How AI Is Changing Contract Management with Nate Kostelnik

Made with speech-to-text software, so a few words may be off.

0:00You're listening to the Contract Heroes Podcast, your one-stop shop for all things contract management. And now here are your hosts, Marc and Pepe. All right, Nate, well, thank you so much for joining us today. I'm excited to be having this conversation, able we were finally to s- get something on the books. Before we flip things over to Pepe to, to give us a little bit more information on what we're gonna be chatting about today, I was hoping you could give the listeners a bit more information on your background and how you got into, you know, the contract management space in, in general. Yeah, sure. So I'm a lawyer. I started in private practice and was there for many years. Eventually, I moved, uh, in-house. Right now I'm with Oncology Nursing Society, or ONS. We're an international non-profit trade group for oncology nurses. Uh, I handle all the contracts and IP and, and everything else with ONS, and so that keeps me busy.

0:59And I tend to write a little bit now on contracting and how to make the process better and smoother, and a little bit more on AI as well, since, uh, everybody's talking about it too, so. And our main topic right now, right? So we would like to go actually back to basics. You know, there's been a lot of stuff going on with AI, and especially how it's been implemented in CLM system. And I think one of the main focus is redlining, because what we always say is like, "You can have the best technology," and CLMs can do, like, really well automating the whole processes. But negotiation, it's always, like, something that it's more into the personal skills of the lawyers or contract managers inside any corporation or organization. So Nate, let's kick things with a big one. Is CLM dead? And what led you float that idea, and how you really see its role evolving in today's AI-driven environment? Yeah, hard-hitting question for a Friday, right? So I think it's a little bit of both.

2:04I don't think it's dead. There are still a lot of core features of CLMs that I need every day, I use every day, that a lot of teams use every day. So intake, approvals, renewals. We're still gonna have our SAS vendors that, that wanna auto-renew you. So tracking those milestones and processes, it's not going away. So we still need those core features. And beyond that, even just the, the repository function, knowing what's in your contracts, being able to find them, having them at your, your fingertips, not just for me, who tends to work with them every day, but for the business team who wants to be able to self-serve a little bit and find their data. So those aspects, I think, of CLM are staying, uh, in full force. But I think we're seeing the new features with AI redlining and kind of evaluating whether CLM, as it w- was developed and kind of pitched to the customers as kind of this be-all, end-all solution.

3:05Some vendors do that, some don't. But that approach, I think we're seeing some refinement with AI to carve out some certain processes, some chunks to say, "Yeah, here's what we still really need. Uh, maybe we don't need this, this bigger solution. We can focus in on some smaller, more discrete tasks or whatnot." Sounds good. And Nate, before we jumping into the AI functionalities and new ones, what would you say that has traditional CLM gotten right, and where has it fallen short in actually improving the contract process? I think back, and I know I've talked to a lot of people who say, "I can't imagine doing it without my CLM. It was a mess before we got one." And I know, know some people out there still have, you know, a mess of just SharePoint folders, if they're lucky. Maybe actual s- physical file cabinets. So that's a mess, and it's tough to imagine how you can really manage your contractual risks and processes productively going from that kind of just messy state into something else.

4:05And so without that, like I said, I don't know what I'd do without that CLM functionality, just to handle intake, processing approvals, and just getting a handle on, on what is there. I think those core features are useful for what vendors... I mean, if you find the, the right vendor for the tech that meets you on those aspects, I think you're doing well. If you're trying to go for, you don't have anything, you're starting from zero, that could be a challenge. Uh, you may not have the processes, and now you're trying to just create something on the fly and, and that's always a recipe for disaster. It's important, right? Because we look at these organizations where they don't have anything in place, like r- right, like you said, and, and maybe if they're lucky, they do have SharePoint, and there are some type of folder structures in there. And, and there's organizations that don't even have that still, which is, you know... It, it amazes me that there's still organizations out there, and, and big ones at that, that still don't have a CLM in place. And it's a huge leap forward when they do decide to obviously purchase a, a software solution, and there's a, there's a lot of work there that needs to be done.

