The Legal Department Podcast with host Stacy Bratcher

No Really, This Is How To Use AI In The Legal Department With Sterling Miller, CEO & Author Of Ten Things You Need to Know As In-House Counsel (E76)

About | Transcript | Episode Resources

No Really, This Is How To Use AI In The Legal Department With Sterling Miller, CEO & Author Of Ten Things You Need to Know As In-House Counsel (E76)

Posted on by Stacy Bratcher

You may be thinking, “not another episode on AI! Haven’t we covered that already?” Well, you’re right. The Legal Department has covered AI, but we haven’t talked practically or tactically about how you would actually use AI in an in-house legal department. In this episode, Sterling Miller, the OG of In-House, shares practical tips on how in-house departments can safely use AI.

Other discussions include:

  • How to deal with confidentiality and attorney client privilege with AI tools
  • Practical, step-by-step suggestions for drafting prompts for different tasks
  • AI use-cases for in-house lawyers
  • How to navigate e-discovery

Sterling also shares his favorite AI tools for legal departments–put together after 2 years of research, so you don’t have to!

This is an episode you won’t want to miss. Bookmark it and listen on a flight so you can take notes!

—

Listen to the podcast here

 

No Really, This Is How To Use AI In The Legal Department With Sterling Miller, CEO & Author Of Ten Things You Need to Know As In-House Counsel

I’m Sterling Miller, CEO and Senior Counsel at Hilgers Graben. My fun fact is that I love to cook. I especially love to cook with a slow cooker, otherwise known as a Crock-Pot.

—

On this episode of The Legal Department, I’m excited to welcome back a three-peat guest, my friend, Sterling Miller, who is CEO and Senior Counsel at Hilgers Graben in Dallas. He is a six-time author of The Productive In-House Lawyer: Tips, Tricks, and the Art of Getting Things Done. We’re going to talk about AI. I’ve done several episodes on AI, but we’re going to get granular about all of the things we’re scared about, how to address those, and what we can be using it for in-house. Sterling, so excited to have this conversation with you.

Me, too. Thanks for having me back.

The Detour To AI: A Skeptic’s Journey

As I mentioned before we got on, you give great content. I’m excited to share this with the audience. I want to first delve a little bit into when I was reading your book, the intro talks about that. You took a detour from writing the book to spend a lot of time learning about AI. I thought that was super interesting. Can you tell us why you had this passion project of writing a book, and then got detoured down the rabbit hole of AI?

The productivity book was something I’d been thinking about for a while. I’d written a lot about productivity in the blog. I was putting a lot of that together, thinking of new things and ways of looking at it. This would have been late 2023. I had started hearing about generative AI, ChatGPT, and this revolution coming. My first thought was it’s a lot of hype because I was there for the first wave, which was about seven or eight years earlier. When I was in-house at Marketo, we looked at AI contract tools, and they were not very good.

I thought, “This is just more of the same, so not going to get too excited about it,” but it was really hard. You got into early 2024 to avoid more and more stories about ChatGPT and generative AI. I was probably about halfway through the book. I like to write about technology that factors into productivity. I thought, “I’ll just give this a look.” A friend of mine, who’s the general counsel at Hunter Fan Company, said, “You’ve got to try this. It’s amazing.”

I went on, I signed up, I put in a few prompts, and I was completely blown away by how good the responses were and how useful they were. Most importantly, this took two or three seconds to get a checklist, to get a clause, and to get something really useful with minimal effort. This is going to change things. I had to stop. I stopped writing. I went and spent a couple of months trying to figure out how to use ChatGPT from an in-house lawyer perspective. If I were sitting in the chair, what would I be trying to do with it? It was fantastic.

I have a really good chapter about it in the book. It was such an inflection point for in-house lawyers in particular, but the practice of law also, that it deserved more than just a passing mention. I’m glad I went down the rabbit hole because I use it almost every day, and I find new ways to use it. I’m eager to see what comes next because the capabilities and the possibilities are virtually unlimited.

AI Concerns For Lawyers: Trust, Privilege, And Confidentiality

I want to dig into a little bit of the why and the how, but also, I heard some skepticism as you were describing how you got into it. Lawyers are not wanting for skepticism, right?

