Latest / Legal Late Night / Practical AI Use Cases, NetDocs' Next Gen Experience, & Ambient AI with Nate Ruiz

Transcript

Practical AI Use Cases, NetDocs' Next Gen Experience, & Ambient AI with Nate Ruiz · Legal Late Night

• Jared Correia (00:00): Hello everybody. We've got to show the promises to be at least mildly interesting for your listening and watching enjoyment. I'm your host, Jared Correia. I'm the CEO of Red Cave Law Firm Consulting. For the monologue, if you're not that deep into AI and are maybe one of those people who keep asking me for use cases, I've got some. In the interview, it's Nate Ruiz, the director of product management for Net Documents. In a counterprogram, we're bringing back holidays on holiday just for Nate. Now, let's talk about the broad-based artificial intelligence use cases you can start with if you need a place to start. (00:46): I recently presented live to about 500 lawyers with respect to the basics of artificial intelligence usage within law practices. You know the usual shit. What is AI? How does it work? When will it go rogue? 500, that's more than I thought would come. Although AI has been part of the popular zeitgeist since early 2023 when ChatGPT broke containment, some folks clearly have not yet been using it aggressively or in some cases at all. The fact is not everybody knows what to even do with AI, which makes the interaction somewhat forced, uneventful, and less than successful. So if you're one of those lawyers who's asking, "What am I supposed to do with this thing exactly?" I have three very broad suggestions for how to proceed. First, collating data. Artificial intelligence is tremendous at aggregating and organizing data. That's important because this is something that humans are quite terrible at. (01:48): If you've ever done research at a physical library, I know, right? You know what it's like when you're looking at reams and reams of information and trying to make sense of it all. Your mind wanders, start looking around, your eyelids begin to droop, get tired, you start to miss things. Artificial intelligence is like a version of you that drinks 65 Red Bulls in one sitting. AI tools can move through tremendous piles of data without needing a break and without losing focus. So if you have a project that involves organization of data into significant chunks, just drop it into an artificial intelligence program, take a nap and come back to get the results. Frontier AI chatbots like Google Gemini, notebook softwares like Microsoft Copilot notebooks, legal research tools, case management softwares, customer relationship management softwares, et al, that include AI features, are all great for this use case. (02:41): Just know that for really, really large swaths of data, you may hit usage limits and will be required to wait out a cool down period or pay for tokens in order to proceed right away. Also, don't add confidential data to any free AI products. In that case, you pay for your usage by allowing the system to train on your data. Second, interactive search. The very worst thing about searching is that if you don't get it right the first time, you've got to start all over again. This is terrible, especially when you're using complex logic patterns like for boolian searches. But even in a natural language searching environment, if you don't get it right the first time, you need to backspace and begin again. With artificial intelligence tools, you can search and if you're not getting what you need, you can instead directly interact with the search platform by asking it further questions or redefining your parameters or narrowing or broadening your focus or even inquiring whether the AI could make suggestions about what you should do next. (03:40): Yes, sometimes that helps. In fact, it helps a lot of the time. Iterating with a search tool is a complete game changer if you've never done it before. Avoiding the productivity trench of having to start, continually restart your searches will save you lots of time in the aggregate. And good news, AI search is just about everywhere. You may be using it already and didn't even know. Creational search engines like Google now offer AI mode search, which delivers an AI overview rather than search engine results pages with which you can iterate by asking additional subqueries in order to revise your search as you go. There are even AI native search tools like Perplexity. Of course, you can also execute traditional web style searches through frontier AI chatbots. Finally, traditional business and legal tech software platforms are starting to add AI based search features, including for product support. (04:34): Third, creative brainstorming. If you have a rough idea that you want to flesh out and your spouse or family members are sick and tired of hearing about your stupid business, artificial intelligence will always be here for you probably until it turns around and tries to kill you. Whether you're creating a new logo for your law firm or trying to develop a substantive or an administrative workflow from scratch or attempting to build long-term revenue goals, artificial intelligence is a great brainstorming partner that won't get annoyed with you, that won't get tired, that won't need to take breaks, though you may get annoyed with it. In the same way that AI can be used to iterate on searches, AI can be used to iterate on your ideas. You can thus utilize a more human version of your initial idea, take it and run with it or refine it further using AI as your revision partner. (05:25): This is a useful strategy even when you're not expecting to get to a final version of whatever you're trying to build inside the AI tool you're starting