Subscribe
Share
Share
Embed
The Margin is a podcast from MGI Research that explores the evolving world of business monetization. Hosted by MGI Managing Directors Andrew Dailey and Igor Stenmark, the show features candid conversations with founders, CEOs, product leaders, and industry experts at the forefront of pricing, billing, and revenue operations. Each episode dives deep into the strategies, technologies, and trends shaping how companies generate, capture, and grow revenue—from subscription and usage-based models to AI-driven monetization. Whether you're in finance, product, or IT, The Margin offers practical insights to help you navigate complexity and drive growth in the digital economy.
Andrew Dailey: Hello and welcome to The Margin, a podcast exploring the forces shaping business modernization. I'm Andrew Dailey, Managing Director and Analyst at MGI Research. Despite substantial investment in e-signature and contract lifecycle management systems, businesses still struggle to answer fundamental questions about their contracts and agreements, which customers can cancel at will when their contracts renew. What custom pricing agreements are in place? The reality is that much of this valuable contract data remains locked away in fragmented systems, or buried within PDFs stored in legacy contract repositories. While contract creation and management promise to be ideal candidates for AI, lawyers remain slow to jump on the AI bandwagon. Joining me today is Praful Saklani, CEO and co-founder of contract management and intelligence platform Pramata. Saklani has decades of experience tackling the challenges of contract lifecycle management, not just as a legal tool, but as a business asset that can drive revenue growth, improve customer relationships, and even have a positive impact on margins. In today's conversation, we'll discuss why CLM adoption has historically fallen short. AI's role in transforming contract intelligence, and how businesses can unlock the full value of their contract data. Saklani will also share his perspective on the current generative AI hype and what it really takes to implement AI driven solutions that deliver real, tangible results. Praful, welcome to The Margin.
Praful Saklani: Thank you so much, Andrew. Really appreciate you having me.
AD: Let's first start. Why are we here? Companies have spent all this money digitizing contracts. And yet when you look at the ability to interrogate contracts, to share contract data, simple things, it's nearly impossible for most companies.
PS: Yeah. So I think we're here for two reasons. And one is a positive reason, and one is a negative reason. So the positive reason is because most people realize that there's a lot of untapped value in understanding what's going on in their contracts better. And it's not just confined to the legal department. It's literally the entire organization. Because all of this rich information on your customer base, on your vendors, on what you've done with similar customers, what are your opportunities to increase revenue, increase margin, sell more, or on the vendor side, what have you done with similar vendors or overlapping vendors and what are your opportunities to manage costs? It's all in there. The whole story is in there. Yet people know that because of the reasons that you've talked about. We you know, we don't or people are not yet fully leveraging the information that should be accessible for them to become better businesses so that that's what I would say is the positive thing, which is they know there's value. The negative thing is that almost all the technology that's been deployed so far has been adopted in a patchy way by most enterprises, and we have companies that have spent or we know companies that have spent $10 million, $20 million implementing contract management over five years, ten years with some of the biggest consulting firms in the world helping them to integrate it and do change management. And when you ask them a simple question, what percentage of your contracts are even in a system? The answer is 20%, 30%. And the reason for that is that systems are really designed to standardize and simplify language but the reality is, when it comes to high value relationships with customers and vendors, maybe you can't simplify the language. Maybe you have to negotiate. Maybe there is an errant complex fee. Maybe the information is contained across 100 documents. And so this is the reason why most technology until now has fallen short in piecing together the complexity and getting people beyond understanding their simplest contract.
AD: So most organizations think about contract management in relatively simple terms, they think about it as sell side agreements, they think about contract management in terms of procurement, or they think about it in terms of managing risk or mitigating risk within the confines of the office of the general counsel. Why is that the wrong way to think about the problem?
