James Pittman:
Ladies and gentlemen, welcome to the inaugural episode of Immigration Uncovered, the new DocketWise podcast where we dive deep into the dynamic world of immigration law, shedding light on the latest developments, cutting edge practice management strategies, and the transformative impact of legal technology. I'm James Pittman, co-founder at DocketWise, and I'm thrilled to be your host on this exciting journey as we empower immigration practitioners with invaluable insights and explore the intricate intersection of law and society. We'll delve into cutting-edge practice management strategies designed specifically for immigration lawyers. As legal professionals, we understand that managing a successful immigration practice requires more than just legal expertise. It demands efficient systems, streamlined processes, and effective utilization of technology. We'll explore innovative tools, software solutions, and automation techniques that can enhance your workflow, save you time, and optimize your overall efficiency. But we won't stop there. Throughout this podcast, we'll leave no stone unturned as we uncover the multifaceted realm of immigration law. We'll take a close look at recent developments, ensuring that you stay informed and up to date on the ever evolving landscape of immigration legislation, policies and case law. From new regulations to court decisions that shape the field, we'll provide you with the knowledge that you need to navigate and advocate effectively for your clients. In addition to the practical aspects, we'll guide you on how to effectively market your immigration practice. Building a strong professional brand and attracting the right clients are crucial elements of success. We'll share proven strategies and insights on effective marketing techniques from harnessing the power of social media to creating compelling content that resonates with your target audience. Unlock the secrets of successful marketing and grow your immigration practice to new heights. But our mission... even goes beyond the nuts and bolts of immigration law and practice. We're committed to exploring the deeper impact of immigration law and society as a whole. We'll bring you stories and experiences that shed light on the human side of immigration. We'll delve into the challenges, triumphs, and aspirations of individuals and communities affected by immigration policies. By understanding the broader social context, we can foster empathy, compassion, and a more comprehensive understanding of the issues. Furthermore, We'll discuss the transformative role of legal technology within the realm of immigration law practice. As the legal industry continues to evolve, we'll keep you at the forefront of technological advancements. From exploring the potentials of artificial intelligence and creating attorney work product, to innovative tools that revolutionize the way we serve our clients, will be your trusted source of information on the latest technological advancements tailored to the field of immigration law. So whether you are an immigration lawyer looking to expand your knowledge, a legal professional looking to stay ahead of industry trends, or simply someone passionate about the impact of immigration on society, Immigration Uncovered is the podcast for you. Get ready for an immersive journey filled with insights, inspiration, and practical advice. Together, we're gonna uncover the intricacies in this world of immigration law and pave the way to building successful and impactful immigration practices. Thank you for joining us on this exciting adventure. Let's dive in. And today on our first episode, I'm joined by two very special guests. I have Jared Jaskot, a practicing immigration attorney and sort of an expert on artificial intelligence and its uses in immigration law. I have with me Roman Zelachenko. He's founder of Laborless and also runs GMI Rocketing and he specializes in assisting lawyers with marketing their practices and building a successful practice, reaching the clients they want to reach. Today, we're gonna discuss some practical aspects of using chat GPT and other artificial intelligence tools in your immigration practice. Right now, I'm gonna turn it over to Jared to introduce himself.

Jared Jaskot:
Thanks, James. Love that intro. So as James said, I'm a practicing immigration attorney. My law firm, Jascot Law, is in Baltimore, Maryland. We focus on, broadly speaking, humanitarian immigration cases. So we have a lot of cases that are in immigration court. We also do have a lot of cases in front of CIS. And we've been using AI in our practice since. low, the end of 2015, I actually became first interested in AI for immigration law. for marketing. We were using it to intake customers. One of the things that, you know, what happens when you market on social media is that if you do have a really successful campaign, you have far more people come into your funnel than you have humans that are able to qualify them, at least in the size of law firm that I have. So I started thinking about AI and trying to figure out how we could use it to talk to people about their cases, get a little information. and make sure that they were a good fit for our law firm and our law firm for them. That was at the first start of my use of AI. We were primarily using what is known as now kind of old school AI, rules-based AI, if then logic. I was building chat bots that were talking to people then. It wasn't too long after that, that I started to see the deficiency of that model and I became interested in the field of natural language processing. Broadly speaking, what that is is... Someone can talk to you, give you a series of words, sentences, even a paragraph. You task the AI with understanding what they're saying and coming up with a coherent response. My bots have always been bilingual in English and Spanish, which adds an additional challenge. Early models in Spanish AI were very poor. So that was always a challenge as well. But as we've gone along now, we've had bots that have spoken to around 125,000 different immigrants. We now have really quite long conversations, obviously natural language processing has gone from what would almost be sort of the intelligence of a kindergartner to passing the law, the bar exam this year. And so it's been really quite incredible to get to see the evolution of AI and to take advantage of it in so many different parts of my practice.

James Pittman:
That's awesome, Jared. It's always great to hear your story. And I know you're doing really cutting edge things. Your Yo Tango bot is certainly a fantastic tool for immigration lawyers. Why don't you just, while we've got you on the mic, why don't you just tell us about the Yo Tango bot?