with. If you launch your logo revision planning in grok, you can still finish it up in Canva or with a graphic designing person, a human. If you work through your process management and Copilot, you could eventually load the final product into Notion. If you're generating revenue projections via Claude, you may ultimately track that via Clio. Artificial intelligence can be described as an affirmation engine in that those tools are trained to be supportive of users, but that same program feature can also help you to bring your ideas from foundation to fruition. Speaking of fruition, we're going to talk to Nate Ruiz of NetDocuments about what the next generation AI experience looks like, when it will be here, and how ambient AI plays into it. (06:17): It sounds fancy, right? Let it all wash over you. That's next. So I've effectively run out of things to say, which is very awkward because this is a podcast. So I'm just going to start the dishwasher. It won't take me long, I promise. Actually, I'll do that later. We've got paper plates. Let's just interview our guests instead. Our guest today is Nate Ruiz, the director of product management at Net Documents. Nate, welcome to the show. Nate Ruiz (06:53): Thanks for having me, Jared. I'm pumped to be here. Jared Correia (06:55): Would it be too aggressive to describe you as a wonderkind? Nate Ruiz (07:02): I would never describe myself, but I'm perfectly open to other people. I'll do it. Sure. I'll take it all day, every day. Jared Correia (07:14): Okay. So we talked to a lot of lawyers here and not coders, but you're a real legitimate software engineer who was coding stuff before there was even AI. So how did you get into that? I talk to all the lawyers and I ask them, "Why'd you become a lawyer? Why did you get into this?" I'm interested to know. Nate Ruiz (07:41): Yeah, so for me, graduated college, not really sure what I was going to do. I actually went to Thailand to teach English for about a year, and while I was over there, obviously kind of an isolating experience since not everyone speaks the language super well. So I was thinking about what I wanted to do in the future and started self-teaching myself, like HTML, CSS, JavaScript. And then when I got back to the States, I went to what they call a bootcamp, a development or an engineering bootcamp. Since then, obviously I worked in engineering for the first five, six years of my career, transferred that into UX design along with the front end engineering, which I worked on a bunch of different products from a bunch of different industries. And then later, when I was in my later career, which I guess I'm in now, I guess, decided to move to product. (08:41): So what do users want? How does that work for the business? And got my first kind of steps in legal tech at a startup called Termscout, which was essentially benchmarking for corporate contracts like M&As, NDAs, things like that. Transferred into a shootoff called Screens, which was AI playbooks for contracting. And we built and sold that in about 18 months. And after getting that integrated - Good for Jared Correia (09:12): You. Nate Ruiz (09:13): Yeah, came to Netdocs where the sky's the limit. Jared Correia (09:17): Okay. You covered a lot there. Let me unpack it a bit. I know. Sorry. Nate, you want to know how old I am? Some of my friends in college, they were programming in Fortran. You remember that shit? Nate Ruiz (09:32): Well, I have no personal memory of it, but I have heard anecdotes. Yeah. Jared Correia (09:38): Thanks. I feel great now. All right. So you guys put together this startup, sold it in 18 months. What was that like? That's a pretty quick turnaround, frankly. Nate Ruiz (09:53): I mean, it was probably one of the highlights of my career. I had been used to startup life, but never at that pace with that kind of outcome. Essentially when ChatGPT came out, it immediately almost rendered a lot of the old machine learning models, like the Roberta models a little bit obsolete. And our CTO at the time saw the writing on the wall and he was like, "We got to run at this in some facet." So myself, the Jared Correia (10:24): CEO, Nate Ruiz (10:25): The CTO kind of made the call to just do this offshoot of the traditional business of Termscout, and then we just ran like crazy. It was probably the most exhausting 18 months of my life, but we got it in customer hands, we got a lot of feedback, we made some sales, and then it turned out well. We got it acquired at the end of that 18 months, which was awesome. Jared Correia (10:48): I like how you guys were like, "This AI think might have some legs, we should start doing that." Nate Ruiz (10:54): Get on the train, right? Jared Correia (10:57): So the legal part of it, did you have any interest in the legal space or were you just like, "I kind of fell into this and I'm going to see where it goes"? Nate Ruiz (11:10): So I fell into that opportunity, that first term scout opportunity, and vaguely interested in legal, but I didn't have any touch points or personal experience. But I learned so much at the year and a half leading up to the offshoot and then the 18 months after, and then all my time at NetDocuments. I learned new stuff every day about the legal tech industry and the legal industry holistically, and it's fascinating. Jared Correia (11:39): So now you're at NetDocuments. Well established legal tech company, what was it like going from a startup to a more established company? The use of the corporate stuff? How's that going? Nate Ruiz (11:55): It's an adjustment. Jared Correia (11:56): A lot of synergy happening, Nate? Nate Ruiz (11:59): Yeah, tons of synergy, a lot of low hanging fruit. Yeah, it was an