[Music] problem in this regard is that I see a lot of a lot of challenges when it comes to my name is Ahmed Elnaggar and this is Monday legal. Good morning and welcome back to Monday legal. My guest for today is Khaled Chivji. He is the CEO and founder of sale and of course the topic of the day is legal technology and AID is a technology specialist lawyer. Thank you very much for coming and for dedicating the time to be with us. Pleasure. Thank you for inviting me. Thank you very much Khaled. I would like to start with giving you a few minutes to explain your legal background and legal career. So our audience will learn a little bit about what brought you into the technology and what brought you into AI. Why did you choose that field? Okay. Well, I um I grew up um actually wanting to become a doctor. I was absolutely fascinated by biology. I'm still fascinated with biology. Um, and I was lucky enough to have received an offer to go to medical school, but I just fell short on the grades um, as I was leaving college. And so, a very major engineering university called Brunell University in in the UK offered me a course on a brand new course called Internet Engineering. Okay. And it was a really fascinating course at the time of the dotcom bubble seeing how emerging technology like e-commerce and open source software were just becoming so readily available and as a result I was able to win a place to work at British Telecom at their contracts division and I was really really into the idea of negotiating doing deals such that a very good solicitor that I worked with called Sabanaki said to me you you really should think about becoming a solicitor, a lawyer as we say in England and Wales because it will give you the professional accreditation and the standing to go into meetings and negotiate with the big hitters, you know, the other solicitors because you have met the same professional standard that they have and so they will no longer have one up on you. And so through that I went to night school for 4 and a half years whilst working full-time and thankfully I qualified and then made my move to the UAE where I've worked as a lawyer for the last 13 years um primarily for technology and telco firms and most recently at Microsoft and of course in today's age Microsoft is one of the leaders in AI and given my technology background I'm also fascinated with it from a technology ology perspective and a regulatory perspective and a commercial perspective and so AI and tech law is my forte and it probably will be for the next 10 years I suppose. Yeah, I totally agree with you. I mean this topic is one of the hottest topics at the moment is the use of technology in other professions out of uh so it's not just something that all the technology companies are working with. I think I see now accounting firms, law firms and even restaurants and people working in marketing and branding and everybody is really so much into technology and since you have been practicing law in a technology- based company I think that fits very well with your passion as well because as I see and I know that you're very passionate about the technology and the use of artificial intelligence and that probably take us to sale and I would like to learn more about your company and what it is is doing in general. Well, thank you. So, Sale S A I L stands for solutions and AI for lawyers. And the reason that it the acronym has a nautical theme is that earlier this year with one of my business partners, we did some market testing um on the brand and the company's model with a number of very senior general counsel looking after major firms around the world. And the most surprising meeting I had was meeting the general counsel for a very very large multinational shipping firm. So, prior to the meeting, my business partner and I thought, we're going to go into this meeting and there's going to be nothing extraordinary to learn about this company. They're probably going to be very traditional. And they turned out to be the most progressive and uh advanced customer that we spoke to about AI. Their general council had already invested in buying solutions geared on AI to test within their legal department. They had hired a lawyer to lead that um part of that change. And being a multinational shipping company, they were already deploying AI to you know forecast risks around the world. So after that meeting I came out completely um you know impressed with that such that I had to name the company after something that was nautically themed. So I called it sale um and use the name solutions and AI for lawyers because AI is AI it's a technology but what do we do? Well we provide solutions If you don't provide solutions using AI, you're creating more headaches for lawyers. And so we what we want to be are problem solvers. We don't want to be people who are creating more problems than headaches for lawyers to use technology and to learn how to use technology as well. And that's why the focus of our company is to empower legal departments and executives in the seauite to build their own AI adoption strategy. And what we do is we do that using our own methodology which helps them to get from where they are now to where they need to be within the space of 12 to 18 weeks and generate an ROI by that time. And that's very very fast compared to the market standard today of a year. And the way that we do it is by empowering legal departments to lead the AI adoption strategy rather than technology departments because technology departments won't consider naturally the compliance risks of using AI. They won't consider the contractual or regulatory risks of using AI but they do provide services to build proof of concepts but at that stage they have to get sign off from legal in in most organizations to proceed with a live uh environment demo and that's