Jared Jaskot:
Yeah, so my bot really is about the initial conversations with clients. It talks to people on WhatsApp, Facebook Messenger, SMS, which is text messaging. It has a, we have a web app and Google maps. And so primarily lawyers who have a lot of leads coming in by social media will then put up the phone number of their bot on their bio or have a call to action. And there the bot can actually talk to people about their case, make sure. You know, most immigration lawyers, in my experience, are niched up. So, they only do certain types of immigration law. Very rare to come across a practitioner that does all types of immigration law. And so the bot can help qualify people, make sure that they're actually looking for the services that you provide. Can also provide value to that person. So how many people want to talk to law firms when they get off work at 2 a.m. if they're a dishwasher? And if a bot is having a conversation with someone on WhatsApp at 2 a.m., that's absolutely not a problem. Uh, and so what that's exactly what our bots do. They have that initial conversation with people working on a integration right now with open AI, the tools that we're going to be using today, that's obviously going to give our bot a lot more capacity than it has now. And we're also exploring, uh, the potential to have the bot begin to do limited legal services for people directly. Uh, which should really dramatically reduce the cost of legal services for immigrants.

James Pittman:
That's wonderful. Absolutely wonderful. Let's give him his introduction. Roman, please tell us about yourself.

Roman Zelichenko:
Yeah, thanks, James. And thanks, Jared. And it's really great to be here and excited for, you know, the launch of this podcast. It's exciting to be able to educate more of the immigration industry about up and coming and existing technologies. So appreciate your introduction of me. I also, in addition to what you said, practiced immigration law. And so sort of came at it first as a practitioner and through my experience practicing specifically employment-based immigration. I found that there were some inefficiencies in the work that was being done by me and really most of my colleagues and went on a circuitous journey to eventually launch my first company, Laborless, which is a tech company focused on H1B visa compliance. The way I got to building my second company, GMI Rocket, was that I had to market my first company myself through content creation, through writing blogs, through LinkedIn, content and engagement. And similarly realize that a lot of my colleagues, my immigration law and other folks in the industry, colleagues were struggling with that as well. And so launched a second business, GMI Rocket, to support individuals who want to get better at creating content for their immigration law firm and or company within the immigration or global mobility industry. For me, I run a tech company, but unlike what Jared built with YoTangoBot, my tech company is pretty rules-based. Laborless is a platform that automates an administrative process. And so, inherently there wasn't, at least in the beginning, too much space for AI in the tool. But what's been really cool, and I think hopefully where I can bring some value to this conversation is how I've been able to use various AI-based tools and platforms at the periphery. of my business to help my business run better. And I'm sure we'll get into this, but whether it's using a platform like ChatGPT to help me write initial drafts of content that I then put out on the web, or using AI-based tools like Fireflies to record meetings with clients and get a summary of that meeting, be able to search an automatic transcript of that meeting, even when a person has an accent. There are little things that you may not first kind of take, or you may take for granted at first, but then when you take a step back, you realize, wow, I'm so much more effective and efficient with a lot of these tools. So excited to dive into that, and hopefully together we'll paint a nice picture of how practitioners can use AI, whether directly with respect to client services or kind of at the periphery of just running their business.

James Pittman:
Absolutely. Let's start the discussion, and I'm going to start with a couple of simple examples that I worked up just using ChatGPT. These are some very straightforward generated templates, and I'll show you the queries. Let me just put them on the screen. Can you see that? Sorry, is my screen shared? No.

Jared Jaskot:
No.

James Pittman:
No, okay. Let's try that again. Oh, I see. Love that now. Yes.

Roman Zelichenko:
Don't

Jared Jaskot:
Yes.

Roman Zelichenko:
move.

Jared Jaskot:
Ooh, Infinity Screen. It's always exciting.

James Pittman:
Yeah, I'm seeing an infinity screen. Let's try that again

Roman Zelichenko:
It worked at first.

Jared Jaskot:
Yeah.

James Pittman:
It doesn't like when you just share the tab that you have to share the whole screen entire screen sometimes it likes

Jared Jaskot:
Top three.

Roman Zelichenko:
Thank you.

James Pittman:
So what I'm sharing right now are some simple examples of chat GPT generated templates of work product. Is that showing now?

Jared Jaskot:
Yep.

James Pittman:
Yep.

Jared Jaskot:
Oh. Nope.

James Pittman:
Still infinity.

Jared Jaskot:
Yep, click that tab.

James Pittman:
Click that.

Jared Jaskot:
That one, there it is.