Yes, I’ve heard that.

I share your trepidation around getting started. There are some legit reasons why lawyers in particular are a little jumpy about getting into using AI tools. I thought we could talk a little bit more specifically about those concerns. How do you trust the bot? How do you deal with privilege, confidentiality, and those things? Can we talk a little bit about what those concerns are and how you were able to get comfortable?

You nailed some of the biggest ones right off the bat. Is my data confidential? How will I deal with attorney-client privilege issues if I am putting sensitive data into this AI tool, this bot, or whatever the case may be? There are several other concerns, but let’s just take that one first, as probably the biggest. First, if you can get an enterprise version of whatever tool you’re interested in using, generally, those agreements come with clauses that they will not use your data to train. They won’t mix your data with other people’s data. It’ll be segregated.

Assuming you trust what’s in the contract, and I understand that that’s still something that people have to get over, then you can feel pretty confident that the confidentiality obligations and privilege obligations are going to be fine. Sometimes, there’s a level below enterprise called business, which isn’t as expensive, but has the same types of protections. Most of the tools have some ability to limit how the data is used. You have to go into the settings and find something that says, “Don’t use my data to enhance everyone’s experience, train, or whatever the case may be.” You switch those knobs to the right place. 

That’s some comfort, but without an enterprise version or something that is a professional-grade, my biggest trick is I just don’t put anything confidential or that would trigger attorney-client privilege issues into an AI tool. I tend to use ChatGPT the most. I’ve also used Claude. They’re both good for legal issues. If it’s a contract on the web, you don’t have to worry about anything. You can run it through ChatGPT, and you’re fine. 

Let’s assume it’s not. If you’re willing to spend the time to get the payoff of the AI analysis, what I do is I just do universal changes. I’ll change vendor names to Vendor Co. and Customer Co. You have to spend some time doing that. You have to decide. Is it worth it? If it is, and you clean that document up, you can certainly put it through. They’re not going to be able to tell who you are or who the other side is. You can ask the prompt. 

I’m sure we’ll talk about prompts in a little bit. “I want to analyze this contract. Set out the pros and cons. Look for these issues. Compare it to a playbook, or whatever you might have.” That’s the cheap way of doing it. There are also so many things that don’t require confidential information. You’re working on a contract. You see that they left out an indemnity process. They have an indemnity clause, but it has nothing about the process. 

I could probably do one from memory. I could go to a clause bank and pull it out. A lot of times, I just have ChatGPT or Claude open. I just go, “I need an indemnity process for a SaaS agreement.” There’s nothing confidential in that prompt. I can look at what it says. I can cut and paste it in. I can edit it. That’s incredibly fast for me to do something like that.

AI Won’t Replace Lawyers: Human Judgment Is Key

It is managing how you use the tool. This isn’t a repeat comment, but people perceive that these tools are replacements for lawyers, or that’s a concern around it. You have to have that legal judgment to use the tool. How am I going to prompt it? What materials am I going to use? I have to own the management of the privilege. 

AI won't replace lawyers wholesale. But lawyers who don't use AI will be replaced by those who do. Share on X

I think I put this in my blog on AI. I don’t think AI is going to replace lawyers, certainly not wholesale. Lawyers who don’t know how to use AI will be replaced by lawyers who do because it’s going to be that powerful a tool. There will always be the need for human intervention if you just think about it practically, so step back. You’re in the boardroom with the C-suite or the board. They’re not going to want to type their questions into a computer. They’re going to want the general counsel, a person in the room that they can ask questions to. 

They may say once you run that through the computer, maybe, but they’re not going to trust a machine, at least not in our lifetimes, over something that they have a person in the room. I had a friend who was the general counsel of a company. She was giving her boss legal advice. The boss ran the question because he wasn’t getting the answer he wanted. He ran it through ChatGPT, and he got a different answer. He goes, “Why do I need you? I could just use ChatGPT. You’re wrong.” The unfortunate thing about that is now she has to go spend time proving the crap that came out of ChatGPT on that issue was wrong. It was wrong because the prompt was bad.