adjustment. Excellent. (12:08): It was an adjustment for sure. But what was really cool was our chief product officer here, Dan Houck, he brought me on board with the assumption that I take the team that had historically been working on some of the things I'm working on today and make it almost in the image of a startup and he would work to unblock me however he could or isolate me from a lot of that. So it's still been an adjustment. There's just a lot of nuance to doing things at the scale of net documents that just didn't exist in startup land. But from a pace perspective, we've been able to move incredibly fast, which that's what I want to do at end of the day, build cool stuff. Yeah. Jared Correia (12:55): So now that you're in the legal space and you're kind of ensconced here, what do you like about working with lawyers and when are they pains in the ass? Because they are sometimes, let's be real. Nate Ruiz (13:07): Yeah. You said it. I didn't say it. I love - Jared Correia (13:12): Don't say the shit, Nate. Don't worry about it. You can build corporate value. I'll tell you Nate Ruiz (13:18): This. That's what I'm doing. That's what I'm doing. I mean, the amazing thing about it is I have never had access to customers at the scale as I've had them at Netdocs, and they're all so vocal about their wants, needs, opinions, which is a two-sided coin. That's amazing feedback to go build some incredible products and it's hard to get in the startup space. But the flip side of that is you have this deluge of feedback and what I've come to the conclusion, every kind of lawyer and law firm thinks they're a little bit of a special snowflake kind of use case. They need special things, and that's really hard to build SaaS around. With SaaS, you're trying to find these repeatable processes. So that's been a blessing and a curse, but having that access and having people speak their minds eloquently in a very detailed way, that helps me build products that meet their needs. Jared Correia (14:29): It's interesting that you bring that up. I find that some lawyers that I talk to are like, they don't want to give feedback to the software companies they work with. I'm like, "Man, you got to tell these guys they love it. It's right up their alley." Nate Ruiz (14:41): Yeah, bad stuff. Yeah, just send it to me. Jared Correia (14:45): Oh man, you're going to wish you didn't say that. Okay. Let's talk about, you're kind of running what is called the next gen experience or next gen experience over at NetDocuments. Nate Ruiz (15:01): That's right. Jared Correia (15:01): As someone whose friends coded in Fortran, what the fuck does that even mean, Nate? Nate Ruiz (15:07): Yeah, that's a good question. It's a lot of things, so I'll try and break it down as succinctly as I can. So NetDocs, old company, old company. It's been built up over the last 20 plus years. The building over time like that and the meeting of different nuanced customer needs just kind of builds you into a corner of this big monolithic thing that's kind of slow to iterate on, slow to innovate on. So a big goal of ours is, "Hey, can we start fresh, jettison a lot of that nuance and really start running at pace?" So that is the next gen experience. Obviously, we need to be doing the same things NetDocuments has done for 20 years, but there's often ways we can rethink how we do them. So at its core, for me, it's two big pieces, a UX, UI, total re-architecture. So it's a totally new look and feel. (16:05): Streamlined workflows should help with adoption of end users using the platform. And then the second piece, and what we're really, really pumped about, the more forward-looking piece is this legal context graph. And that legal context graph, at time, documents are uploaded into net documents. We'll classify them, we'll extract key data, we'll summarize them, and then we'll start connecting the pieces across an entire repository, and that becomes amazing context for legal professionals, and then also increasingly for AI agents like a Claude or a ChatGPT to go find the things they're looking for and get you a higher quality output. Jared Correia (16:47): Right. I mean, so on the first piece of this, a lot of legal software, I feel like they strive to be like, "Let's just copy Microsoft because lawyers are used to that." But it sounds like you're going beyond that and thinking about what does the user experience look like for an attorney specifically without necessarily regard to these traditional software tropes that you see. Is that fair to say? Nate Ruiz (17:14): It's fair. It's a fine line to walk, right? We could spend a lot of time reinventing the wheel and folks could say, "Hey, we're not going to use it because the change management is too high." So we're trying to ride that fine line that we want our software to feel familiar to these legal professionals, often working most of their life in Microsoft products, but we want to have Jared Correia (17:39): Enough Nate Ruiz (17:39): Of the own NetDocuments flavor and sometimes ideally do better than Microsoft as far as workflows, user flows, et cetera. Jared Correia (17:49): Shouldn't be too hard. Have you seen Teams? Nate, in terms of the legal context graph, I think some people might hear that and they might be like, "Wow, immediately over my head." Do you have an example or two about how that could work in context and why it's valuable? Nate Ruiz (18:08): Sure. So at NetDocuments specifically, people organize their documents into what they call matters. They can call them workspaces. The name of it doesn't really matter, but it's a collection of files, folders, et cetera, of documents. All of those documents, when they come into