James Pittman:
All right. So, Ricardo, we're going to start from this point. Just start from this point. So what I'm sharing right now are some examples of chat GPT generated templates. The first one was I simply asked the tool to draft an employer support letter for an I-129 petition for an H-1B visa, where the beneficiary is using qualifying experience to satisfy the requirement of a four-year academic degree. So you can see what it produced below the dashed line here is what it produced. It didn't fill in a specific USCIS address. That's not surprising, right? But you'll fill that in, dear sir or madam. And overall, I found it to have all the bases covered. I mean, I'll just read some highlights here. We're writing to express our full support for the I-129 petition filed by so-and-so. Actually, you can see this is now, this actually should be. the petitioner's name, it's put the beneficiary's name here, unfiled by, but that actually should be the petitioner's name on behalf of, and this should be the beneficiary's name. So that leads into what we're gonna talk about in that while it can produce a very solid first draft, you will eventually find some errors that you have to correct, and that is one of the reasons why it's critical that you review it carefully. So. filed on behalf, we believe that the beneficiary possesses the necessary qualification, skills, and experience to fulfill the requirements of the H1B category. And it goes on to sort of talk about hiring the best and the brightest at our company. We've reviewed the qualifications and credentials of the beneficiary. And then it mentions their experience, their performance while they've been in the employment of the petitioner. And here it's where it talks about the... using qualifying experience to substitute for the academic degree. Although the beneficiary does not hold a four-year academic degree, they do possess a substantial amount of qualifying experience that we believe meets or exceeds the requirements outlined by USCIS. And then it has some boilerplate language here about the company being committed to fully complying with all immigration laws. It has some sort of language that just adds polish, like we believe that the beneficiary makes significant contributions to the U.S. economy and enhance our organization's capabilities. We kindly request that you give full and fair consideration to their petition for an H-1B visa. So it does sound very much like the style that you would use when giving an employer a draft of a support letter that they can adapt. And then it has blocks here for the name, title, company name, et cetera, for the petitioner. So overall, I think very solid and only one obvious slip up there reversing the parties. Second one I did was even simpler. Please draft a file closing letter to an immigration client telling them that their marriage-based adjustment of status case was successfully concluded, resulting in a grant of conditional permanent residence to the foreign spouse. So here we have the address here of the firm, the client's address. The subject line it created was, successful conclusion of your marriage-based adjustment of status case. That's perfect. Okay, dear client, right? We're delighted to inform you that your marriage-based adjustment of status case has been successfully concluded, resulting in a grant of conditional permanent residence to the foreign spouse on behalf of our firm. We would like to extend our heartfelt congratulations. Okay, and then it summarizes what the work was that was done. We diligently prepared and submitted the necessary forms and evidence. Your case was adjudicated favorably, and your spouse was granted conditional permanent residence. And they talk about what goes along. I mean, this was not even part of my query, but it does have a paragraph in here talking about what the rights are associated with having conditional permanent residence status, the right to live and work in the United States for a period of two years. Within the 90 day period before the expiration of the conditional permanent residence, you must jointly file an I-751. It has that. Very important paragraph where you are notifying the client of the I-751 requirement that they will have to fulfill 90 days before their conditional card expires. Some nice language here, again, that adds polish. We understand that this journey's been an important and sometimes challenging one for you and your family. Our dedicated team has worked diligently to make sure every aspect of your case was meticulously handled. We encourage you to reach out to us. This is, again, nice language here. Once again, congratulations on this significant achievement. We're honored to have been part of your journey closing the letter. Very nice. I think, actually, a beautiful letter coming from a law firm that no one should be have really second thoughts about using, I think, once you've reviewed it. And then here I asked it to please draft an attorney-client agreement for an immigration lawyer located in the state of New York. that is compliant with the New York Rules of Professional Conduct and specifying that the scope of services is limited to representation before the EYR immigration court, specifying that the law. Now, I'm very detailed. When I give a query, the more detailed you are, and just separate your clauses with commas, but put in all the details into your query, that's when it's going to give you your best response. Try to really drill down and zero in on what you wanted to tell you, what you wanted to do for you. specifying that the law firm charges $300 per master calendar hearing appearance plus $5,000 to prepare applications for relief and conduct individual marriage hearing. So it produced an attorney-client agreement here in a form that seems very much like templates that you'll find if you network with colleagues. So it's a format that is this or something similar to it is used by many, if not most immigration lawyers. You... must outline the scope of representation, what you're going to be doing for the client, you're going to be providing representation in immigration court in New York. And then here it includes preparations of applications for relief, representation at hearings, representation at individual hearing, does not extend to other legal matters, goes on to specify what the responsibilities of the attorney are, what the responsibilities of the client are, what the fees are, it did mention exactly the fees that I specified in my query and what they were for. It has a paragraph here that it put in relating to costs and expense associated with the case, which are not included in the attorney's fee. And a paragraph here about complying with the requirement of confidentiality. Attorney's going to maintain all information shared by client in confidentiality in accordance with attorney-client privilege and the New York rules. Talks about termination of the engagement. Talks about the governing law and jurisdiction. And then has a signature line. for the attorney and you would put in a signature line there for the client. I think overall a very nice starting template that you can polish up and add to depending on your preferences and your firm's policies, but a very solid example. So I mean, again, I think the value proposition here of using the tool is clear. So let's turn to our guests to get some more examples. I either guess Jar, do you want to give some examples?

Jared Jaskot:
If I could, I'll just first comment a little bit about the examples

James Pittman:
Yep.