5:06But I think when we look at some of the, the functionality that AI is allowing us to do less of, right? I think, you know, Pepe, as you mentioned, right, redlining. And you said that this is kind of a key focus area, right, when it comes to maybe the evolution of AI and, and allowing, taking things off your plate, right? This is kind of a, a focus area. So how do you see this evolving, and, and kind of where is it at right now? It's a great question. I use AI redlining for, like, a first pass review for a lot of contracts, both that you might consider high volume, but also more bespoke or, uh, less frequent. Not necessarily because I'm always looking for the efficiency and the time saving, but I'm really just trying to get used to building it in as part of our process and seeing where AI works really well for first pass review and where it may fail, where it may not catch something, or it may flag something and, and I'm not quite sure why it did that. So I use it for kind of that, uh, iterative- Uh, refinement to always, uh, find ways for it to improve as part of our process.

6:09But even beyond that, I mean, you hear what everybody's talking about these days is eventually we're kind of working towards AI negotiating with AI, and two AIs are just gonna negotiate an NDA or a SaaS agreement on their own. I don't know when we'll get there, if we will, but if that's kind of what we're working towards, what steps do we have to put in place now that get us closer to that? And so part of it is developing that playbook, identifying the issues you care about, maybe stuff you don't. Do you have your thresholds? What types of activities would you be okay with eventually delegating to AI, even if you don't do it yet, and what ones would you still wanna hold onto for now? And so I think kind of saying, "Okay, if we're here now, and that's where we might go, what are the gaps and what are the tasks and steps that we have to fill in to see if we can even get to that point?" And let me throw you one real case, right? Well, very, very common case, and I'm thinking about our friend, Nada AlnaYafe, uh, from Contract Nerds- Yeah... and, uh, Redlining Etiquette.

7:16She's been on our podcast a couple of times. And, uh, one thing with contracts and red lining, right? First is, uh, maybe a drafting problem, right? When organization has, uh, contract templates with terms and conditions that nobody will accept, like unlimited liability. That's, like, the most common one. Like, nobody is going to accept that- Yeah... and they're going to get it back with a red line. And one of the things that maybe that you need to have something in between, right? Not start a n- a negotiation with, with something that extreme. Maybe try to look back into your previous negotiations with other constructs and try to put something in, in the middle so you don't get that much pushbacks. And if you have more than one of those type of clauses, right, or unilateral confidentiality, like, why, right? And on the other side, and I will say this is more like, like the cultural one, right? Es- especially in law firms and sometimes also in-house teams, when you receive a third-party paper and you know it's okay.

8:17Okay, this is like, like standard terms, but sometimes, like, trainees or more junior lawyers, they don't want to just send a document without any review and tell the, their boss this is okay, because they, they're afraid, like, okay, why you didn't put any comment or, or any red line. Like, are you actually doing your work? So what would you say there? Because even if you have the best AI and it's trained with hundred of thousand of the same type of document, and it's already, like, talking to a human being with 40 years of experience, like, where do you think there's, like, a balance between how AI can handle this, these drafting problems, and also the cultural problems? Yeah. Great question. So I think about this a lot, and I think one of the benefits of, you know, AI playbooking or giving AI instructions for how you want it to review your contracts is you have to put a little bit of work in upfront to do that high-level thinking, to do the hard work, to say, "Here are the clauses I care about. Here are the ones I don't. I'm not gonna include them in my playbook." The ones that I do care about, uh, as you said, do I wanna ask for unlimited liability? If I do, I'm gonna have my AI flag that in every contract.

9:32I'm gonna get pushback on the red lines, and now I'm gonna have to do more work and more negotiations. So it's not the AI's fault, 'cause it's doing what I told it to do. That's my fault, 'cause I didn't put in the time and the thought to go through and kind of validate that request as something I wanna ask for. So unlimited liability is a great example. Uh, when you're developing your playbook, look around and see if you can benchmark your playbook against some market standards. There are plenty of them out there, and that's a great place to start so that you can say, "Hey, what we're asking for is reasonable, and we're starting from a more reasonable position." Maybe we can have fallbacks from there and escalate if needed, but that's a great place to start instead of asking for stuff that you're probably not going to get accepted anyway. And then I think the other example you mentioned for unilateral confidentiality is another great example.

10:23When you're developing this, it's useful to think about these are red lines you're gonna send to the other side, and you kind of want part of that collaborative process. And so if you're developing instructions for your AI, can you reframe some of your asks as mutual positions instead of taking unilateral? So you're gonna s- fly through things much faster if you can insert a mutual provision rather than something that's unilateral. So even just building that concept into your playbook can save you some time in red lining negotiations as you go along. And, y- you know, let me ask you this, uh, Nate. So obviously we're talking about playbooks right now. This is some low-hanging fruit for where AI, AI is right now. Do you think as we move into the future that more functionality... Right? The AI's gonna become more complex, and, uh, do you think it's going to, to kind of change the way consultants are looked at, and there's going to be this kind of new role that becomes maybe a, a prompting consultant and, and training is more focused on the AI than the actual configuration of these CLM tools? Yeah, I think so.