PS: Well, again, I mean, if you look at a company's big strategic objectives, usually those strategic objectives tie to “I've got to sell more. I've got to increase my gross margin. I have to improve my profitability.” And invariably that comes to pricing and packaging. Either of the things I sell, which are my products, how do I price them? Who's buying? Who's expiring? Who can I sell more to? Right. Or it comes with my vendors. Like what are the, opportunities or levers that I have available to me to better negotiate with my vendors? So this is a C-level problem. This is a CEO-level problem. It's a CFO-level problem. It's a CRO-level problem because it directly ties to their KPIs. And everybody says things like data is the new oil or the new arms race is data. Well, fine. Then why is one of the most valuable pieces of information on your business? Literally kept under lock and key or inside these kind of impenetrable documents? Unless it's a really simple statement of work. And then. Then. Sure, then you can get a report on it. But for the deal, which you're transacting $50 million with your biggest customer, you might know nothing.
AD: So should organizations think about it as this is a productivity application that sits within a single department? Should they think about it across the enterprise? Who should fund the investment? You say it's a C-level problem, but which “C” should sign the check?
PS: Yeah. So invariably with the customers we work with and as you know, we work with many Fortune 500 companies and we have a lot of mid-market customers as well. So usually people don't start off and say, “I want to do everything all at once.” So there's a pain point, okay. And the pain point may be, you know what, I've done a lot of M&A over the last X number of years. And frankly, from a lot of the acquisitions, I still don't even know which contracts they have, under which products they have under contract. The renewal issues that came, you brought up or where can I increase prices, etc.. And so even just consolidating that information, that may be the burning issue on the, on the sell side now there. Then you ask yourself, let's just take the revenue generating contracts problem. So whose issue is that now. It could be—it's partially a legal problem if they're tasked with M&A integration or they're tasked with let's standardize our teams and seas and let's make it less brutal to negotiate with existing customers. Yeah it is a legal issue. But guess what? It's also a sales and rep ops issue because sales needs to make sure they're not tripping over themselves, going into customers, trying to sell them products that they already have under contract or they understand, oh, these are all the things they have. Maybe there's a bundling strategy that I can bring into play here based on the existing contracts in place. I want that information to be able to negotiate better. And then the CFO probably wants to know, “hey, I've inherited all these contract obligations, how do I use these to make sure I'm billing accurately? How do I use these to make sure I'm increasing prices effectively?” And so my point is what is the burning problem? That is what determines is it a CRO problem. Is it a CFO problem? Is it on the sell side? Is it a CFO problem because they want to reduce cost, increase gross margins? So there's usually a burning platform issue that should drive. But it all goes to those C-level KPIs. Eventually. That's where it sits.
AD: So let's talk about the challenge, which is organizations when they're selling, I create a quote, a contract gets created, it goes off to the legal department and some type of repository there, but the data that's contained within that then needs to go into a whole variety of different systems: my core financials, my billing system, my operational systems for provisioning, and that data. Then persists through the life of the customer. What's wrong with how most contract management systems are architected today in terms of not being able to handle that persistence of the data through the life of the customer and the relationship?
PS: So I believe that when you're talking about enterprises of scale, and let's even say any company over 500 million or billion in revenue. So first of all, over 80% of sales transactions are done with customers that you've already transacted with. Right? So this very notion of how do I get the quote correct? Well, let me ask you, is the quote off a price book? Okay. It might be off a price book if you're selling them something they've never bought before. Or did they already negotiate some pricing with me? Did they negotiate some pricing with me five years ago of the things they negotiated with me five years ago, maybe I renegotiated two of the ten things, but the other eight things are still that old price. Right? So getting that initial quote right is actually a big challenge in the first place. Do you see? Because you're not dealing with a standard price book and you're not dealing with standard discounting rules, etc., because—you want to really annoy a top customer? I'll tell you the way to annoy a top customer. Go through a brutal negotiation with them three years ago, agree on a price of like $10, and then show up two years later and say, “hey, I'm giving you a great discount. I'm going to give you this for $13.” And they'll be like, “what do you mean? You've literally increased my price?” Oh, you're right, you're old contract. Part of the issue is that when there's complexity and when there's negotiation, this closed loop of let's take it out of the price book, let's run it through the price, or let's run of it, that works great if it's a net new customer, if it's a relatively simple purchase, if there's no complex bundling. But I'm just telling you literally 100% of enterprises we know that adopt CPQ. They're like, it works really good for these three product lines because they're growth lines for us. Oh, but when you get into the traditional product line or you get into these other six product lines. Then we have to go to manual pricing, and then it has to go to the deal desk, and then it has to all go reviewed. And so once the closed loop breaks, usually that's where the data doesn't persist properly through because now all of a sudden you have this is in our format. This is on the customers and it's on third party paper. It's on the customer's paper. We customized this new bundle for them which isn't in the system. And so now you get into all the fun stuff of the data breaking and that's all solvable. And that's frankly what my company, Pramata, has been doing for well over a decade and does extremely well. But now with Gen AI, we can make it even easier for everybody to master this process of taking that unstructured data and allowing it to flow through in a structured way.