Debunking AI Hallucinations: Treat It Like A Summer Associate

Let’s stop there because that is the next concern. How do I have confidence that this bot is going to produce work product that I can rely on? In one of your blogs, you compared it to working with a summer associate.

You have to understand AI’s limitations—that's a really good way of thinking about it. Share on X

You have to understand the limitations of the tool. That is a really good way of thinking about it, a smart summer intern or a first-year lawyer summer associate. If you remember when that was you, and you went into the office with the partner, how much time did they spend with you explaining what they wanted? “Here’s the context. Here’s what I need. Here’s what it should look like.” You bring something back to them, and they’d go, “Sterling, this is okay. Great first start. Let’s follow up with the following.” You have to iterate on what has been produced. You would never have a first-year lawyer, for example, review the work of a first-year lawyer and then send that work product over to a client. It’s the same thing with ChatGPT. 

It’s going to give you a really good start, a good base, but you still have to put the human touch on it and look at what you’re getting. The better the data, the better the response. You hear about hallucinations, which is a terrible word if you’re thinking about, “My tech project is hallucinating.” What that means is it analyzes all kinds of data based on whatever question you’ve asked it. It’s coming back with things that aren’t correct or true. For lawyers, you generally tend to hear about this in terms of citations, case citations. Someone has used the tool to write a brief and put citations in, and didn’t bother to cite check, Shepardize it, or anything like that. They just submitted it to the court. They’re still doing it even though it’s been in the news.

It’s unbelievable. There’s a court opinion that cited twelve made-up cases. I’m going to just do a shout-out to an episode I did with Alex Alben, who is a technology professor at UCLA law school. He framed it this way. He said the AI wants to please you, especially if you know, like, “I know there’s a case.” I’m hearing a law firm partner tell me, as I’m a junior, “I know there’s a case.”

There’s a case out there that says yes.

Professor Alben said that it wants to please you. If you keep asking it the same question, you are more likely to get a fabricated hallucination answer. That’s just a callout. I want folks to know that you got to be the captain of the ship and bring your common sense into using the tool.

How you’re approaching it also goes down to the prompt, the questions that you’re asking it, the specificity, and then what data are you pointing it to? If you have curated data, say a tool that specializes in legal data, that’s going to be better than a tool that just searches everything in the known universe in terms of getting you answers that are more reliable.

Mastering The AI Prompt: Beyond A Google Search

Let’s stop on prompts for a minute. You wrote a blog on the hundred AI prompts for legal.

They are practical prompts.

Real basic, a prompt is not a Google search. Let’s dig into a little bit of what a prompt is, especially from a lawyer’s perspective.

I think of a prompt like a recipe. It’s a series of steps to get to something good at the end, like a cake, a roast, whatever the case may be. You are setting out a number of parameters for the AI tool or summer associate, however you want to think about it. “These are the steps I want you to follow. This is how I want the end product to look.” That list of things can be incredibly long. I even put one in the blog at the very end. I think it’s two and a half pages long of a prompt. That’s how complex you can go. 

Most in-house lawyers don’t need to do that or aren’t going to have the time or the willingness to do that. What you need to be thinking about as you’re writing out your recipe is that you give it a persona. You are what? You are an expert lawyer in data privacy issues or contracts. Who do you want the generative AI to be when they’re giving this answer? You’re a school teacher, but you probably don’t want the school teacher weighing in on the legal issues. Tell it what you want it to be. What do you want in terms of format? “I want a grid. I want a checklist. I want a PowerPoint presentation. I want an email.” Whatever it is, tell it what you want. 

What’s the deliverable?

What tone do you want? “I want you to be persuasive. I want you to be neutral. I want you to be humorous.” Who’s your audience? “I want you to write this at a level that a high school student could understand or a sophisticated business client.” All of those different steps go into it as you’re crafting the prompt. What’s the context? Again, think of it as you’re sitting across the table with that eager first-year lawyer summer associate. You’re explaining, and you’re trying to do that data dump. It is like a data dump.