when legal needs to be brought in traditionally and they've not been involved at that stage and so there's a delay they have to upskill and learn how to use AI they have to learn about the PC the proof of concepts and it takes time so our methodology Y is is empowering about empowering lawyer to lead that change and we think that investing that time for them to understand how to lead that change and how to actually embred uh embed AI in their organizations provides that extra time to really really get their hands and get crisper on what AI could do for their organizations and how to protect the organization from potential risks. Yeah. Yeah, and the time invested helps to save organizations an immense amount of time in in the long term as well. I think big part of the conversation, I totally agree with you and everything you say and this is really something that will empower the industry in general. A big part of the conversation is going towards how AI can help us. Um I I feel that there is a little bit of a missing point here where we don't focus much on how can we integrate it easier into the skill set of the legal professionals because there's a lot of legal professionals who are amazing or whatever they do and they are also enthusiastic into using the AI but it's not easy to understand that now they can review the contract in a different way than they used to review the contract. can find the emails in a better way or navigate the documents in using the AI in a better way. How can these um solutions happen? What what sort of help can you do in this regard? Well, how long do we have today for today? Go ahead. Well, before I go ahead, the first thing I want to do is to recognize and acknowledge all of the work that legal technology vendors have done to build solutions and technology platforms for law firms and in-house legal departments. I can safely say that without the effort and the drive and the investment they've made into the legal profession, I wouldn't be where I am today. They've created that market. They've trained lawyers and skilled them up on how to use AI and legal tech and they deserve recognition for that as well. My methodology is built around using an organization's technology stack. what they've already bought in terms of licenses, for example, to use Microsoft 365 C-pilot or for example to use Gemini for Google Workspaces and to invest time in maximizing the opportunities to generate outputs using those platforms rather than merely buying another solution or another platform from the market. And the way that we do that is by first of all encouraging lawyers to get over their fear of technology. And again I want to do another acknowledgement here that becoming a lawyer is very very difficult and people who've trained and invested time and money to go to law school had given up and made sacrifices elsewhere in their lives. Of course that may have been um to sacrifice the opportunity to use technology and to become better at using technology along the way whilst they were studying to become lawyers. And so I'm very cognizant of the fact that lawyers haven't had the chance to expose themselves to technology as much as perhaps other disciplines have in finance or for example engineering or or other disciplines as well. Um and so what we do is we help lawyers to get over their fear of AI and technology and we start with prompting. We start with teaching them how to prompt using a generative AI tool for example like chat GBT or clawed or complexity. And we use um a combination of nonconfidential data and prompts to enable lawyers to practice how to extract information from documents, how to then present information which they managed to extract from documents and then how to develop prompts to help them to get better at doing that going forward. Now, that in itself provides a lot of skills-based training on the way that lawyers prompt and then hopefully with the right tools in their enterprise, they'll have the ability to use those prompts. And so, here's a health warning for everyone um who's listening and watching today that never ever put client documents into a public premium yes, free-touse AI tool. Um it is so so risky to do that um because any data that goes in there and any prompts that go in there will be used to train the foundational model and you really don't want someone halfway around the world who's asked for an M&A template um to be spitting out the names of the parties that you yourself have been working on advising um simply because that was a template that CH GPT was trained on at that moment in time. The best way to do this however is to use the generative AI tools that your enterprise has purchased for use. They sit on your cloud and as a lot of the hyperscalers and AI studios say that data remains your data. It is not used to train foundational LLMs that power that generative AI agent or platform as well. Right. I think your your your professional help is not just um important to make the lawyers know how to use AI in a better way to serve their work. I think it is also important for not to use AI in the wrong way to leak data or to make mistakes that can cost them the license for example or the cost them the confidentiality of the information of their clients. Absolutely. Absolutely. I mean we so this strays onto the topic of what's called safety AI safety and the broader topic of responsible AI right and the two disciplines merge when it comes to client attorney privilege and how to protect that when you're using artificial intelligence to generate outputs of work for clients and the basic rule that we've discussed today is just don't use a premium tool to do that but do experiment using dummy data, synthetic data. You can ask um CHBT to generate synthetic contracts for you or synthetic reports that are based on fictitious data and to then