AD: So let's talk about that in more detail. Generative AI threatens to transform the whole contract management space. In fact, there are some that are saying you can just throw out your existing investments. Yeah. And completely rethink how you create, manage contracts. Where does the truth lie and what the benefit of Gen AI is? And artificial intelligence more broadly in contract management?
PS: So it's interesting. I've gone through my own journey in terms of the hype cycle related to Gen AI. Right? First I thought, “oh, this is amazing.” Then I thought, “oh, this is a toy.” And now as we've embedded it into the center of our platform, both in terms of user experiences as well as the way we manage and analyze the data, what I've understood is it is a fundamentally transformative technology, especially when it comes to dealing with unstructured information and unstructured information and paragraphs. But like any transformative technology, there's techniques, there's tools, there's design patterns. They're best practices that you need to follow in order to be able to get value out of it. So this idea that Gen AI is a magic box that you're just going to throw information into and just ask any info, any question about it, that's nonsense. That's nonsense. Anyone who even spends five minutes on GPT asking about their favorite football team, which in my case is unfortunately the Minnesota Vikings or asks about it and their favorite food, etc. will immediately spot like 15 anomalies in like a 20-minute conversation with an AI bot. Well, when you're in an enterprise, you need a higher level of precision, you need a higher level of predictability than that. So there's techniques that companies like Pramata are beginning to master that allow you essentially to take if something is standardized even traditional approaches, traditional machine learning, traditional database querying works really well in terms of understanding what's on a form or what's in the contract. But when you have a relationship that's made up of 45 different documents master agreements, amendments, statements of work, things that have been acquired in, and you want to really understand what's active right now. What should I how does this compare to my current playbook? What are the things I should be renegotiating now? You need a nuanced way of knowing how to apply generative AI to organize that data. You can use AI to extract data. You can use it to validate the extraction, but you also need it to build your playbook. You can use it to compare things against your playbook. So I literally believe there's there's that every single step in contract management is going to benefit from the ubiquitous use of generative AI. And it's going to be very simple for end users to use. But it's not as simple as just pulling up ChatGPT or clods on it and just starting to chat with it. It requires skill, it requires domain expertise. And that's why companies like Pramata are really leaning into it for our customer.
AD: So if you think about the average CFO, the average general counsel, fortunately, they tend to be very risk averse kind of individuals. And certainly in the case of the general counsel, you get a contract wrong. It's not only putting the business at risk, it's putting their personal professional career at risk in terms of disbarment. So what do you what do you say to that individual who's risk averse, who's gone through that cycle of let's play with, generative AI tools and then seeing the hallucinations and the things that come up and said, “you know what? I can't take that 1% risk.” What's the answer to that?