NetDocuments are unstructured data. It's just Jared Correia (18:31): Legal Nate Ruiz (18:31): Pros. It'd be like if you set an agent on it, they'd have to read the entire document to get the gist of it. Jared Correia (18:39): Well, I feel like people are starting to talk about this issue of structured data sets for AI. So keep going. This is great. Nate Ruiz (18:47): Yeah. So essentially, when you upload a document to NetDocuments now, we try and structure that data automatically and in the background without you having to do anything. So out of the box, it would just work. You upload, say, an NDA to NetDocuments. We would send it through what we're calling our doc intelligence pipeline. It would say, "Okay, I understand this is an NDA in this industry." Based on that understanding of what type of document it is, it'll then go say, "Okay, this is an NDA. This is the key information I need to pull for an NDA." And it would do the same for. (19:29): It could be the same for a claim construction debrief, or it could be whatever area of law, it'll say, "Okay, what type of document is this?" And then based on that, we'll extract relevant data for that specific type of document. So we're talking things like governing law, entities like people and parties, important dates, you could have dollar amounts, things like that it'll pull automatically, and that all becomes context that can be served up to the AI agent instead of sending them the entire document. So what we've found is it's just the same as working with a human. If you put a paralegal on a new case and you're like, "Tell me what this case is about," they would have to go read through tons Jared Correia (20:14): Of Nate Ruiz (20:14): Documents to assemble that knowledge. Same thing with agents. But if you gave this associate a summary of the entire matter and the key people and key dates of this matter and key fact data points and fact patterns, then it can get right to work with a deliverable potentially. So that's kind of the goal and the value behind this legal context group. Jared Correia (20:40): So it's the same as working with a human only better and faster. Gotcha. Ideally, Nate Ruiz (20:44): Yeah. Jared Correia (20:44): I like that's cool that it's done automatically. I would love that as an attorney. Do you find people being like, "I'm worried about that. Can I change the settings? Can I screw around with this?" And do you have the ability to do that? I'm assuming the answer's yes, but. Nate Ruiz (21:01): So yeah, this thing that we're calling doc intelligence, we really want it to be this out of the box. It just works. You don't tweak any of the knobs and dials on it itself, and it gives us this consistent layer that we can build experiences off of. So traditionally, every law firm would have different kind of profile attributes or metadata that they would want, but it's hard for us to build SaaS experiences off of that craziness between firm to firm. Jared Correia (21:32): So this gives us - At least V1 probably has to be an automated version. Nate Ruiz (21:35): That's right. That's right. Yeah. And so we still have the ability for people to add custom attributes and do the same automatically into their custom kind of taxonomy, but we always want to have our out of the box baseline to start with. And then if you choose that you need something else, you can go build it. Jared Correia (21:53): I can tell that you've started to work with lawyers because you're like, "Yeah, I get a lot of requests like this." Nate Ruiz (21:59): Like I said, especially Snowflakes, we have this one thing that we like to pull out. You can do that, but we're going to pull out all this other stuff too. Jared Correia (22:10): Right. 97% of it's going to be the same that every other law firm wants to do. Okay. I know the legal context graph stuff is new. People should check that out. For sure. If they're NetDocuments users currently or if they are not particularly aware of what's happening there, or even if you've never seen the product before. I always tell people to grab a demo. Obviously you've been around the space for a while. I like the product. What about other features in the system that you think are underutilized? Do you have any that you want to highlight? Nate Ruiz (22:46): Features, features. I mean, at the end of the day, we're a DMS, right? We're kind of where documents go to be stored, but we're increasingly becoming where work gets done. So our recent integration and partnership with Microsoft allows you to collaborate real time like you would in a OneDrive, but with those files being stored in NetDocuments. We also just released our tabular review, which is you could drag thousands of files, say, "Hey, I want this data out of all the files," and it'll go make a table of all that data. That kind of brings us to parity with some of the big AI workflow tools out there, or it gets us near parody. But what I am most excited about is what I like to call ambient AI that we're just kind of baking into the experience, and that's AI that works Jared Correia (23:42): Without - Oh, right. We talked about this a little bit before. Yeah. Nate Ruiz (23:46): Yeah. Jared Correia (23:46): Keep Nate Ruiz (23:47): Going. It's AI that works without you having to ask it. So the legal context graph is a piece of that. Another one that we're kind of doing as an extension to that legal context graph, versioning of documents. So when you're working on a document, you decide to version it up, make some edits in this other version, just really at that time that you save that new version, understanding the differences between the past version and the current version, and then putting a really succinct