generate legally useful prompts um that you can use to test in chat GPT if you like to wish to wish the the way you want to go to extract information and to pretend if it was a real contract and how to get good at analyzing it using AI. And part of what we're doing is what you are doing is also to recommend because maybe these firms or these corporate councils have no idea about they want to start you know they don't want they don't know where to go to from that point it's like we want to use AI but they don't know whether they use chatbt or any other legal tech model that is already in the market. So you probably are able to direct them based on their requirements. We do that and and we do that a lot and the reason for that is that generative AI platforms are powered using something called a large language model. Some of them have what's called a small language model. Um but broadly speaking they're large language models. They've been trained on billions of what are called tokens of data and this is called training data. And a lot of these LLMs have embedded in them what are called parameters and weights. And these are the ways that the LLM can analyze data that it's been trained on. And this is how it gets good at doing specific tasks or um performing different functions. I'll explain. So for example, some generative AI engines that are powered by chat GBT are very good at analytical reasoning. So if it's powered by chat GPT 4.5 or chat GPT40, it'll be very good at providing analytical reading reasoning and what's called chain of thought reasoning, showing you how it's working out uh a problem and then helping to guide you in terms of the solution that you may need. There are other generative AI platforms that are very good at financial models and analyzing those as well as other models which are very good at analyzing contractual information because they've been trained on contractual data. And so what we do is we first of all analyze um from what the client wants in terms of the use cases in terms of how they want to use AI. And using our industry knowledge, we pretty much have a good idea already about what they're likely to want to use AI for. And we then seek to match them up with the right model that they can use internally on their cloud um so that they can benefit from the use cases that offers. Do you see that the cost of such subscriptions or such models is something that would be a wall between the legal professions and the GCs into um using the AI because training is a cost. The subscription model of whatever solution is also another cost. The time that the lawyers will waste not waste sorry will invest into because billable hours my friends. So the time that the lawyers are not going to be billing they would be spending in in in these things. So do you think this would be a little bit of a obstacle on the in the way for professionals to develop? I do. I do and I think that the reason for that is perception. Lawyers who've been trained to bill by the hour have been trained on how to maximize those hours in order to get as much client work done such that they can build for their time. It's the way that the legal model has worked for hundreds of years. Yeah. And that is the way that they've been trained and that knowledge had been passed down to them by very senior lawyers who have trained them. What we try to do though is with AI is we try to show lawyers how on an individual productivity basis AI can help them in their day-to-day work. We show them use cases and we demonstrate use cases. for example, when they arrive in the office, how AI can perform automatic summaries of cases that have been issued by the courts the day before or provide a summary of the latest news or legal news for them without them having to research legal databases or the uh news outlets to get that for them. And we then show them how it can be used to then accelerate client work um going forward such that they can actually get more work done in the hours and be more efficient such that they can start talking to their clients about more strategic work which actually comes with it a much higher value to their clients and remember that lawyers are all about not just uh providing legal advice but also providing solutions and their clients expect them to provide solutions. So the way we could do that is by helping them to move to um matters which are more solutionsorientated. There are and it's worth mentioning that there are state bar associations like the state bar of California that have recognized this paradox. They've recognized the paradox that look lawyers are expected to use AI. Their clients are expecting them to use AI and they're expecting them to use AI because they're charging $2 to $3,000 an hour to produce legal work and their clients want to save money. So the California State Bar said to its members, okay, if you use AI on a client matter, you can bill your clients for the cost of the time that you spent learning how to prompt and prompting to get solutions out of the AI engines that you used. And so it enables the lawyers in California to enter enter into some sort of cost recovery for the time they spend training using AI or using AI to maximize the results on a client matter or maximize the output advice on a client matter. So I think that that's a very very good balance to provide in this day and age in terms of what lawyers could do using AI. uh it certainly doesn't speak for what might be the case in say 5 years time but it does mean that lawyers have a fighting chance of being able to argue that the billable hour model for them should persist there is this term that from time to time I hear it and I think I heard a few things about it as well the term that comes is um AI agent and uh I learned or what I understand about it is that you can create an AI agent to do a specific task