Lastly, a good prompt breaks things down into steps. One of the reasons you get hallucinations or bad responses is if you just put too much in at once. You can say, “I want you to do this step first, then do this step, then do this step.” One of the steps I learned that makes a huge difference is, for example, if you want to look at a contract, the first step should be to read the contract. If you don’t put that in there, it may go, “There’s a contract that does this.” It’ll just go off and start thinking about contracts that are like that without ever reading the contracts.

That’s a really important point. 

You got to know about baby steps here.

That is important to focus on because the output is so sophisticated. A person could get under the misconception that it is super smart and it knows exactly what you want because when it communicates with you, it communicates with a sophisticated tone. It is really important to say it’s still a machine. It needs to know exactly what it is you want it to do.

AI isn't thinking, it's predicting. Share on X

It is not thinking. It is predicting. The whole science behind it is all around predicting what the next sentence should be and the next word should be based on the algorithm that it uses. The most important thing about a prompt in my experience is that after you get the answer, iterate going back. “I want you to expand section five and talk about these things.”

Again, it is just like when you were a summer associate and the partner was talking to you. “I want you to go back. Sharpen this or talk about this. Add this and this. You missed this whole point.” That’s funny. You get these back, and you go, “You missed this entire argument. Can you go back and add that?” “I’m sorry, Sterling. I will go back and add that.” It’ll do it over. That iteration is really important because that’s how you get to the end product that is useful.

The Human Touch: Why Lawyers Still Reign Supreme

This is another point of why AI is not going to replace lawyers. Even if that CEO can type in a search and get an answer, it is not a nuanced answer that is framed with legal judgment. As you said, the prompt was wrong. It didn’t incorporate all of the steps that were all the factors that the bot needed to consider.

That is right. You wouldn’t have the first-year lawyer advising the CEO. They’d be so intimidated. The CEO is going to say, “This is right. Correct?” “Yes. That’s right, sir.” That’s what you get.

Let me ask on tone. Do I need to ask it to be persuasive, neutral, or whatever? Is there a default tone if I just said, “Help me write an indemnity clause for a SaaS contract.”?

The default at that is something I hadn’t thought about. I think it’s just a matter of fact that is the default, at least in my experience. It’s not trying to be funny. It’s not trying to persuade me. It’s just, “Here’s what I found.” You can go back and tell it to change the tone in some way. “Make it more professional. Lighten it up. It’s so dry. Add a little humor to it,” depending on what you’re asking it to do. As a default, very Joe Friday, if you remember the old program. That’s how I usually see the answers coming out.

Aside from those legal worries, one of the things that’s on my mind about AI is not folks in our vintage who have been trained in the manual go to the books for Shepardizing and all that, but folks who are early in their career, where there’s less time and people are working virtually. There are no libraries anymore. They may not have the training or expertise to discern what the AI work product is.  

That is such a great point. One of the things that is incredibly helpful is just being old. You can see something and go, “Yes, that’s not right,” or “Eighty percent of that is right. The other part isn’t,” because you’ve just dealt with this stuff before. You’re using generative AI as a helper to help you get something done faster. If you are a first or second-year lawyer, you just don’t have that background.

You have to be very careful how you are using it and understand that you’re still responsible for making sure that the citations are correct if you’re using it for that. It’s been Shepardized and all the things that you would want. My daughter’s a second-year lawyer at a big law firm here in Dallas. We’ve talked about this a lot. They use Harvey, which is a really powerful AI tool for law firms, expensive but good. She goes, “It’s so easy just to ask Harvey to do something for you, but don’t let that become you because they don’t need you.”

Don't let AI become you — it’s a great tool, but your unique expertise is why you’re needed. Share on X

You don’t know what you don’t know.  

It’s a great way that you can read it, like someone handing you a sample when you’re a young lawyer. I can remember someone saying, “I want you to draft this type of contract. Here’s one that we’ve used before. Read it.”

It is a template. 

“Understand what’s in there and what the different sections are. If you have questions, ask me.” First-year lawyer, what’s an indemnity clause? They don’t teach you that in law school. Think about that, unless you take a very specific course, maybe in contract drafting, which there wasn’t one when I was in law school, at least at my school.

No, they just called it contracts, right?

That’s right.

It is Hadley versus Baxendale.