summary right there in the user's face so that they don't have to go open both versions, run a complex compare tool on them. "Hey, at high level, was the changes cosmetic? Were the changes structural? "And that just happening without you having to say," I would like to compare these two versions. "That's kind of what I'm most pumped about because what we find in AI adoption is half the time that blank page, what do I do with it, is the hardest question to answer for these legal professionals. Jared Correia (24:49): I don't know, man. Ambient AI is a term I really like. I don't feel like ambient AI would be running the singularity on me. I feel like I'm just chilling. I don't know. Do you feel like there's a couple different camps here? Because when you talk to people about self-driving cars, I think some people are like, "Fuck, it's the end of the world." But I'm like, "Bro, if I can take a nap in a car that is driving me places, sign me up yesterday." I mean, yeah, that's what I want. I drive my kids all over Creation for sporting events. Could I put in a location and then literally just sleep for three hours? That would be life-changing to me. Ambient AI, let's make this happen. Nate Ruiz (25:38): I mean, I've found that it's different strokes for different folks. Some people really want that level of control and we want to make sure we keep that level of control at their fingertips. But then you have a whole other camp like you and I who would love to take a nap and they're going to be really - I don't care Jared Correia (25:55): About control. I care about naps, so I want. Me Nate Ruiz (25:59): Too. I'm right there with you. So it really varies in the customers I've been talking to, and we're just trying to not be all things for all people, but give a little flavor of everything, give you an out of the box experience, let you tweak it. Jared Correia (26:14): And if you want to be a setting sound, you can still be a setting sound and for all your shit. 100%. Anything else you can tell me about that you're working on that's cool that's not going to get you fired or should we go down a different road? Nate Ruiz (26:28): So I'm hitting the Iltacon greatest hits right now because we were just at that conference Jared Correia (26:34): Last nig. Give me the IltaCon greatest hits. Yeah. So that would be about probably three weeks ago by the time this show comes out, but yeah, go Nate Ruiz (26:42): Ahead. So another way we're extending that legal context graph that we kind of announced at Ilticon was what we're calling legal authorities, right? So again, as soon as the document is kind of uploaded, we'll understand (27:02): The citations in those documents to legal authorities cited elsewhere. And then we'll highlight those in our previewer and allow users to click through to go see that case that was cited or that precedent that was cited right now on Court Listener, which is kind of an open source kind of database of that. We're going to be looking to extend that in the future, so folks could potentially hook up, we'll see, but potentially hook up their Lexus or their Lexus to get those citations coming from Lexus instead of CourtListener. But yeah, we're just kind of marching this legal context graph down the road and seeing where those extensions make sense. And then we're also just being really cognizant on proving the value of it. So we released a couple weeks ago, we released some evals on how the legal context graph not only increases accuracy, but reduces token cost for AI agents. (28:07): So it doesn't matter what AI agent you use, could be Claude, could be ChatGPT, could be Harvey, because the AI doesn't have to go regenerate that context every time you ask it a question. Jared Correia (28:19): Totally makes sense. Nate Ruiz (28:20): Yeah. It'll just drive down costs pretty significantly. Jared Correia (28:24): That's cool. Okay. How was Iltacon this year? Was this your first one or one of your first ones? Nate Ruiz (28:31): This is my second Iltacon. We used to go to Klock before, but Jared Correia (28:36): IltaCon I Nate Ruiz (28:37): Guess is where NetDocuments shows up. It was great. Jared Correia (28:40): ILTA is like the, what is it? International Legal Technology Association for those unaware, they have a conference every year. That's right. And then KLC is for in-house counsel. Nate Ruiz (28:51): That's right. Jared Correia (28:53): Iltacon this year, Vegas, Disney? Nate Ruiz (28:56): It was in Nashville. Jared Correia (28:59): Oh, shit. Okay. Yeah. Nate Ruiz (29:01): So I was actually there Jared Correia (29:02): For - At the, what is it, the Opryland thing? Yeah. Oh my God. That is just like a sprawling metroplex. Nate Ruiz (29:13): Yeah. It takes 20 minutes to get from one meeting room to the next. It was wild. Jared Correia (29:19): I was lost there for 45 minutes a couple years ago all inside the building. You guys have your own conference as well, right? Nate Ruiz (29:27): Yeah. Yeah. So we have Jared Correia (29:28): Our - When's that one happening? That's happening soon, isn't it? Nate Ruiz (29:32): Yeah, it's later October. It's called the Inspire Conference. I should know this off the top of my head, but I've been running. I think it's like the - That's Jared Correia (29:40): Why you have a fucking website. Nate Ruiz (29:42): Yeah.That's Jared Correia (29:43): All we need Nate Ruiz (29:43): To know. Jared Correia (29:44): Inspire Conference. I'm already inspired. Check it out. Okay. How do you and your team use artificial intelligence in your work? Nate Ruiz (29:55): Oh, that's such a great question. So part of the reason why we've been able to run so fast since I came on is that we've started building