on your behalf or to perform a set of tasks in sequence or on occasional basis. Can you please explain more of that because we need to understand what is an AI agent? Okay. So, generative AI platforms today in 2025 are very manually driven. They're manually driven because at this moment in time, the only way that we can gain responses from those prompts uh or the generative AI agents is to type or speak to them. Yeah. Um using natural language and they will then spit out the response for you. But people are starting to become very wise about using these generative AI engines such that they might want to then ask the generative AI agent to automate the next steps for them. So I'll give you an example. I have on my cloud a generative AI agent uh platform called Manis. Uh it's got mus and it runs based uh it runs on the deepseek model. There's an open source model and Manis is very good at chewing through large amounts of data and providing really excellent outputs. But I'm a lawyer. I want to know about what's happening in the news every day. And so I've set up Manis to automate the retrieval of the latest news stories concerning AI every single day at 8:45 p.m. such that it then produces a report that I can read on my phone. And then I can analyze it and then it will then provide me with the option to then go deeper into some of those reports that it provides because it's already gone and learned what I want it to do. It learned about the prompt um that I've inputed it in terms of information that I I want it to retrieve. It knows that it has to run that prompt every day. It knows about the sources that I trust and it will only look at those sources and it will then produce a consolidated report for me rather than me having to at every iteration step by step telling it to do that at 8:45 p.m. That's a very good example of how an AI agent works. But that's only on a standalone basis. The real power of AI agents comes when standalone agents start talking to each other. And this is where the magic really happens. Um, let me get this straight because I I I want to make sure that I understand it very well. You're saying that you can train an AI agent or you can create an AI agent to get some information and you can get another AI agent to receive that information and do a certain task. Is that what you're trying to say? Absolutely right. So the way to conceptually think about this is to think about it like an orchestra. I love music so this is a great example um that I love using. You nominate one AI agent to be your conductor. That's the role that you assign to what's called an orchestrator. And the conductor stands in front of a very large or uh orchestra. And she or he will enable the orchestra to play a beautiful symphony because she or he knows when each category of instrument should be playing and how fast they should be playing and can detect when some of the notes are being played um out of tune as well. It doesn't mean that they have to know how to play every instrument, but they do have to know how they work together and what the sound should feel like to them and how it should resonate with them. And that's what an AI orchestrator does. But the real challenge that a lot of these AI agents have, whether they be for example, Co-Pilot AI Studio from Microsoft or whether they be Gemini AI Studio from Google or others is that they need a way to talk to each other. So going back to our analogy, a conductor stands up and waves her baton and she uses the baton to direct different parts of the orchestra to start playing at certain in uh certain points during uh a symphony. And the way that translates into how AI agents work is that they have to have a common language. And thankfully that's been developed now. It's available already. So Google led the charge by developing what's called the A2A model, agent to agent, and Microsoft signed up to that as well. And what it now means is that once that goes into AI um in the mainstream, AI agents will be able to talk to each other and a they'll be able to orchestrate responses from each other. Now, first of all, you might ask yourself, but why? I can just use a standalone agent to do all the work that I want. Why would it need to talk to other AI agents? It's because as we covered earlier different agents have been trained using different LLMs and different training data which leads that the agent perform well in certain tasks than other agents which means that I'm going to get the best agent who works with analysis to transfer some information to the best agent to do contracts and then goes to the best agent to do mathematics and someone else who's going to do I don't know sales job or something and by orchestrating all of them I can um create a company I don't need um more than just a couple of AI agents theoretically so there are platforms out there that sell this idea to people that you can establish a and have AI agents running that company for you. Um, there's an open- source platform that's being heavily advertised called N8N and it's very powerful and it's cheap to use and people can install and run it locally on their laptops or on their cloud if they set one up. But it's worth having a go at playing with it and actually using it because it gives you an idea about how to get the process between one agent and another agent talking to each other. And lawyers might be quite surprised once they start using this platform to how intuitive it is. It's very very visual. It doesn't require much coding. It's what's called a low code solution. And so there isn't much to do in terms of learning how to code. So the the there's a bit of a cheat that you can use here which is to have um N8N running on your laptop at home and on your laptop also having a separate instance of chat GPT running. You can ask chat GPT to talk you through the steps that