Jared Jaskot:
that you gave there. I thought both letters were really quite excellent too. One of the things that I think chat GPT, a lot of people aren't necessarily using it for yet, but that those letters show a good example is there's a lot of areas in firms where maybe it would be nice for us to do a little more in terms of client communications, documentation. potentially informational materials that we give clients. But you know, our time is limited and we're trying to make profit as well. And so when you now have a tool that can do such a nice job on these things so quickly, come up with templates, wow, maybe you just offer a much higher level of service and information to your clients for the same amount of money. I think that's a really, really great use of it there, James. And then on the retainer, I agree, it was a really good starting point, but I think The retainer also shows in a way the edges of chat GPT's skillset because, you know, like in my firm, we spent a really long time working on our retainer and thinking about like, what are all the problems that we're seeing in our firm on a regular basis? And so, and that's the sort of the human element from your firm, the local knowledge that chat GPT does not have, it will not have, but right, if. James had included those things in its prompt, like about, you know, you gotta get us documents within X amount of time and things of that sort. Of course, it could do it. So as James said, I think also, prompting with specificity is really what it's all about. One thing that I keep talking about when I talk about chat GPT, I think everyone maybe rolls their eyes. So it could either be that I'm missing something or maybe I'm extraordinarily weak in this area, or maybe no one's getting what I'm saying, I love ChatGPT's help for email. When you're emailing with other attorneys, when you're emailing with the courts, when you're emailing with clients that are occasionally angry with you, for me, it's a real challenge to sort of turn down that emotion. Take away the sort of like, especially if you're feeling attacked as a person receiving an email, it can be hard for me to then respond with an email that doesn't have a couple of attacks in there as well, especially because, you know, one of the reasons I became a lawyer is that I like to fight. but I don't really want that to be a part of my email experience. And so what I find myself doing time and again with Chad GPT is I give it a nice long prompt. You know, you're an immigration attorney. This is the nature of your relationship with this person. Your objective is to respond to their email point by point, but I want you to be professional, direct and aggressive. I mean, if necessary, but not provocative. It's just so good. You know, while I was on vacation last week, I did it one and then I was reading it to my father who was, he and I were sharing a room on our vacation and he was like, wow, that person, how are they even gonna be able to come back to what you just said? It was really, really impressive. And yet it lacked those little barbs that I tend to add in when I'm doing my email and struggling with it. And so in that way it was better as well. And then here's the real cherry on the top. Tough emails can take 30 to 45 minutes to write. You know, not here. I mean, we're talking about five to 10 minutes, which in the case of my vacation, literally gave me a half hour of my time back to enjoy the beach. But it also kind of kept my emotional levels down because sometimes as you're tapping these emails away, you're kind of riding the roller coaster of emotions. But putting it into Chad's GPT and getting that answer can be so much easier. In fact, sometimes when I'm not using chat GBT to write my emails, I do find myself thinking, what would chat GBT say here? And it's sort of, it's teaching me to do a better job with email. So I do, I find that for email. I think it's really. What are you using it for, Roman?

Roman Zelichenko:
Well, I kind of wish I had chat GBT back in law school when I probably sent a few emails that I shouldn't have when I maybe got rejected for a job of some sort. Yeah, I love the way you put that, you know, what would chat GBT say as sort of really the most unbiased and emotionally detached third party there could possibly be.

James Pittman:
Brilliant.

Roman Zelichenko:
So yeah, I love that. And I want to, you know, if I could just comment briefly. One thing that I would say about let's say the client retainer agreement you kind of touched on this Jared But a you can give it a little bit more detail so, you know when I get to my section and show kind of the few ways that I May use chat GPT there for my business or kind of Show my clients how they can use chat GPT for their business Some of the results were really good and other results were kind of missed the mark on a few things But did a really nice job on other things and you could follow up with a prompt that says hey take the above and then change this or add that or say it a little bit differently. And it will literally revise the thing that it just showed, that it just provided with whatever additional piece of nuance or whatever you wanted to. And so I would say this sounds obvious, but maybe for somebody who hasn't really used it yet or is hearing or seeing this for the first time, they might say, well, look, the output wasn't perfect, so I'm going to have to, you know, why would I even use this? Well, Just like with a person whose work you're editing, you can go back and say, this was good, good start. Now here are a couple of things that you can add. And knowing what things Chat DPT can continue to improve, and then at what point you say, all right, this is as far as Chat DPT can go. Let me download this into, or put it into my own Google Doc and put my own finishing touches on it. I think once people use it more, they get to understand the how to use it. apex of what it can do and then where they can step in. And when you find that right balance, you can truly optimize some of what you're saying, whether it's container agreements, emails. Like I imagine Jared, you wrote, let's say you had an email drafted and then you said, don't be provocative, but be, but then it came out a little bit too soft. Right. And you're like, okay, that was good. But can you maybe you know, hone in on this one and then it'll give you kind of another iteration of it and then you'll find what you like. I think again, my point here is iterating with chat GBT is something that people should also bear in mind, not just that sort of initial output.

Jared Jaskot:
I completely agree. I think going over and over on things is a really good way to do it. I found the editing process, like you mentioned, Roman, to be a place where I can do that too, where I'll ask it to write me a rough draft, then I'll edit and add things, then I'll put that in and ask it to edit again, and it can sort of be a passing the ball back and forth thing, and it is really good there. And I've also seen it, like, when I've used it to... create like a list of arguments for a conversation with a prosecutor or for a brief, I can then say to it, okay, on your third point that you made in this list of arguments, I want you to blow that up. Like, you know, make that

James Pittman:
Thank you.

Jared Jaskot:
third point into five more things and see like sort of how far it can go. And it's, it's really quite remarkable how good it is at sort of revising itself, thinking about itself and quickly and expanding and reducing areas. So yeah, I agree with you totally, Roman. Like keeping going with it, running pass after pass, I think that's definitely the best way to use it.

James Pittman:
Yeah, I've used it. I've used it a number of times where I've written something. I've asked it to polish what I've written. And then I've given it a query, taken its draft, revised its draft, given it back to it. I've asked it simply to elaborate on points, make something longer, make something shorter. It's good at condensing things too, without really losing the substance of what you're talking about. So it's pretty incredible. And you can be quite specific about what you want it to magnify or reduce. Yeah. Did you guys have any examples that you wanted to show?

Jared Jaskot:
You

Roman Zelichenko:
Trey, do you

Jared Jaskot:
want

Roman Zelichenko:
want

Jared Jaskot:
to

Roman Zelichenko:
to go first? Oh, yeah.