software with AI coding agents. So especially at the turn of this last year, so last December into January, Claudecode just started becoming a lights out tool for product spec development all the way to engineering the thing and QAing it. So we've been able to really leverage Claudecode here at NetDocuments to run much faster than we ever have before. So we can say on a given Monday, we could say, "Oh, I have this idea of this cool ambient AI feature." A couple days later, we could see that in our own product. So it's incredible to be able to run at that pace and more importantly, get feedback from customers at that pace. And that's really how we're going to be successful in this whole next gen push is that speed of shipping things and speed at iterating with feedback. Jared Correia (31:05): So I think this is kind of like, if you're a developer already and you know your shit, you're not a danger when you're vibe coding. Should lawyers be vibecoding? Nate Ruiz (31:17): I've seen them do it. I think so. I think it's always good to dip your toes to learn new technologies. I just gave a panel about MCP. So if you don't know about MCP, for example, best thing to do, find a low stakes MCP connection Jared Correia (31:39): And start - Can you tell people what that is if they're new to. Yeah. Nate Ruiz (31:45): MCP is model context protocol. The simplest way I can think of describing it as the USBC for AI agents. It's just like a connector. It's infrastructure that is widely adopted by the tech industry. And basically what it allows you to do is have a central kind of front door in your AI agent of choice, whether that's a Claude, Harvey, whatever, and then have connectors, MCP connections from your billing software, potentially Netdocs as your DMS, however you want. Outlook has its own MCP connector, Microsoft. It's just this new way of integrating with third party services that isn't point to point like you used to have to do with APIs. Jared Correia (32:32): So I guess USBC for everybody except Apple. Nate Ruiz (32:36): They're just on it now. Yeah, they're new. Did Jared Correia (32:39): They move? Nate Ruiz (32:40): Yeah, they moved. Jared Correia (32:41): My hatred of Apple is long and deep, but I'm glad they're on the USBC now. Good for them. They Nate Ruiz (32:47): Had to get dragged along for sure. Jared Correia (32:53): Lawyers should be using Claudecode Cowork. Just get into the chatbot to start with. Let's say somebody's out there and like, "This sounds cool. I don't want to do the thing where I give Claude access to my email and then deletes all my email." What's a safe way to start if somebody's a relative noob here? Nate Ruiz (33:15): Yeah, it's interesting because I think there's a lot of this anxiety just around that. Will I spin something up? Yeah. And starting locally on your own computer is the best way to start. Being really cognizant of what you hook up to is the second thing. So Claudecode, I could spin up a new website in 10 minutes, but if that website isn't hosted anywhere, it all stays local on my machine, it's not touching anything. The risk is super duper Jared Correia (33:46): Low. Nate Ruiz (33:47): So to that end, I'd say go for it. Just be really cognizant of what you're connecting it to. As soon as you start giving it connection to data sources like your email data sources like your CRM, that's when it starts to get a little bit fuzzy and you should probably start to talk to some engineers about how to move forward. But there's a lot you can do locally on your machine. All Jared Correia (34:14): Right, so let's talk beyond that because it's one thing to be like. I mean, I think if you're listening to this and you're a lawyer and you've used AI to be like, "Hey, Google Gemini, I've got some salmon in my refrigerator. What should I make for dinner?" That's a whole lot different than just being like, "I'm going to connect this shit everywhere." But if I'm an attorney, I'm probably thinking about I can connect my net documents to Microsoft 365. Should I also connect my claw to Microsoft 365? That's potentially appealing. How should people go through answering those questions with relation to governance, like data confidentiality, that kind of thing? I'm not expecting you to have an answer, but what kind of questions should people be asking? You kind of jumped into that a little bit just now. Nate Ruiz (35:02): Sure. Yeah. So I think it's an exercise and it's a continuum. First thing I always like to say is how clean is your data? Because if you hook up these AI agents to bad data, you're probably going to get bad outputs. So first making sure you've got some pretty clean data wherever you're trying to hook it up. Second thing, how confidential is that data? Is it got PII in it or is it something like a Slack channel? There are different gradients for level of risk. Jared Correia (35:34): Not everything's got personal identifying information in it. Not everything's got HIPAA data inside of it. Nate Ruiz (35:39): That's right. Yeah. And so you can choose something that's pretty low risk to hook up to start. And then also just understand what actions you are enabling. So we just talked about should I hook up Gemini to my email or is it going to go in and delete a bunch of stuff? Jared Correia (35:57): Those Nate Ruiz (35:58): Are actually skills that are surfaced to that MCP via Outlook. So you could say, I want to turn these skills off for myself or for my entire organization. I don't want to do anything destructive. And it's up to vendors like Net Documents to make sure you have all those knobs and levers to pull so that your governance stays in place. Alongside all the stuff that Net Documents is famous