My first job out of law school was writing legal content for a large legal website for consumers. When you type a legal issue into Google, this website shows up 95% of the time. And they show up because they’ve written articles about everything, in every state, and then wrote about it again, just to make sure that they had it covered. Every day, I was tasked with writing six articles, five days per week, and I did that for about three years.
It’s probably because of that content-churning gig that the potential of the first artificial intelligence platforms — which focused on writing content — really struck home with me. I remember that the first A.I. writing model releases were postponed by fears that the tools would be used for spam or misinformation (and they almost certainly are). Eventually, these artificial intelligence bots were let loose on the wider world, and I couldn’t have been more excited to try them.
Excited and, well, maybe a little disappointed, once I actually did try them.
Try asking general longform AI content writers (as opposed to one trained on actual legal contracts and authority) to write about a niche legal topic, and if you actually understand the legal topic, you’ll have some good laughs when the robot gives the absolute wrong answer or just writes meandering content full of warnings about “how important it is to get legal help, because legal help is help, and if you don’t get help, you stand to lose everything.” It kind of sounds like some of the drivel written by some of those large legal marketing agencies’ content mills that staff their teams with freelance nonlawyers, each paid $20 per article to produce something deliverable in under an hour.
Wait a second - these models are trained by scraping the Internet. Maybe the old spam is responsible for creating the new spam-bots.
However, in the time since those early tools came out, the AI has gotten smarter and more user-friendly, and we now have tools for short-form content (like tweets and ad taglines), long-form blogs, legal document review, and even image and art generation.
That’s right: as of a couple of weeks ago, I was actually able to get my hands on software where I could ask the robot to generate an image and it did so - a royalty-free, wholly original, image generated by artificial intelligence. Robots are painting pictures, y’all!
The point is this: for tech-minded lawyers, especially those of us running our own practices, this is an exciting time, and A.I. is going to give you superpowers. You don't have to hire a copywriter and graphic designer to make marketing materials for your business: you can have the robots do it in a few minutes. If you get writer's block while coming up with ads, the robot will get you started. If you need a second look at a contract you just drafted, there are great tools that can help you research, draft, find missing clauses, and highlight terms that you might want to negotiate further.
Sounds great, right? Here are five categories of tools that you can put to use today to supercharge everything from law firm marketing to substantive legal work:
When OpenAI first made their AI text generation available to software developers, the market was flooded overnight with hundreds of new software startup companies that would allow you to generate short paragraphs, marketing taglines, and later on, full blog posts. Personally, I snapped up an early tool that had unlimited text generation credits, mostly as a toy rather than an actual tool. However, over time, it has gotten better and better, and when I get writer's block on a topic, I’ll pull it up once in a while and ask it to generate a blog introduction or outline. Sometimes the content is good, often it is bad, but usually, it will break the ice for me.
For years, there was rudimentary content spinning software that was largely ineffective. It would take someone else’s carefully crafted words into the machine, blend in synonyms, and produce nonsensical “plagiarism-free” output that was only good for spam blogs. Today, with the power of AI, these bots are absolutely incredible. If you have written something that is too long, you can ask the software to paraphrase and shorten it. You can also ask the software to rewrite it at the level of a fifth-grader.
Us lawyers tend to be long-winded and use lots of big words. Sometimes we do it as a means of demonstrating authority, and sometimes it’s just because we don’t know any better and use legal terminology so much that we forget how to communicate with the average Joe. Software like this can take complex paragraphs written on our level, and translate it down to a lower reading level using shorter words and shorter sentences, making it more approachable to the common person.
Suggested tool: Quillbot, hands-down. It has paraphrasing, simplification, grammar editing, AI co-writing, plagiarism checkers, and more.
For a marketing person, stock photos can be frustrating. You go through hundreds of pictures to illustrate your ads or blog posts, only to see someone else has purchased the same picture and is using it on competing ads, which ruins any distinctness your ads had in the first place. Plus, you are always trying to find “that perfect picture,” and you rarely succeed.
We are already at a place where I can ask a robot to generate the perfect picture. It does require some training when writing the requests — you can ask it to make the pictures photorealistic, you can ask it to make the pictures in the style of a certain artist or movie cinematography, and you may have to try a few times before you get an actually usable image, but it is a ton of fun along the way.
My first attempt was for a blog post illustration: It produced something…
I then described my favorite person and watched as it created a little mischievous girl with the same nose as my baby girl.
I even tried to describe myself fixing my car, and it produced a melted face with missing appendages, which was both hilarious and prescient, since two days later, I would shoot automotive coolant into my eyeballs at 16 psi.
Fortunately, unlike my A.I.-produced doppleganger, I didn’t lose any appendages or melt my face.
Even when it produces a bad image, you can usually get a laugh. And once you get the hang of writing requests to the bot, it really produces magic.
Because these images are newly generated and unique, you don’t have to worry about having the same image as anyone else, and you don’t have to pay for stock photos. There are even tools where you can upload a picture of a person and then have the software adjust that picture, such as uploading the picture of a lawyer and asking the software to put that person’s face on the body of a cowboy riding a horse — I haven’t quite gotten that working myself, but I’ve seen it done.