Jared Jaskot:
go, Roman? Sure. The example that I was going to show is actually this is an example of a plugin that I. So there's a couple of things in here. The first is that I'm using a prompt that's really popular right now amongst lawyers called Fallacy Finder. where you can go and you get an argument or something from someone else, and you ask it to go in and find all the logical fallacies in their argument. This is a really powerful tool for attorneys. And I think that it can be used two ways. The first way is your own arguments. For the folks watching this, I'm sure they're aware of it, probably less than they wanna be, that when they write things, there's logical fallacies. Why not just go in there and take them out, or at least know where your own weaknesses is while you're writing? The second thing is, you know, when you get an opinion from a judge or you get a response to a motion from opposing counsel, or you get an RFE or a notice of intent to deny from CIS, plug that baby in and watch this thing take it apart. That's really, really nice. It's so detailed. So here what I did actually, and this one that I think is, I think it's interesting, is for those that don't know, there's a character limit in the input side of Chad GPT. And that can be a problem with the law, because especially if you're asking it to look at legal decisions or memos, it can actually be, can be quite a limitation. So what I did here is I actually used the plugin, the Bing browsing plugin. And so what I told it to do was I said, I want you to act as fallacy finder. You'll be on the lookout for invalid arguments. It can call out any logical errors or inconsistencies. You know, and if you look at this Fallacy Finder prompt, there's a lot of them out there. And then what I told it to do was browse the internet and read the Supreme Court decision, Citizens United versus FEC. Because when I went with my first pass, what happened was it told me that the decision that I pumped in was too long, I went past the character limit. But then when I pointed it at the internet, it went to a page, read the decision, and then it was actually able to do it. Now digging into the decision, It actually, it first just gave a lot of summary. And that's the strong thing about this app. It's really good at summary. It gives kind of what the main arguments are for me to justice. And then it kind of pushed back on me. As a fallacy finder, it says, it's important to know that logical fallacies are not necessarily present in these decisions and kind of goes on and why. And I just went right back at it and said, no, please list the logical fallacies you found. And then it actually, it did. And it found two. And it actually shows a lot of intelligence here because it says, arguable next to the slippery slope and appeal to tradition. Two fallacies that Justice Thomas and Justice Roberts used. And it goes in and sort of talks about those fallacies. And so I think I love this example because, number one, is it's beginning to use this plugin. And this is a little hack around. It's an off-the-shelf prompt fallacy finder that I like a lot. And then it also shows how you can go back and rework chat GPT if it doesn't actually give you what you want out of the gate.

James Pittman:
So the Fallacy Finder is a freestanding app that only serves that one function.

Jared Jaskot:
No, it's a prompt.

James Pittman:
It's just a prompt.

Jared Jaskot:
So what we're seeing, I think one of the things that is really emerging, and I do encourage people if they want to go read the Brainiacs email, I get that. It's from a law professor who's done a really nice job of coming up with sort of like email. He did 100 emails in 100 days of different things. But one of the things that's emerging is that lawyers were beginning to acquire all these prompts. And these prompts are very powerful. So you can sort of, you have your own little database of prompts and then thinking about how you want the app, maybe you go over to your Google doc where you have all these prompts and you say, oh, I need that, I need this specific prompt, or maybe you combine them. And what turns out that having all these prompts in your arsenal and understanding which ones to deploy at which time can really help the tool go a lot further.

James Pittman:
That's actually fascinating. That's sort of like a whole new dimension to things where we have to, in order to make the most out of it, we really have to arm ourselves. We sort of have to have a library of prompts at the ready to really focus in on the specific things that we wanna highlight. That's amazing.

Jared Jaskot:
There is one more tool that I'm not allowed to say the name of yet. It's going to be dropped at ALA when you guys are there on Wednesday.

James Pittman:
Okay.

Jared Jaskot:
And what it is sort of it's chat GPT. Well, it's an open AI's API. It's actually going to have, uh, access to a lot of proprietary immigration books preloaded.

James Pittman:
Mm-hmm.

Jared Jaskot:
And so, you know, one of the problems with chat GPT is that a, it makes things up B it doesn't cite its sources. what we're beginning to see in our market are gonna be stuff that have proprietary content and will actually have footnotes and sort of list. So that's gonna be coming out this week. You guys will see it, it's exciting. I've got to play with it some, it's really good. But I think when you start to combine this proprietary content, your prompt library, something that's got footnotes, less hallucinations, probably less creativity, but we're really starting to see, emerge a whole new set of tools that are gonna be coming out. really right now.

James Pittman:
Yeah, that's incredible because then the tool has a very, sort of let's say detailed and in-depth education in a specific field. So it's not just surfing the net up to 2021 to find information, it's got a library of very detailed subject matter and the results that it's gonna produce are gonna be so much more concise, so much more detailed, so much more precise, that the quality is just gonna be incredible.

Jared Jaskot:
Yeah, yeah, it's exciting.

James Pittman:
Roman, you want to talk about some of the marketing aspects? Sure.

Roman Zelichenko:
Certainly, and I will share my screen and show some of the. Let's see here. All right. So can you all see my screen

James Pittman:
So,

Roman Zelichenko:
here?

James Pittman:
yes.