for, the government piece being specifically important, how do we make sure agents don't have access to anything you as a user wouldn't have access to? Jared Correia (36:34): Nate, good stuff, man. That was fun. It's Nate Ruiz (36:37): Exciting. Jared Correia (36:38): I'm going to have to have you back on sometime so I can grow you some more about AI. I love Nate Ruiz (36:43): It. Jared Correia (36:43): Will you stick around for one more segment? Nate Ruiz (36:46): Sure. Yeah. Be happy to. Jared Correia (36:48): All right everybody. We'll be right back with Nate. Welcome back everybody. It's the Counter Program. We're here again at the end of the show. It's a podcast within a podcast. This is a conversational space where we can address usually unrelated topics that I want to explore at a greater depth with my guests. Expect no rhyme and very little reason. We're back here with Nate Ruiz from Net Documents. He's building corporate value like a motherfucker. Nate, how are you? Nate Ruiz (37:20): I'm doing great. I am building that value. Jared Correia (37:22): Thanks for coming back. (37:25): Nate, with Labor Day and the end of summer approaching, sadly, actually as of the recording and release of this podcast, summer will pretty much be over. I'm going to bring back, I know, one of my favorite segments. It's called Holidays on Holiday. Here's how it works. I will read off the name of a defunct holiday and a description for same. You just have to tell me whether it's real or fake. Easy, right? Yeah. We're going to get deep here. We're going to go back to the middle ages, going to be all over the place. We're not using any skills or any AI connectors. So this is all you. Okay. Number one, Robagalia. I'm sure you celebrate this with your family every year. Nate Ruiz (38:18): Yeah. Classic. Jared Correia (38:21): This was an ancient Roman agricultural festival held every April 25th where the Flaming Cornelius, that's what people used to call me, ritually sacrificed a rust colored puppy and a sheep at the sacred grove of the god robigis to protect ripening wheat crops from stem rust and devastating agricultural mildew. Did the Romans have a fucked up celebration where they had to find a rust-colored puppy and then sacrifice it for the sake of their crops? Nate Ruiz (39:02): It's so wild. I'm going to go with yes. Those Romans were nuts, so it's oddly specific too. Jared Correia (39:12): First, where do you even find rust-colored puppies? I didn't know that was a thing. And you have to sacrifice the sheep too? Nate Ruiz (39:18): A vicela? Is that just viceles? Jared Correia (39:23): Maybe. Okay. You were correct. This is true. This was a real feast day. Unsurprisingly, it was suppressed during the Roman Empire. I don't know. I got to sign with the Romans on this one. Nate Ruiz (39:36): Yeah. Good riddance. I'm glad we don't celebrate that anymore. Jared Correia (39:41): Let the puppies live, even the rustcard ones. Nate Ruiz (39:44): Exactly. Jared Correia (39:46): Are you ready for number two? Nate Ruiz (39:48): Yeah. Jared Correia (39:48): Hit me. The second celebration is called Noctus Lucerne. It's a Byzantine, not Byzantine as lowercase B, capital B from the Byzantine Empire. A Byzantine maritime observance where merchant gills gathered in dockside warehouses at midnight precisely to conduct blind barters in absolute darkness, believing that removing site eliminated visual prejudices, prevented haggling disputes, and guaranteed strictly honest valuations for valuable foreign trade goods. Did merchants during the Byzantine Empire trade in absolute darkness at midnight for the sake of this festival? Is this real? Is this fake? Nate Ruiz (40:43): I'm going to go fake for that one. And I don't Jared Correia (40:46): Really have a reason. You Nate Ruiz (40:47): Give me a reason. I mean, it sounds like Jared Correia (40:49): Something that - Nevermind. Nate Ruiz (40:50): Yeah. I just don't like the name. Jared Correia (40:55): You're like, "I want to barter with my eyes open." You are correct, sir. Amazing. Two for two. Now I have to tell you, Nate, I don't want to put additional pressure on you, but we had our first ever perfect score a few weeks back in 20 years of doing this podcast. So if you can pull that shit off, that'll be a next gen user experience, let me tell you. Okay. Nate Ruiz (41:20): Amazing. Jared Correia (41:21): Number three, Festim Asenorum, a riotous medieval European festival celebrating the biblical ass associated with the holy family's flight into Egypt after Jesus was born, during which townfolk led a lavishly adorned real donkey straight into a church where priests and congregants enthusiastically braid responses instead of singing traditional Latin chants. Is the feast of the ass a real thing? Nate Ruiz (41:55): Oh, I'm so Jared Correia (41:56): Nervous. This is just like everyday shit during the Middle Ages. Nate Ruiz (42:00): They Jared Correia (42:00): Just don't wild stuff back then. Nate Ruiz (42:02): Yeah. I mean, they were killing puppies, so why not talk like a donkey? I'm going to go with false and I'm only fifty fifty on that. Jared Correia (42:14): This was a real festival. Nate Ruiz (42:15): Oh my God. Really? Jared Correia (42:18): It was originally celebrated in France as a byproduct of other feast days celebrating the donkey related stories in the Bible. Nate, there were a lot of donkey related stories in the Bible. There were. Nate Ruiz (42:29): Lots of donkeys Jared Correia (42:30): In the Bible. They love donkeys. Nate Ruiz (42:31): Yeah. Jared Correia (42:33): Unfortunately, this was banned by church authorities in the 15th century, so you need to keep your brain to yourself from here