and, you know, have a sense for what they're not as good at. And building really strong technology and AI models and products is not, you know, what has, what has made an investment bank great. It's the relationships, it's the business acumen, it's the, you know, the ability to negotiate and get a great outcome for, for a client. I think we're going to work with those sorts of banks to produce tools that mutually enrich us and make them more effective and help change their business models. But I think it's much more of the sort of partner scenario than the either build yourself or just buy an off-the-shelf tool. And by the way, I think that's been core to some of the large legal AI companies as well, right?

Yeah.

If you speak to someone like Winston at Harvey, a large part of his vision is working with law firms to produce products that they can jointly offer to customers.

Yeah, that makes a ton of sense. Yeah, I imagine there's a tension, right? And, and again, I, I see this with law firms. I'm, I'm, I'm not even speaking to investment banks or rogo, but it seems like there's a tension when we go and we talk to lawyers at some of these big firms. It's like, oh, of course we know AI is coming, but we'd rather use our own tools and our own data rather than, you know, some startup or even again, one of the big models who is clearly like specializing in this stuff.

There definitely will be a tension. I mean, I think. I think that folks in finance have not kind of woken up to the reality that like more and more of their day jobs are going to be automated in the way that folks in legal have.

Yeah.

And so I think it's partly just going to be a mindset shift over the next coming years, 5 years. And then I think the other reality is like people, you know, firms have not really seen job replacement yet, right? Like, you know, there haven't been big banks or big professional services organizations or big companies of any size, like, you know, Apple or Disney or Microsoft, who have really replaced jobs with AI yet.

Right.

Um, and so I think until we start seeing that, you know, these, these firms are going to start, keep, continue functioning as business as usual, but maybe with slightly healthier margins or a slightly better product. And when you start seeing the ability for AI to really automate human work, well, then these professional services organizations are gonna have to take a step back and say, well, how does this change my business model? And that's when I think we'll start to see the tension you're talking about.

Yeah, I totally agree. And I think there's probably— I mean, look, I don't know anything about running a law firm or an investment bank, but there's probably an opportunity right now where if you run ahead, you can totally just like, you can arm your staff with these superpowers and like you said, get massive increases on your margins. It almost makes— I don't know, if I was running a law firm, I'd probably do that right now.

Yeah.

No, I know.

But there's also people who are starting to build law firms or banks. But, right, you know, from the bottoms up with technology, there's a great company called OffDeal started by a friend, Ori, who was an investment banker. And his idea is, hey, how do I start, you know, the next Goldman Sachs just with technology and have fewer bankers? So I think people are going to attack it from both ends, from, you know, large incumbents who figure out how to digitize their workforce versus, you know, upstarts that figure out how to just build AI natively from day one. But not as tech companies, but as professional services organizations.

Yeah.

And I mean, this is part of the trend you're seeing where, you know, VCs are levering up and buying, you know, typical sort of cash flow private equity businesses and then tech enabling them.

Yeah, these roll-ups. Yeah, totally. What do you think about, you know, looking ahead, what do you think about the future of Rogo? How do the types and flavors of your customers start to change and evolve? How do the use cases start to change and evolve? Right now it's like investment bankers, PE, etc.? Do you imagine a world which in the future it's anyone who's doing financial analysis, you know, finance departments of every flavor of company? You know, help us think through that.

Yeah, I get asked about this sort of internal corporate FP&A stuff a lot. My really— maybe I will regret this when someday we're, you know, a huge company and we're, we're eking out more TAM in an Investor Day presentation. But like, I'm not that interested in like FP&A departments at corporate. Okay. Right. Like, that's interesting. That's less exciting to me.

Why?

Helping— it's more like accounting workflows than it is, you know, thinking like an investor would about what companies are exciting, what are the great business models you want to invest in, who are the founders you want to back, or as a banker, what's the most strategic way for someone like Bob Iger at Disney to think about capital allocation or acquisitions or the land or the, you know, sort of the market.

So it's more like his corp dev team than his—

Yeah, exactly.

Yeah.