Roman Zelichenko:
All right, so I'm gonna go through a couple of different things. The first one I'm gonna do is I'm gonna talk, so I was talking about marketing, right? And I work with immigration law firms and others on marketing, specifically marketing on LinkedIn, but really it could be on any sort of platform. And I think one of the big challenges For a lot of folks in the industry really anybody who's trying to grow their brand through LinkedIn is you know content creation and Feeling like they can be consistent and consistently write good content And so you know one of the things I've really been sharing of course with people who? Want to write a lot of content on their own is the ability is how chat GBT can help them at least create a nice first draft of a LinkedIn post so I, the screen that I have pulled up is, you know, it's a chat GPT, uh, the screen. And I have a prompt here up top that says it's telling chat GPT, right. A 200 word LinkedIn posts targeting immigration lawyers about the importance of using LinkedIn for growing their law firm brand. And I said, touch on subjects such as how many potential clients there are on LinkedIn, the importance of having a digital brand in today's day and age. And that LinkedIn is actually free to use. And then I gave it, so that's sort of the content of it. And then I gave it kind of specific formatting prompts. I wrote, use emojis as bullet points, add an empty space between paragraphs and select the top three immigration related hashtags in the end. And it's a fairly basic prompt with some specifics that anybody who wants to get better on social media can learn about on the internet in terms of using emojis and kind of adding some space. And in about 30, 10, 15 seconds, it pushed out a pretty good initial draft of a post. And for folks who are listening, they may not be able to see it, but what we're looking at is something that looks like a pretty well-crafted LinkedIn post in terms of use of emojis, in terms of the language being used. It shows, it provides some facts, but it's still pretty conversational in tone. There's enough space there that it doesn't feel like one long paragraph. That's difficult to read It does end with some hashtags at the end Etc. Would I put what I copy and paste this onto LinkedIn? No, but this is an amazing way to go from 0 to 1 to use that term And and then kind of play around with it. For example, I don't love that it has too many for me It has too many hash. Sorry. It has too many Emojis, right? I just it's not my style Some people say it's not my style. Well, that's okay. Like we said earlier, you can use this as a draft. You can then provide a secondary prompt to kind of revisit the post and tweak it or just kind of use this and then plug in your style manually and kind of finalize that. So I think for immigration lawyers who want to create content around their kind of area of expertise, the reality is what you're gonna put on social media is going to be not necessarily generic. right, but it's not going to be specific to a client level. Maybe it's specific to a client type, right? Maybe it's specific to, if you're an employment immigration attorney and you work with, I don't know, oil and gas companies, you can write content about oil and gas and have it be something that sounds specific enough for the industry, but is general enough for different companies within the industry find value out of it. So to me, this is a really great way to use LinkedIn for marketing. So there's that. I also want to show an example here of when you're talking about LinkedIn, for example, a lot of people also struggle with their LinkedIn profiles. So not even just content, but the basics of their profile. And you can add a picture, you can put your experience there where you currently work, where you previously worked. One of the tougher parts for folks is the quote unquote about section, sort of their bio, where they're expected to... you know, write a little bit about themselves. What a lot of attorneys do, specifically attorneys, is they, especially if they're at a larger firm or maybe previously were at a larger firm, they have this sort of really formal sounding third person kind of, you know, couple paragraphs that again, talks about it would say Roman Zelichenko graduated, summa cum laude, I didn't, from Brooklyn Law School in the year 2013. He received this award and that award and this fellowship and that fellowship. Um, you know, Roman has represented, uh, companies in XYZ ABC industry and whatever. Right. And typically it's stuff that you would see in a CV and a law firm website. Nothing wrong with that. But what I typically advise immigration law firm clients are frequently, frankly, anybody, uh, is to be a little bit more kind of personal, right? Imagine your LinkedIn profile. is the same as you introducing yourself in person, in real life to somebody. You're not gonna talk about yourself, first of all, in the third person. And second of all, you're not gonna jump into your sort of accolades and accomplishments. You're gonna say, here's who

James Pittman:
Thank you.

Roman Zelichenko:
I am, here's why I do what I do and what I love about it. Maybe you can say one thing that I was successful in, that I really am proud of, was this particular case or what have you. But there's a way to... present yourself in person that I think isn't translated really well on LinkedIn, people get stuck there. So what I did with the second prompt here was I gave ChatGVT a prompt and I wrote, I'm an immigration law firm focused on employment-based immigration, specifically financial and FinTech industry. And that's a little bit of background about me. And then I said, write a LinkedIn about section explaining that I went to Brooklyn Law School focused on immigration law, yada, yada. Um that i'm excited and love immigration law and then I wrote end the bio with a call to action to reach out To me on linkedin or visit my website. I made up a website url And then I gave similar prompts that I mentioned before I said use emojis as bullet points and use an informal But still professional tone of voice So if you scroll down again, if you're watching this, you'll see this if you're listening Um, really what it's come up with is a really nicely structured linkedin about section It talks about myself. It's it's first

James Pittman:
It's good

Roman Zelichenko:
it

James Pittman:
fun.

Roman Zelichenko:
goes in the first person It is professional but formal enough where it sounds like it's me talking It's not sort of you know, so a marketing department trying to write something really impressive about an attorney Etc. It uses emojis as I asked there's a call to action at the end It's not perfect it added hashtags which I didn't ask for And then at the bottom, it wrote, this LinkedIn about section has an informal, but still professional tone of voice. JAT GPT misunderstood. I was telling it to use an informal, but professional tone of voice. And it thought that I wanted it to write that this profile has been

Jared Jaskot:
Note

Roman Zelichenko:
written

Jared Jaskot:
that.