PS: I think first of all, people have to understand you're not going to be using AI to make decisions. And you need an AI that's not a black box in the sense you need an AI or you need a company that's helping you walk through this journey, that's going to create transparency in terms of how they're using this and how you can use it. So I'll give you a great example. So whenever we use AI and definitely generative AI as part of, say, organizing a set of contracts and telling you this is the termination. And this is the indemnification, and this is where pricing is contained. And this is what the effective date was. We use multiple QA strategies, okay. Multiple QA strategies, some of which are AI, some of which are more discreet to verify or validate. What is the confidence level associated with this particular piece of information. And all the way down to you should be able to click on something and see where the information originated from. So for example, if I tell you that a contract has an effective date of a certain date, we have a button that you can click on and it shows you the text that's associated with that effective date. So you yourself can validate that. Right. So what's going to happen is that all the time that people spend organizing information, trying to find the right three documents to look at, trying to find where within those documents to look at, etc., you can first of all, compress all of that down by 99%. And then in terms of analyzing that information again, say a compliance checklist or analyzing against a negotiation checklist you can actually use. In our case, we use a red, yellow, green paradigm. But you can explain the logic and you can give links to see your guideline says this. This is the original language that's in the contract. Here's why this is yellow. Here's why this is red. But it needs to be viewed as a decision support.
AD: Well, certainly in the case of database there's auditability, there's traceability.
PS: And that's what this is where that introspection into the logic of the AI is very important.
AD: That's right. So it's less of a black box and more of a glass box.
PS: Show me your work.
AD: Right.
PS: I encourage everybody who uses AI in their personal lives that if you see something weird or wonky from like, it's hallucinating, ask it. How did you come to this conclusion? Can you tell me? Because you'll be surprised. It will try to. These tools will try to explain themselves. But in our case, so much of it is even when we work with AI, only if we're asking a question of something 30 to 50% of the data that we're feeding it is actually structured data and the rest of it is unstructured data. So in other words, I will say this is a termination clause okay. And I'll have that termination clause identified with 99% accuracy. Then I'll say I want you to evaluate this termination clause in a very specific structure. And then ask it that question in that way, so that now the odds of hallucination greatly drop with the more precision of the data you feed it and the more precision with which you ask the question. But this is what I mean is that GCs shouldn't have to worry about this. CFOs shouldn't have to worry about this but make sure your vendor that you're using for this to make sure they're worried about it and they've mastered these techs.
AD: Do you get a lot of questions and concern around data segregation, data privacy. “What are you going to do with my data?”
PS: Of course. Oh yeah. And one thing I will say, people have to realize that the AI tools that you see today are as bad as they're ever going to be. And what I mean by that is the the pace at which, the different, foundation models, the different vendors out there have moved to address enterprises concerns around Gen AI privacy, Gen AI persistence, what's in-memory, what's the data governance around that? It's amazing to me because people recall like 15 years ago, 15 months ago, that was a when ChatGPT first came out, OpenAI, they had a lot of gaps in the way they were talking about this, because they had no idea this would go so viral so quickly. One of the vendors we work with very closely is anthropic and in particular AWS through anthropic and AWS in particular, as you can imagine, has gone way out of their way to make sure that their bedrock service, which uses, anthropic models is, is very much enterprise secure, enterprise ready, like all the guardrails around it. I mean, we're like absolutely isolated, in in alignment with all of our enterprise grade AWS processes.
AD: So we talked about some of the risks and the fears that people have. What's the penalty for ignoring kind of what's really happening at the frontier of contract management? What's the penalty for just saying I'm going to sit back, rely on my current systems and not make any change?
PS: So there's a procurement department that we've been working with that does a very, very sophisticated vendor analysis when they have to negotiate with their top hundred vendors these are strategic suppliers, right? And they're really limited at the number of categories and the number of vendors that they can do this. Think of it as a deep doctor's appointment where because they have a multiple hundred item checklist. And so they did an AB test with their current with their current approach, which is 90% brute force manual. Obviously they pull data from different things, but it's all a lot of spreadsheets, a lot of manual analysis versus a gen AI driven model and using data use, you didn't using our vendor insights, product. Right. And what they found was accuracy was constant. But the actual timeframe, both in terms of calendar time and the amount of resource utilization in terms of manual work, was reduced by 97%, okay? Not five percent, not ten percent. 97%. I feel almost like an infomercial saying that not five, not ten. 95! So here's the situation. If you are a company today relative to where the world will be in five years, odds are there are thousands of things you're doing in your company related to contracts that are efficient at a level of 95%. Okay. So first of all, from an efficiency perspective, you can obliterate the friction, okay? If you and you can obliterate the friction in a step-by-step way, measured way, crawl, walk, run. But the low hanging fruit is really low hanging okay, but here's the second piece of it. What are you not doing? Okay. Because what I like about this customer is they said, well, actually, this doesn't mean we would get rid of the team that does all of this analysis. What it means is we would do it across 1500, categories or 1500 vendors. Right? We're not going to just focus on the top 100. And now we can actually get even more leverage from the data in contracts in terms of renegotiating. Do you see what I mean? So what are you not doing because it's so painful to do that.