You learned indirect damages and all that stuff that doesn’t come in useful when you’re drafting. If there is anything for the young lawyers, don’t over-rely on it.

Practical AI Use Cases For In-House Counsel

It’s just a tool in the toolbox. I know a lot of your focus is on the in-house crowd. I love that. Those are my people. I was with a group of my GC colleagues, and AI is always at the top of the list. We’ve got business partners that are using it or want to use it. There is an expectation that the legal department is going to get on board, too, with AI. My colleagues and I were talking. “What would I use it for?” In-house, you don’t do research really. Maybe you look something up, but you don’t have time to do research necessarily. You might use it for drafting. There was just a lot of uncertainty about, “Do I need this?” What are some use cases for the in-house crowd?

We mentioned the blog. That’s Ten Things You Need to Know as In-House Counsel. I wrote a hundred prompts that were all aimed at being an in-house lawyer. A practical example, a useful example, and one that wouldn’t require you to have an enterprise or business level subscription is, let’s take a term sheet, a very common document. The business sends you an email. “Here are the basic terms. Can you guys prepare a term sheet?” Sure, it could take you an hour or an hour and a half, maybe. It just depends on how elaborate you want it to go.

If I wanted to use AI, here’s the prompt that I would use. “You are in-house counsel for a mid-sized business that is seeking to buy another company. You’re a corporate law specialist. The target company sells analytics software. The company has 2,000 customers, owns five patents, and generates $25 million in ARR. You will have a 60-day exclusive period to negotiate with a purchase price of $50 million. Draft a term sheet for an asset purchase of the target company. Delaware law applies. The term sheet should be non-binding other than for confidentiality and exclusivity obligations.

The deal is subject to due diligence. $50 million will be paid as follows, $25 million in cash, $15 million in stock, and $10 million in an earn-out over two years, with the conditions of the earn-out to be negotiated before any definitive documents are put together. The two founders of the target company must agree to become employees of your company and help operate the acquired assets for at least two years. Include other standard provisions for an asset purchase.” If you put that in, and that is a good paragraph, you will get a two or three-page term sheet that is 90% of what you want. You will have to go in and tweak it, but I guarantee you. You will be amazed at what you get. Have you ever used AI for anything like that?

Not for that.

That’s one of my favorites. There’s nothing confidential in there. If someone saw that, they wouldn’t know who the company is or who the target is. It would have no name. There’s nothing in there. You can even put fake numbers if you’re worried. Double the number. When you get it back, you’re going to cut it back.

I got to say. Back to one of the questions I asked you about prompts, that’s super clear that this is not a Google search. To the point for our in-house colleagues, to get up to speed in using AI is really getting good at that prompting because that’s the legal work. As you describe the different terms of that deal, that’s the legal work. 

I’ve created several presentations on generative AI, which I like to do for legal departments or whatever conferences. The first ones, I probably spent a lot of time talking about the history of AI and how it works. That’s a little interesting because I could look at the audience, and their eyes are glazing over. “Who gives a shit how it works? I just want to use it.” The second half of the presentation was, “Let’s pull up live, and we’ll do some prompts.” I have a list of prompts. I have some documents I can scan, and all this stuff. Everybody gets excited.

The newest presentation that I have is all prompts. It is a little bit of, “Here’s a bit of the background of what’s going on. Here’s how you use it every day.” That gets people excited because the first thing it does is it shows you it’s easy. It is incredibly easy to prepare a useful prompt. You don’t have to have two or three pages. It can literally be a paragraph, maybe even shorter. I had a client who called me up and said, “Can you get on the phone and talk about insider trading with me? We’re thinking of going public. What is it that I’m going to need to know?” I thought, “It’s been a long time since I’ve dealt with insider trading.”

I just went to ChatGPT and said, “Give me a checklist of things you need to know about insider trading and creating a policy.” Three seconds later, I could look at it. I could know what was good or what was bad, but that one sentence got me something that I could get on the phone. I had something that I could look back on and refresh my recollection. I had the experience, so it wasn’t that I was just making it up. I can, but that is the way that it can be incredibly helpful.

I just want to emphasize. It’s a tool. It’s not a substitution for that legal training and experience.