Roman Zelichenko:
in an informal, but professional tone of voice. So, but that's okay. Right? Like it's a robot and it didn't understand the sort of, maybe I used my sentence structure, wasn't perfectly logical for it. So I would just, you know, this is a really, really great starting point. I would just prompt it and say, hey, take out that last note and get rid of the hashtags. Otherwise everything else is perfect. So, yeah, go ahead.

James Pittman:
I'm sorry, I just wanted to throw in a couple of comments. Perhaps you told it to note that, you know, this piece should have a professional but somewhat informal tone and it, you know, it thought that you meant for it to write that as a note. Also, I found that the fact that it can use the emojis to me is fascinating and pick out appropriate emojis. I mean, there's sort of a buzz going around. People are talking about the fact that chat GPT can explain why a joke is funny. This reminds me of the same thing. If it has an understanding of which emotional emoji makes sense in the context of the point that you're trying to make, to me, it's fascinating. I mean, I would love to understand more deeply how it does that, how it knows that, but I think it's amazing that it can do that.

Jared Jaskot:
Right,

Roman Zelichenko:
Again,

Jared Jaskot:
plus,

Roman Zelichenko:
look.

Jared Jaskot:
like, I mean, this is far better at emoji use than most people over the age of 40.

Roman Zelichenko:
Right.

James Pittman:
That's true. Good point. That's right.

Roman Zelichenko:
And again, look, to me, I'm a writer. I wrote for magazine for a decade. You know, now I write content really as kind of my profession. And I think, frankly, you know, content writing is one of the first things that people say is going to go. And I think still nothing is as good as a human being, at least for now. Really, this isn't close. But it's good. It's really good. And I think to me, the most important thing is that this is a really, really wonderful place to start. The hardest part about being a writer, is when you have your topic and you crack your knuckles, you open up your computer, or maybe you're writing on a pad and you're staring at a blank page. That is the hardest part of writing. Editing, I've found at least, and I think many of my writing colleagues would agree, is far easier. It's easier to criticize what's been there already rather than to create something out of nothing. So that's one of the reasons I really love ChatGPT. Now the third thing I'll show, so we did a LinkedIn post, we did a LinkedIn bio. The third thing that I would show would be a blog. And so this prompt is quite long. And so I'm not going to read the whole thing out loud. But I effectively want to chat. By the way, I have the free version of ChatGPT. So Jared, you can speak about the upgraded version. But I'm just showing the free version, because when people are trying new tech for the first time, if they could try it out for free, it's great. LinkedIn is free, and the free version of it is really powerful. Same thing I would argue with ChatGPT. So for example, here for a blog post, I said, I told it, I'm an immigration law firm looking to write a 500 word SEO optimized blog post about the labor condition application process. And I said, I want you to write the first section, should explain what the LCA is. The second section should talk about the LCA posting requirements. The third section should talk about how traditionally employers posted using paper, but now there's tech. And then I said, please refer to this particular. DOL bulletin and conclude with a call to action for a reader to use laborless my software and link to my company. So chat GPT put together a pretty decent draft of you know a fairly basic blog post. But here's what it did. It gave me a title and then it wrote introduction and then it gave an introduction. The next paragraph said section one. It wrote out the word section one and you know it gave me some information and then it said section two. wrote section three, and then it gave a call to action at the bottom there saying, please use Laborless to automate this process. I realized that I told it that I'm an immigration lawyer, and then I had a call to action at the end for a tech company, it doesn't make any sense. And so I told it, I wrote, I gave it a second prompt and said, hey, rewrite, I didn't say hey. In my mind I said, hey. I wrote, rewrite the previous response, but instead of the introductory paragraph coming from an immigration law firm, it is coming from Laborless as a company. Makes sense, right? I made a mistake in the beginning. And it did exactly that. It rewrote the introduction, talking about Laborless, and then the rest of it was effectively the same. And then I was reading it and I said, you know, I don't really like, I can't copy paste this and post it right away, right? I don't like that it says section one, it literally says section two. You don't wanna read that when you're reading a blog. It's awkward. So then I gave it a third prompt and said, take out section one, section two, et cetera, and instead draft it with section headers and space it out as this SEO optimized blog post. And that's exactly what it did. It gave me a title and then it took out the word section one, section two and just wrote out the title of the section, which I would put into an H1 or an H2 format on my company blog below that. So what started as an error, first in terms of content and then formatting, I was able to work through with pretty clear and pretty simple prompts thereafter, and got what's a really good workable blog post that I can now feel comfortable copying and pasting into my blog on my website, making some final changes and going from there. I want to show these three options because we went from a LinkedIn post to a LinkedIn bio to a sort of- blog post for a website, there are different types of content, and it did a really good job with all of them as a starting point. And the thing I'll end on is to say that the pushback I typically get from immigration lawyers, or frankly anybody who's a professional trying to grow their business on LinkedIn, is, well, yeah, I'm not a content creator. I'm an accountant. I'm a lawyer. I'm this. I'm that. I'm not good at these things, or I don't have the time, or whatever. This is a great point. way to start to feel comfortable with not having to draft, you know, sort of the initial thing. And then going from there, you can become a content machine without being a content professional, so to speak.