on out. Nate Ruiz (42:41): Where's the harm? I mean, who doesn't want to bray like a donkey every now and then? Jared Correia (42:47): I would if I could. Two for three. Yeah. Very good score. I got three more for you. Nate Ruiz (42:55): Oh, amazing. Jared Correia (42:55): The next holiday is vindfilated. I think I'm pronouncing that correctly. An archaic Scandinavian winter tradition where coastal villagers curled salted parsnips, dried cocktails and carved driftwood effigies into frigid fjords during the year's final gale to appease northern wind spirits and prevent spring ice dams from crushing their timber ships as one would. Did the Scandinavian celebrate something called vindfallated where they threw salted parsnips and dried cocktails, among other things, into the sea? No animals or humans were sacrificed, which I think is a win. Nate Ruiz (43:49): Yeah, that's a win. I'm going to go, yes, it happened. Why not? Jared Correia (43:55): Also fake. All right, we're on a bad run here. Nate Ruiz (43:58): I started so well. Jared Correia (44:00): You started strong, but you can bring it back. We got two left. You'll have a chance to finish over 50%. Yeah. Oak Apple Day. Oak Apple Day. A former British public holiday established to commemorate the restoration of King Charles II to the throne in 1660. I believe that's the glorious revolution, during which citizens wore oak leaves to honor his famous tree hideout while violently thrashing any unpatriotic passerby caught without one of the oak leaves with stinging nettles. You got to love a holiday where you can beat people with stinging nettles. Nate Ruiz (44:44): Just Jared Correia (44:45): Violent holidays. The British did some fucked up shit. This is right in line with that. Okay. Was Oak Tree. What is it fucking called? Oak Apple Day. Was Oak Apple Day a real British holiday? Nate Ruiz (45:05): Now I'm shook. Yes. Jared Correia (45:07): Now you're like, "Why did I agree to do this shit?" Yes, mate, it is a real holiday. Well done, sir. Let me give you a little background. After losing the Battle of Worcester in 1651, tough B, King Charles II hid from enemy soldiers in an oak tree and people wore those oak leaves on May 29th to show loyalty to the king. It was officially canceled in 1859, but some people still observe this tradition locally. Nate Ruiz (45:41): Wow. Jared Correia (45:41): Crazy shit from the British. Nate Ruiz (45:43): Nettles and all? Jared Correia (45:45): You got to have your nettles ready in case somebody's not wearing their oak leaves. You got to do your civic duty. Nate Ruiz (45:51): That's right. Jared Correia (45:52): Wow. I got to say I prefer our normal holidays like Labor Day. Okay, I got one more for you. Feast of Fools. This is a raucous medieval church holiday celebrated around the new year by minor clergy who elected a mock Lord of Misrule or Archbishop of Fools. Dawned bazar costumes, burned old shoe leather instead of incense and played dice directly upon sacred cathedral altars. This sounds like the antithesis of anything anyone would do in church. But this was the Feast of Fools. Throwing dice in church, I believe frowned upon most of the year, but is it okay to do it during the Feast of Fools? Was the Feast of Fools a real holiday? I'm going to Nate Ruiz (46:49): Say no. I Jared Correia (46:50): Like how you're working through this. No. How come? Nate Ruiz (46:55): I mean, the shoe leather, that seemed like an oddly specific detail, and these were lesser clergymen throwing dice on the altar. I mean, if it's Jared Correia (47:05): True - Hey, you're not. Are you not burning shoe leather in the room you're in right now? I think that's fucked Nate Ruiz (47:12): Up. I always have a pile going, but this is the clergy. Jared Correia (47:17): Real holiday. Another real holiday. Officially suppressed throughout Catholic Europe between the 15th and 16th centuries. Due to growing protests, what a surprise. Nate Ruiz (47:29): Yeah, weird. Jared Correia (47:31): Nate, you can come to my house for the next Feast of Fools celebration. Amazing. Nate Ruiz (47:36): Amazing. We'll burn Jared Correia (47:36): Leather Nate Ruiz (47:37): Together. Jared Correia (47:39): Pretty solid effort though. That was a difficult quiz and you hit the 500 mark, three out of six. Thank you, sir. Nate Ruiz (47:46): Yeah, it was fun to play. Jared Correia (47:49): I appreciate you coming on. It was a real treat today. We'll have to have you back sometime. Nate Ruiz (47:52): Likewise. Really fun. All Jared Correia (47:55): Right. Take it easy. Nate Ruiz (47:56): All right. Bye. Jared Correia (47:58): Thanks for our guest today, Nate Ruiz, who is the director of product management for Net Documents. To learn more about Nate and Net Documents, visit netdocuments.com. That's netdocuments, all one word, dot com, netdocuments.com. Now, because I'll always be a 90s kid whose pants were a little puffier than Dan Marino's, but whose passion is burning disease for anyone who would listen, I'm now just doing the modern version of that, which is creating Spotify playlist for every podcast episode that I record where the signs are tangentially related to an episode topic. For this week's playlist, we've got signs about going on trips. That's right. What do you do during the holidays? You go on vacation. This vacation songs. At least I hope some people are still getting their vacation on during the fall, post the summer. And it's all sponsored by Royal Caribbean Cruise Lines. Not so, but I wish it was. (48:56): Some of those cruise ships are pretty dope, but Frank is scheduling regular excursions. You got this Royal Caribbean Cruise Lines. Join us next time when I learn how to drive stick.