The other thing it can’t do is this. I can remember this. I’m sure all your readers, if you’ve ever dealt with clients, there’s so much body language and nuance to conversations, inflection, tone, and voice. You can tell, “Are they asking me to do something squirrely or completely legit?” You can ask a lot of hard questions, but if you’re a tool, you’re not sussing that out. You’re just answering whatever it is that person asked you.

You’re trying to please them, which is certainly something you want to try to do as a lawyer, but your obligation is to the company. That is your client. If you think someone is trying to do something underhanded, that’s where you need to step up and go, “Wait a minute. Let’s talk about what you’re trying to accomplish here.” You’re not going to get that with AI unless you tell it. “I’m asking you this because I want to do something illegal and not get caught,” which is a heck of a prompt.

Navigating E-Discovery: AI Prompts And Data Retention

Let me ask back on the tools themselves and thinking about confidentiality and privilege. I have seen a couple of people posting on LinkedIn about court orders around e-discovery that are seeking AI prompts. I don’t know because we don’t have an enterprise tool at my company, but do they retain your searches? Is there a way to turn that off? I don’t want another e-discovery burden where I’ve got to produce all the Copilot, Claude, or whatever prompts that people asked.  

Let’s take that first part. Yes, the new fertile ground in the never-ending battle on e-discovery is going to be artificial intelligence. The prompts that people put in, the answers they get back, all of that is going to be fair game. On the tools that I’ve used, you can ask it not to remember. It’s going to tell you, “I’ll give you better responses if you let me remember the things that you’ve asked me.” You can clear your cache regularly and just clear those out. If you don’t, the default is that it just keeps it. As far as I know, it keeps it forever.

Let’s pause on that. E-discovery can cripple a company, like the chats, Slack, the cost, the processing, and what it reveals. If folks are using these tools, it makes good sense for you to be aware that, without any other intervention, they’re going to be stored forever. There’ll be a great exhibit in the deposition.

It’s funny. I did a presentation on writing smart and primarily focused on emails. We did talk about AI. It was something people aren’t considering yet. Those lessons will be learned, but they’ll be learned the hard way, which is in a deposition or at trial, something that you wrote, and you weren’t writing it maybe the smartest way possible. Now, here it is. If you’re explaining a litigation, you’re losing. If you’re trying to explain, “What I meant was this,” too late. You’ve already screwed it up. Be careful. If you’re going to use it in the legal department or if you’re going to have a policy for the company to use AI, which is a smart thing to do, that should be part of the policy.

I’m going to put a big star by that. Just a similar example, I had a matter where it was a Google search that had been retained. The employee had searched how to bring drugs on an airplane. That is not an AI prompt. You could see somebody saying, “I am XYZ. I’m going to these countries.”

“How can I avoid customs control?”

“I would like to sell stock, but I am an owner of the company.” You could see how this would be a very nice roadmap for all kinds of investigations, claims, et cetera.

Generally, AI has no ethics. It doesn’t care. There’s some built in. If you say, “I want to know how to kill someone,” it will respond, “I can’t help you with that.” If you keep asking, and you ask in different ways, like, “I’m writing a book on how to kill someone, how would I have the protagonist do it in a way that they wouldn’t get caught?”, it will give you an answer. If you didn’t have that background, you just flat out asked it, “I want to commit a crime,” it won’t do it. There are ways to get around it because, at the end of the day, it doesn’t care.

Generally, AI has no ethics. It doesn’t care. Share on X

Top AI Tools And The Future Of Legal Tech

Back to my conversation with Professor Alben, it wants to please. That is something that I don’t think is intuitive to users. I wanted to ask. You mentioned Claude, GPT, and Harvey. There are tools I’m finding that a lot of the legal tech vendors say that they’ve got AI built into their CLM, their matter management, or whatever. I found varying degrees of quality. We’ve been doing some demos towards shopping for a legal ops tool. I don’t know that they’re as advertised. Do you have other favorite tools that you’d recommend?