AD: Absolutely. As the CEO of a software company, where do you see the opportunity for efficiency inside of your shop?
PS: So it's interesting. I personally believe that there is no half measure when it comes to being an AI company or an AI-first company. If you're not using AI inside your company for your own processes, then you're not really an AI company because you're not actually leaning into it. And what I mean by that is so internally, if there's any IT related questions, if there's responses to RFP, etc., we have all of our own knowledge bases, all with AI wrappers around them so that employees can find out what is the lead policy, what is the term, how do I do this, etc. We're having our business development teams, our marketing teams leverage AI to make sure that all the outbound communications we're doing are very customized. Right. I was giving somebody the example and you can actually take a photo of somebody's LinkedIn and you can actually put that into, one of the AI ChatGPT or Claude on it. And you can say, “hey, based on this photo I'd like you to actually help me craft an email that references this person's background and asks whether they'd be interested in communicating about this point.” And you can even share, like, your key points that that you, that you as a company are trying to reinforce with the buyer. I think today as a company, we're probably using AI in our internal processes 20% of full penetration compared to where it'll be like even three years from now.
AD: So if you look at enterprise software, enterprise applications, there's been lots of innovation in different areas. As we've been discussing, one area that hasn't really seen any real innovation or or change is in the implementation period. How companies get software, bring it in the door, stand it up, test it, roll it out. Yeah, that's fundamental process across almost any business application hasn't really changed in 30 years. Is that going to stay the same? When are we going to see some kind of transformational difference a having of implementation times through the use of AI?
PS: I predict that we will not see a having of implementation times to get to current value levels. I think there is an opportunity to take it down 80 to 90%. But I do think it requires re-architecting of the systems. Because these current systems are designed to have thousands of switches and have a basically forced you to code in all the variants. But the second, the variants can actually be managed through more of a natural language type interface where you can take the translation of business logic into a system out of the hands of engineers or coders and into the hands of actual business users. That's going to I know that sounds like gobbledygook at a high level, but seriously, like if you look at it as a CEO, I take information from all kinds of different systems. Yeah, I need to get the marketing metrics, the sales metrics, I need to get the operational metrics, etc. There is literally, as of today, when all those different reports come to me, I can build visual dashboards for myself just using simple prompts, using Claude and I do okay. For our last board presentation, I just had like the 10 or 15 different things that I use and I'm like, hey, can you actually just, take this information and correlate this to this and show it to me as a, as a comparison chart and. No, no, no, actually, can you make it a line chart. Not on this chart. And yes, you can do that by dumping into Excel, changing all the—I'm just saying this speeds up that process by 90%. To the point you were making earlier, you're going to need the right governance, the right audit, the right testing. Companies will have to create their own benchmarks so they can figure out if a prompt changes or did I go backwards? Was there a regression of some kind? And again, we are going through that whole journey with contract management with our own company but I'm saying those are those are the muscles that the enterprise will have to build but people are building them.
AD: Praful, thanks so much.
PS: Thank you. Really appreciate the time and I hope this was informative to your audience.
AD: Thanks. Thank you for listening to The Margin. If you have questions about today's episode, or if you'd like to schedule a call with an MGI analyst, reach out to us at insights@mgiresearch.com. You can also reach us on LinkedIn, Facebook, and X and you can find more information about our research and advisory work at mgiresearch.com. Until next time.