But that said, I mean, look, I, if when I look at the companies I'm— some of the companies I model us off of and most admire, admire companies like Bloomberg. Bloomberg never said, okay, great, you know, we've had— we have a good enough hold of the finance market. Let's start selling to everyone. They said, no, let's continue to innovate for this one buyer type in this one market that has almost a limitless ability to pay for tools that make them smarter and more effective. Right? Like, I would always continue to invest in creating better and better tools for investors and bankers than try and veer off course. Just because I think, I mean, it's the one market on earth where there's an unconstrained ability to pay for something that makes you smarter, right? You can go to a great hedge fund investor and say, hey, I'll make you a 20% better investor. They will pay you whatever you want. Whereas if I go to a corp dev or FP&A team and say, hey, I'll make your life a lot easier, there's a ceiling on what they're willing to pay, right? Like they're not— they don't have unconstrained upside in their decision-making. Investors do. Right, right.

Yeah, that's totally true.

Like imagine if I went to Lightspeed and I said, hey, I'll make you guys a 20%, you know, better investment firm. How much would you guys pay for that? Yeah, we will. $30 million.

Yeah.

Yeah.

I imagine where you could probably see a ton of market expansion and maybe you're already doing this then is in consulting, the McKinseys of the world. It's Accenture. Like, are you thinking about that? Is that already part of what you do?

Yeah. No, we already have a few consulting firms on the platform. We actually have one of the big 3. who is rolling out the product not just for their front office consultants, but for the, you know, sort of the research teams they have internally, which is kind of a funny thing to me because you would think that Rogo would replace the— replace research teams that service consultants rather than be used by research teams to further help consultants. But yeah, I think that sort of professional services work is right sort of in the bullseye of what we can do.

Totally. So when you, when you go to sell this product into any of these organizations, professional services, investment banks, etc., Do you, do you think about it sort of from a pure kind of sales top-down motion, or is there something here where you can actually have a bit of a kind of a PLG flywheel where, you know, one, one banker uses it and then shares the data with somebody else and that unlocks a new seat? Like, help us think about that.

I would love to have more of a PLG flywheel. And it's funny, our, you know, new investors thrive. There's a, there's an investor there, Vince, Vince Hanks, who, who, who's awesome and super smart and has always pushed me to have a PLG product. Bankers, The domain is going to be blacklisted. Like, they have such strict InfoSec policies that, like, if I had, you know, if you could just go to app.rogo.ai and log in, you still wouldn't be able to do it. I mean, you have to, you have to sell the hardware. Things can't be shared, even internally. Yeah. And I think that, you know, there's PLG possibilities on the buy side. At a place like Lightspeed, I'm sure people just go and use ChatGPT on personal devices, and then the firm thinks, okay, great, we may as well get an enterprise account now because we don't really want people using personal devices. But throughout financial services, it's a harder market to sell into from a PLG point of view. I mean, you can use PLG or some product-led things as, as a way to generate top-of-funnel, right? But I don't think you're going to see the same type of sales motion that a Notion or an Airtable has where it just gets some bottoms-up adoption and actually significant revenue traction too. And then you parlay it into enterprise.

I guess all those kind of like sharing and viral loops just get broken, right? Even if you Even if you had a URL and you shared it in Slack, like it just, it wouldn't even unfurl. It wouldn't show you any content. Like you would just have to have all this stuff saved locally.

Yeah, the MD on your team would say, I can't open this. There's a firewall.

Right, exactly.

Yeah.

That's gotta be really hard. I mean, and maybe again, going back to like how you learned security, how did it, how did you as a small team learn sales? You know, how'd you guys do that?

Sales was a hard learning experience.

Really? Yeah.

Tell us about that. Oh yeah. I mean, I, was computer science in undergrad. I went, worked as an investment banker. I'd never met someone who worked in sales. Like, I just had never been surrounded by that archetype. It's hard. Enterprise sales is difficult. Like, it is, it is not an easy thing that, you know, oh, sales is not that, not that hard. You just show the product, you get people signed up. Selling to enterprise is there's 40 stakeholders. They all care. It's political. You have to figure out what people want to hear. What are the pain points? How to thread the needle? It's a difficult thing. I mean, it was all learning.

Yeah.

We were lucky to hire some great, great folks early on who had experience selling into this market, and that really expedited my learning and our success. But it's not, you know, enterprise sales is not something to trivialize.