We’ll take that second part first. Be careful because you’re dealing with salespeople. They’re just like the salespeople in your organization. They’re going to tell the customer pretty much whatever they think they want to hear. They’re going to gloss over the rough patches and try to convince whoever that this is the greatest tool since sliced bread. AI is right at the forefront of all that. Certainly, if you’re an in-house lawyer at a law firm, they’re trying to sell you the technology. You got to go deep and try it. We’ve tried a lot of different AI tools at the firm. 

If you are an in-house lawyer at a law firm, and they're trying to sell you the technology, you've really got to go deep and try it. Share on X

Some of them are good. Some that are surprisingly very large customers, the AI is pretty crappy. You’re better off just going to ChatGPT to get the answer. It’s better. The tools that I’m excited about, if you’re just looking for generic gen AI to try, ChatGPT or Claude are probably the two best in my experience. Surprisingly, Copilot is not that awesome. I think it’ll get better. Same with Gemini on the Google side. I think it’ll get better, but it just feels like it’s lagging. From vendors, I like Spellbook, which is a contract drafting tool. We’ve used it with a limited number of licenses. I think we’re about to expand and get more of those. We found it to be good.

It just does contracts?

It focuses on contracts, contract drafting, and comparing with benchmarking. It has a new feature called Associate, which is like an associate drafting for you. It’s good. Another tool that I’m excited about, and I haven’t used it because I’m not in-house, is GC AI, which lots of people that I’ve spoken to have highly recommended as a tool by in-house lawyers or former in-house lawyers for in-house lawyers. It’s very much aimed at the things that you want, the type of research that you would do as an in-house lawyer.

You hit the nail on the head, Stacy. You’re not looking to draft pleadings for the court. I need the down and dirty. “I want to fire someone in Tennessee. What do I need to know? What’s my checklist? Is there anything I need to be aware of?” California is someplace a little more complicated. It can do that type of research for you. LawVu is another great tool. It does a lot of different things. It’s a neutral platform as well. It can integrate with other tools that you have. It’s matter management, legal ops. I have found it to be exceptional. I’ve recommended it to several in-house legal departments. They’ve implemented it, and they love it. That would be one. 

That wasn’t on my list for legal ops vendors. I’m going to take a look at that one. Thank you.

Ironclad is another contract CLM-focused tool, but the tools that I’ve seen from Ironclad to analyze the contracts that you have in the database and pull out the information that you want are probably in the top three of the companies that do that. The whole new frontier, I’ve been hearing the hype building up, but I haven’t got into it. It is agentic AI, which is AI that’s going to connect different applications. You can give it a task, and it will go use those applications for you. For example, let’s say I’m a law firm lawyer. I have meetings with potential clients. I could say, “I want you to check my calendar for meetings with potential clients, then go pull news stories about that company from the last six months. Put that into a package that I can review before I meet with the client.”

It’s going to look at your calendar, and it’s going to go out to the web or whatever. It’s going to pull sources and prepare a report. It’s even more human-like in terms of the tasks that you can tie together. It’s a little bit like, if not then this, or whatever that tool is, where you connect all your apps. Zapier, I guess, is a better example. It’s going to do that stuff for you. All the concerns we just talked about are magnified a thousand times. It’s going to go into my calendar. Am I okay with that?

The hair on the back of my neck stood up when you talked about that. I was pumped about the others, but I’m like, “I don’t know.”

That’s coming. That’s one of the things I’m going to be spending some time on.

I’m going to wait for your next blog on it.

It’s coming. My next one that I do every August is I do cool tech, where I try to find technology that is useful for in-house lawyers. I’ll probably have some of this stuff in there. I also try to find things that are very inexpensive or free that are cool. I like technology. I’m not a technical person. I did not do computers, computer programming, or any of that stuff. I just find it helpful and interesting. If I can use it, then all of your readers need to know that they can certainly use it, because I just don’t have any special skills other than I know I can’t break it in general. I just get banging away on it and see what happens.

Sterling Miller, you are helpful and interesting. That’s why you’re a three-peat guest. I appreciate you being here and, hopefully, getting folks a little more comfortable with using AI in their department. I want to thank you for being here.

My pleasure. 

The Soundtrack Of Success: Music And AI Insights

I forgot the pump up song. You are a three-peat, so people have heard. I know you’re a musician, but I don’t know if you have a song of the summer that’s in your mind these days.