11:24Part of it is you look at as early as we are and just how we can use the technology, but in thinking through, you know, how you implement it and how you prompt it, I really think you see the value in testing things out, seeing what works, seeing what doesn't work, especially because once you know what doesn't work, then you can really start thinking, "Okay, here's what I, I wanna be able to do with it. Here's what I wanna leverage. Here's something that may not quite be ready for full delegation." And so you can kind of see, we always talk about kind of getting it out of your head and into the tool, and that's really the value in the process. So if consultants can do that, uh, I think the closer they are, uh, to the business and, and understanding it the way you guys do, you'll see much more value than someone who maybe not as Involved with, uh, here are just five rules that you should have for an NDA and just run with it. That was going to be my actually next question, Nate. So maybe for people that haven't used not... Well, I'm pretty sure, like, by now a lot of people have used at least ChatGPT to run some document a- analysis.

12:28But can you give us, like, uh, some example on how you have trained AI? Like, actually what are the steps so our, uh, our audience can start to think about, like, how maybe before jumping into hiring a new software, like, what's the homework they need to do in order to start building, you know, like, playbooks on, on how can train a model to analyze those contracts? Yeah. Way it comes back to CLM a little bit, because it's kind of the same process that we talked about before. You can't just expect that you can just install the tech and it will magically fix everything. That's part of the, the risk with AI, is that it really it seems like it can do that, right? You can, you can use gen AI and it, it'll spit out something that's polished and looks perfect. Uh, but if you don't a- approach it with a critical eye, you may miss something.

13:17So what we've seen with CLM, and, and you guys know this probably better than most, is that you have to go through that hard work with developing those processes, identifying them first, before you can actually implement anything. If you don't, you're gonna have tech that either isn't gonna be adopted or is just gonna be doing stuff that you don't anticipate or don't expect. Same thing with AI. When you develop your playbook, it's just gonna follow the instructions you give it. So before you jump in and say, "Hey, we don't have a playbook today, but we're gonna get AI and we're gonna now use a 70-page or 70-clause playbook," that's probably not the way to jump in. You probably wanna sit down and say, "Look, here's what's really important to us. Here's what we care about. Here's what we want to ensure that are in all of the other contracts we review." And think about it in terms of, as you said, taking it off your plate, what does that mean, you know, taking it off your plate? What issues would you feel comfortable giving to the business team enhanced with AI?

14:15And so one example that I like is, is really low-hanging fruit. With my organization, we have kind of a, a group of similarly named entities that are actually different. And so one tool that, or one rule that we have is, is this contract with the right entity? And here are the three entities, and you can check for that, and that's something that now is an easy lift for AI. It's something that, that the business team can quickly review and validate. And you just kind of get it under your belt as an easy win, whereas it's not identification, it's not limitation liability, but it is something that you can have the business team do with AI, lets you kind of get your feet wet and move along. No, yeah, you know, Nate, I was gonna s- ask, uh, to wrap up, you know, one maybe takeaway or, or a piece of low-hanging fruit, but that, I think that's a great action item for folks that are out there that are just getting started. I mean, anything else come to mind like that for maybe folks that are just trying to dip their toes in the water when it comes to, to AI and the functionality that it can help with? I like to say test and try and fail.

15:14There's that Gandalf prompt engineering game that's out there. Try and find a way to do that with a contract. Put in a clause really well and see if you can get AI to redline it the way you would expect or catch something that maybe you wouldn't think of. And it's not so much that you wanna use it to do that going forward and get that efficiency immediately, but it's really to see where it's doing something that you may not expect. So then you can refine it, you can change how you think, change how you prompt, and develop it from there. That's great. No, I think that's a, a great takeaway. So Nate, thanks so much for joining us. It's been a pleasure. Yeah, same here. Thank you, guys. Yep. And thanks everybody for listening to another episode of Contract Heroes. This episode of the Contract Heroes is sponsored by Koho Consulting. Koho helps organizations of all shapes and sizes find, implement, and manage their chosen CLM tool. Whether your project is buy or sell side led, Koho can help you navigate the waters of the CLM space to make sure that you're getting a product that's going to fit like a glove years to come.

16:15If you'd like to learn more about Koho, visit their website at C-O-H-O consulting.com.