Of course, if you have a lawyer's brain, you are probably thinking of all the bad ways in which the software can be used. And there will probably be plenty of legal work suing people back and forth over art done “in the style of a living artist” or when someone puts my face on the body of a chimpanzee. How’s this for an idea? Use the image generators to make ads for image generation malfeasance cases.
Casetext introduced search features powered by artificial intelligence a few years ago, which not only crawls citations or keywords like we always have, but actually uses AI to dig through the facts, litigation history, and jurisdictional authorities of your actual matter to find more and better results, faster. It is searching for concepts and semantic meanings, rather than keywords or bluebook citations. That’s huge. A study done way back in 2018 showed 21% more relevant results, 24.5% faster research times, and much happier lawyers - one can only imagine how much better that AI model has gotten since. (It should also be noted that both Thomson Reuters and LexisNexis have added AI to their platforms since then.)
Suggested tools: pick your favorite legal research platform - they’re all doing it.
I was on the waitlist for Spellbook for months, and the wait was worth it. Rally produced this magnificent artificial intelligence powered Word add-in that can analyze your contracts and documents and find missing clauses, give you tips on which terms seem like they need some more negotiation (such as salary amounts that fall well outside the norm for your market and position), or just define terms in the contract that you neglected to define already.
When I first test drove this platform, it was okay for my legal niche (qualified domestic relations orders, the orders that split retirement accounts after a divorce), but they did produce some irrelevant suggestions for missing clauses. I mentioned that to one of Rally’s founders and within a couple of hours, he had trained the model with additional qualified domestic relations orders. When I tried it again, it had gotten smarter — smart enough to produce almost perfectly relevant suggestions.
Using generative AI, Spellbook gives you superpowers when drafting and reviewing contracts. Some of the many things it can do include:
See how Spellbook can be used to draft new clauses from a simple prompt in the video below:
Document automation is a pretty big deal for running a more efficient legal practice, but it isn’t a massive leap forward (conceptually) from where we were back in the 1990s. When I first started practicing law, we already had HotDocs and form automation. It has simply gotten better, and is now cloud-based, so we have the ability to access those forms from anywhere or have clients fill out forms for us. Document automation has incrementally improved across the board from what we had before, so while the day-to-day use is greatly improved, it doesn’t feel like a revolution.
This “artificial intelligence as a superpower for legal drafting” isn’t that sort of incremental improvement – it is a massive, massive leap forward. Sure, experts in their field may have already noticed a lot of these missing clauses or would have gotten around to defining the few extra terms, but for bringing laypeople and less experienced attorneys to that higher level of practice, the power of artificial intelligence to check your work, to find your mistakes, or to simply make your work product better, is already massive, and this is the first iteration of this technology.
We lawyers aren’t going anywhere: Tony Stark’s Iron Man suit is incredible, but it doesn’t fight bad guys on its own. Spellbook is one of those tools that takes an ordinary lawyer and lets them work faster, smarter, and even catches a few of their mistakes - legal superpowers, in other words.
Suggested app: Spellbook, by Rally.
When I first got started running paid search ads for lawyers and companies, running paid search was a huge bore. You had to come up with a few text ads, upload them to the ad network (usually Google Adwords), manually decide how much to bid for a click (on these networks, you usually pay per click on your ad), check in every few hours or days to change your bids, and try to figure out what is working and what isn't from a few clicks.
Today, you give the ad networks a dozen headlines, a handful of longer ad text sentences, some image assets, and once the platform has enough data, it can optimize itself using machine learning and artificial intelligence to get you leads for your law practice at a specific dollar amount — $50 per probate lead, $300 per auto accident lead, etc. It will decide on its own what the most efficient bids are, and can even make adjustments for demographics, geographic location, time of day, the device that a person is using, etc.
These machine-learning algorithms can make more adjustments than any human could ever make. Once you get the machine running with enough data, it leaves you free to check your ads every few days, or even less often, and you may not need as much help from a professional ad manager as you did ten years ago. Instead, the ad algorithms will make all of those decisions and adjustments automatically and eventually turn down or turn off your ads if it doesn’t feel it can meet your cost per lead or return on ad spend goals. It will test different ad copy, and different images, until it finds the stuff that works best.
Suggested app: Uh, Google has a monopoly here, right?
Ten years ago, if you were trying to start a law practice on a small budget, you would have too much on your plate: You not only have to learn how to be a lawyer, but you also have to run the marketing and business side of things to try to generate the leads that became clients. Either you would spend a lot of money on a marketing agency so that you had more time to lawyer, or you would settle for less legal business in order to spend more time doing business development.
Today, with the superpower of these artificial intelligence tools, you can put brilliant machines to work on everything from drafting your marketing copy to running the ads to generating the images that go next to the ads. Then, you can ask the robots to help you do real legal work. These superpowers can be used throughout your business, from writing ads to closing the case. And the best part is, we are just getting started in this nascent industry, and these models will only get smarter over time, which means you’ll have more help and more superpowers with each passing year.
If you are as excited about the potential for artificial intelligence in your law practice as I am, reach out to Rally and get on the waitlist to test out Spellbook. The team values your feedback and is developing the next generation of artificial intelligence tools for lawyers. We would love to hear about what features you are looking for in your drafting assistant.