James Pittman:
Absolutely, fantastic use of the prompts there and very impressive, both very impressive pieces. I mean, well, I would be remiss if we didn't talk a little bit more about some of the pitfalls. And I have some points that I laid out for some of the pitfalls. Let's talk about that. Let me put that up on the screen. Let me share that. Here we are. And I'm talking now about the potential pitfalls. So the accuracy, I mean, we've hit on these points, but just to sort of sum up and conceptualize what the pitfalls are, the accuracy and the reliability. I mean, sometimes you'll get text from ChatGPT or other tools that appears coherent and accurate, but actually there are errors or inaccuracies in it. We've said this. So you cannot rely solely on machine-generated content without careful review and verification. Otherwise, you're risking potential legal and factual errors in your work product. You have to keep in mind your ethical and professional responsibility. For attorneys to have an ethical obligation to provide competent and diligent representation to their clients. So when you're using AI-generated content, you have to ensure its accuracy. Otherwise, you're potentially breaching these ethical obligations. Third, you have to be cognizant of confidentiality and security. All of these language models operate by processing large amounts, huge amounts, vast amounts of data, including potentially sensitive or confidential information. So you want to be very careful. If you're sharing that kind of sensitive information with an AI system, that raises real concerns about data privacy and security. Attorneys have to be careful about protecting client information and ensure that they comply with relevant privacy regulations. Also, as it is a robot, there will be a lack of legal judgment and contextual understanding. And we've mentioned a couple examples of this during today's discussion, but language models aren't people. They lack nuanced understanding of legal principles and the sort of moral theories that those legal principles are based on, lack understanding of legal precedence, and lack the ability to apply human legal judgment. which attorneys would possess. So they may not adequately consider specific facts or weight facts in a way that a human being might. Also the legal context or strategic considerations when you're talking about undertaking any given legal process in the legal system, it will not understand those strategic considerations and cannot weigh options of one course of action versus another course of action. And as that relates to a particular case or legal matter. And last, accountability and liability. When you're using AI-generated work product, questions can arise regarding accountability and liability for any errors, emissions, or legal consequences. So you have to clarify the roles and responsibilities between the attorney and the AI system in order to ensure proper accountability. And as an attorney, of course, you are ultimately responsible for everything that goes out of your office, as all of you know. But those are some pitfalls, things to keep in mind. We're going to see how that evolves as the technology gets better and better. This has been a fantastic in-depth discussion of the practical applications of some of these tools today. I want to thank our two guests, Jared Jaskot, Jaskot Law, Roman Zalichenko, founder of Laborless and also of GMI Rocket, a marketing agency catering to immigration attorneys. Before we close, do either of our guests have anything that you want to add?

Jared Jaskot:
I would like to say two things. One, thanks for having me, James. It's always really fun to be with you and to hear your insights. I've been a docket wise customer for a long time and I continue to be a happy customer. Second thing is... This is like any other thing in this world, the best way to get better at it is to use it, use it, use it more. There's absolutely zero reason to be intimidated by it and stay on the sidelines, right? Even if you're intimidated by it, don't stay on the sidelines, just start talking to it. And it's not like it's in public or whatever. Your first efforts can be halting, they can be messed up, and guess what, you just keep on going. That's the best way to learn from it and see what its capabilities are. So I think every lawyer that isn't about to retire in the next year needs to learn how to use this the sooner, the better. It's going to be a competitive advantage for law firms that adopt faster, really frankly, a very large competitive advantage. So start using it right now and get the paid version, start paying for it. The paid version passed the bar exam. The free version failed the bar exam. That's enough said on that. So thanks again, guys.

James Pittman:
Thank you, Jared. Roman anything?

Roman Zelichenko:
Yeah, I can't add anything onto that, but I will say, James, thank you. It's always a pleasure talking to you and Jared. And I don't practice anymore. I would be a Dakiwise user if I was, but I'm not. But I have been a big fan and friend of Dakiwise for a long time, so excited to see where things continue to go. And again, thank you for having me.

James Pittman:
Thanks very much and join us again for our next episode of Immigration Uncovered. Bye bye now.

Jared Jaskot:
James laughs He's just like click That's really weird

Roman Zelichenko:
Bye bye now. See ya. Oh, that was great.

Jared Jaskot:
Thanks again Ricardo and Tom, anything else

Roman Zelichenko:
Yeah,

Jared Jaskot:
from us?

Roman Zelichenko:
yeah, thank you guys.

Jared Jaskot:
Sorry. Totally agree. Use it

Roman Zelichenko:
Yep.

Jared Jaskot:
however you like.

Roman Zelichenko:
Same here. All good. Thank you.

Jared Jaskot:
Let us know so we can promote. When do you think it's gonna go out?

Roman Zelichenko:
I

Jared Jaskot:
For

Roman Zelichenko:
think

Jared Jaskot:
me,

Roman Zelichenko:
James...

Jared Jaskot:
I'll just say, just put jascott.law. That's fine for me.

Roman Zelichenko:
Yeah, I think, and I think James has some whatever I've sent before. It's Roman Zolchek. I mean, yeah, you could put laborless.io and gmirocket.com or something. I'm okay.

Jared Jaskot:
Yeah, I think both of the words are so small that I just.

Roman Zelichenko:
Yeah. Managing part. I'm a vice president. Yeah.

Jared Jaskot:
and

Roman Zelichenko:
Yeah,

Jared Jaskot:
assisted.

Roman Zelichenko:
exactly. You could just, yeah,

Jared Jaskot:
I'm out.

Roman Zelichenko:
exactly. Yeah. I'm good with just the name and my name and the company names

Jared Jaskot:
screen.

Roman Zelichenko:
or company URLs, I guess. But yeah. Cool. Thanks y'all. See ya.

4o

‍