Yeah. And because you don't have things like PLG, you can't even employ sort of like this land and expand motion where some random person starts using it and then you go and you convince the— yeah, CIO or whatever to buy it for the whole org. I'm guessing you're literally just doing like top-down sales, just like banging on doors.

Yeah, more and more now we can sell like 5 seats to an investment firm and that's a little bit easier. And then we go back, you know, in 3 months and say, hey, you have 5 seats, but these 20 other people have asked to get on too. Why don't you just get an enterprise license? So we see that more on the buy side. For banks, yeah, I mean, you're running pilot, you have big stakeholder conversations, you need to prove out the ROI. Um, it is not the sort of thing that can be sort of PLG level.

Yeah, totally makes sense. So we talked a little bit about what you would and wouldn't do in the future in terms of potential customers and you know, TAM expansion. But, you know, maybe where do you see the company in 5 years, 10 years from now? And maybe like the banking industry more broadly as a result of AI?

I think that 5 years from now, 10 years from now, Rogo is going to be the most effective analyst on Wall Street. I mean, I think it will be very hard to run an investment firm or a bank if you don't employ Rogo, right? If you don't literally have, you know, an army of Rogo analysts helping you be smarter and more effective, you won't be able to, you know, submit a bid in the first round of a transaction fast enough because you won't have this army of AI analysts doing diligence. You won't be able to win the sort of the S-1 bake-off because you'll be a little bit less sophisticated in how you pitch to clients without a tool like Rogo. It'll be every firm's smartest, most productive analyst. The sort of flip side of that is we will also be able to offer financial services to types of companies that beforehand didn't get those types of financial services, right? Like giving M&A advice. Running fundraising processes, that's a type of high finance financial service, right? We don't think of it that way because it's not like a checking account that you can get through Chime or like a financial service, like, you know, being able to buy and sell your own stocks through something like Robinhood. But these are still important financial services that a lot of companies don't have access to. If you have a, you know, $20 million EV HVAC business, who's running your sell-side process? No one.

Right.

So you're getting a 20% worse price when you sell it to some private equity-backed roll-up. If we have the technology to help folks do that, I, I'd love to directly offer it, um, and, and help make private markets more efficient and help democratize some of these high-finance financial services that, you know, not everyone has access to today because you can't pay Goldman Sachs $5 million to, you know, run the process or give you advice or, you know, help you fundraise.

That's super interesting. So yeah, I mean, we talked earlier about, you know, you don't want to get into FP&A, but I think, you know, we were talking about Disney and Bob Iger and we're like, oh, His, his corp dev team could use this. When you look at it through that lens, I think to your point, I mean, this is something startups could use, you know, small businesses, small, medium-sized businesses, again, to get access to the types of services they might only get access to if they're a really big company and have internally, or they're hiring somebody like Lazard. Is that kind of what you mean?

Yeah. Well, I mean, look, if you're, if you're Bob Iger and you're thinking about maybe you're selling the company to Apple, And you're thinking about who should I hire as my investment bank? Should I hire Lazard or Goldman or, you know, whoever? What if you don't need to hire a banker to produce a fairness opinion?

Right. Right?

Like, what if you don't need to pay $25 million to do that? Or what if you can use, you know, let's say Lazard has an AI product that can help generate fairness opinions and it's at, you know, a quarter of the price and double the quality. Could you just use that? Um, and so there are types of financial services where I think like firms will be able to go direct with technology. And then there's others that you can't, right? Like if you're, If you're brokering a transaction and it takes a human in the loop and you want to shake someone's hand and you want to trust who you're partnering with and negotiating with, like, that's when a banker is going to be invaluable, especially a banker with a Rolodex and, you know, one that, you know, knows your business and your industry and all the counterparties.

Makes total sense. Gabe, this has been fascinating. I really, really appreciate all the time you spent with us. And yeah, wishing you and Rogo all the best of luck. And we should do this again sometime.

Mike, appreciate it. Thanks for having me.

Thank you for listening to Generative Now. If you like this episode, please rate and review the show, and of course subscribe. It really does help. And if you wanna learn more, follow Lightspeed @LightspeedVP on X, YouTube, or LinkedIn. Generative Now is produced by Lightspeed in partnership with Pod People. I am Michael Magnano, and we'll be back next week. See you then.