I do not have a song of the summer, but I like to listen to old songs. One that I’ve found that I like is a band from the ’70s called Roxy Music. They had a song called More Than This, which is just a fabulous song. There was a group in the ’80s and the ’90s called 10,000 Maniacs. They did a remake of that song, and it is just gorgeous. More Than This by 10,000 Maniacs has been near the top of my playlist. I would just say to your readers, if you’re interested in this, reach out to me. If you want a presentation for the legal department, I enjoy doing that presentation, especially the live prompts, because you never know what’s going to happen. It could fail. It could be awesome. It’s like being on a tightrope, and it’s a lot of fun. Throw prompts at me. Let’s ask it something.

There is exhilaration in the legal department.

When people see that, they get so excited because they know, “I can do this.” That’s the fun part. It’s here. Get used to it because it’s not going away.

Sterling, where can people reach you?

You can reach me at the firm at Hilgers Graben. If you go to HilgersGraben.com, you’ll see me right at the top row, the people. My email is there. You can go to the blog, Ten Things You Need to Know as In-House Counsel. All my contact information is there as well, or on LinkedIn. I love to connect with people on LinkedIn, the last sane bastion of social discourse on the internet. I’m still there for now.

Thanks so much.

My pleasure, Stacy. Thanks for having me back.

   

Episode Resources

   

About Sterling Miller

The Legal Department | Sterling Miller | Legal Department AI

Sterling Miller is CEO and Senior Counsel. at Hilgers Graben He has enjoyed a wide and varied 30-year legal career. He is a three-time general counsel, having served in that role at Marketo, Inc., Sabre Corporation, and Travelocity.com where he managed international teams handling a myriad of complex legal issues including “bet-the-company” litigation, M&A, IPO, corporate secretarial, data privacy, government affairs, and commercial agreements.

Out of law school, he was an associate at Gallop, Johnson, and Neuman in St. Louis. He began his in-house career in 1994 at American Airlines. He led the legal team for Sabre Holdings’ successful IPO in April 2014. In January 2019, after leading the legal team that worked on the successful sale of Marketo to Adobe, he returned to the Dallas office of firm Hilgers Graben PLLC for a second stint as Senior Counsel, focusing on commercial matters, litigation, and data privacy. He has published six books.

Sterling earned his J.D. degree from Washington University in St. Louis (Order of the Coif/Law Review) and his bachelor’s degree from Nebraska Wesleyan University, where he was student body president among other honors. He is certified in data privacy/security by the International Association of Privacy Professionals (CIPP/US). He is a highly sought-after speaker for legal department offsites and for legal and industry panels including presenting at the South-by-Southwest conference in Austin, TX and keynoting several ACC events. Under his leadership, Inside Counsel magazine named the Travelocity Legal Department one of the “Ten Most Innovative” legal departments in the United States (with the Sabre legal department later winning honorable mention under his tenure). He has been a multi-year finalist for the Dallas Business Journal’s “Best Corporate Counsel” award.

Sterling serves on the Board of Directors of Terrasoul Superfoods and the advisory boards of Travefy.com and Kyte. He publishes a widely-read and award-winning legal blog entitled “Ten Things You Need to Know as In-House Counsel.” The ABA published four of his books: “Ten Things You Need to Know as In-House Counsel: Practical Advice and Successful Strategies” in 2017, “Ten (More) Things You Need to Know as In-House Counsel: Practical Advice and Successful Strategies (Vol II)” in 2019, “Showing the Value of the Legal Department: More Than Just a Cost Center” in 2021, and “The Productive In-House Lawyer: Tips, Hacks, and the Art of Getting Things Done” in 2024. His third book – a cookbook titled “The Slow-Cooker Savant” – came out in late 2018. Sterling’s first book, “The Evolution of Professional Football: 1920 – 2015”, was published by Mill City Press in 2015.

You can follow him on Twitter @10ThingsLegal and LinkedIn where he regularly shares his commentary and insights on cutting-edge legal issues.

Leave a Reply

Your email address will not be published. Required fields are marked *