are required to install N8 and to set up an AI agent and it will walk you through those steps and it might even be able to see your screen while you do it. So it can point out any errors or enhancements that you can do and will train you up on how to use it and generate a bit of JSON code if you need that as well. And that is a really really efficient way of learning how to use these low code programs and platforms because it doesn't require prior coding experience. And so I'd advise all of your listeners and and viewers to try this out. You really can't go wrong with it. But getting over that fear of using the technology is half the battle once you're there. I think the lawyers that listen to your show will be very very surprised about what it can actually do for them. I mean one of the best things is that 50% of the audience are Gen Z uh in terms of the age and uh maybe 80% of them are lawyers. So I think it's uh your message uh received by the right audience and I think um I'm I'm myself right now listening to you thinking of how many applications I'm going to ask you after the episode that you can help me with because this is really great. There's so many repetitive work that I don't want to do and there's so many repetitive work that my colleagues are doing that I would like them not to do because it's better for them to u optimize their time work more on creative work or more analytical work or I don't know spend more time with the kids at home than uh just do this repetitive work that is time consuming and I think um they can help a lot It's 100% agree. Yeah, 100% agree. I think that AI does provide that productivity gain for people to visually see how long their day is going to be and to cut down on a lot of administration that they have to do, whether it be producing invoices or checking client matters if you're working for a firm to see just how much work you've done. Everyone hates time sheeting and um we have to produce if you work in private practice time sheets and so AI can actually go back and look at the outputs that you produced on a client matter and can make a reasonable calculation about the amount of time you spent working on those matters um simply based on the complexity of a document and how long you spent working on it and it can extract that information and produce a time sheet for you. Um if you are conducting legal research um Lexus Nexus recently teamed up with a legal tech platform called Harvey. Yeah. To combine the Lexus Nexus search function with the AI um that Harvey can offer and other legal databases are doing similar things as well such as Thompson Reuters and the practical law company too. So this is where law firms can benefit and I think that that speed of execution in terms of being able to analyze new legislation is going to really help lawyers to really be able to provide a more comparative view for their clients because the main advantage for them is that they can track legislation and how it progresses around the world and cross compare how that legislation compares to best practice in other countries. And so one of the tasks I did last year was to compare the progress of regulations concerning AI and I used AI to do that for me. So I downloaded the draft bills from the United States uh executive orders um bills that were going through in the United Kingdom, Europe such as the AI act and other countries. And I asked the internal AI platform that I used to generate a cross comparison study of what those legislation might cover if they went into law. And as a lawyer, it was fascinating to see how they differed and the approach to risk that they took in terms of what they were saying to the companies that they were regulating about how they should use AI as well. This is just one case study of hundreds, thousands of case studies that you can use AI for, which lawyers can now start to exploit generative AI for. Well, I think is fascinating. All what you're saying is really fascinating and it just shows how the future is going to be uh very AI uh driven and it's not only the legal profession, I think in in every field. What what's your advice then? I mean it it seems like too there's so many things that we have to catch up on and uh what what's your advice when it comes to I don't know I'm from the private practice um field and there's a lot of in-house councils and so on. What do you advise each of us to do? Well I think the first thing is to find something about AI that you really really are passionate about. No one can become an expert in every field of discipline concerning technology. It's impossible. And just as lawyers, we specialize in specific disciplines of law. Yeah. Pick a topic about AI that really really gets you excited that you would read an academic paper about that or you'd want to watch a documentary on YouTube about that particular topic and study up on it and become really good at it. And 50% of your listeners are from the Gen Z generation. Well, Gen D, they're wonderful at sharing information online. Yeah. So, research, talk to people and find out about what other people are thinking um in the legal community about that particular topic and how it impacts their lives and their professional lives as well. Because by coming up with topics that you're interested in, not only are you brushing up and getting crisper on those topics, but you're also developing your professional brand in the market. And that's really important today in the legal profession to have a professional brand online and to be known for a specific subject. I know lots of lawyers online just as you do who are regularly posting about topics such as arbitration or data protection and they are the people that I would go to for advice if I ever needed to look up sources concerning those disciplines. Well, your listeners can do exactly the same thing. And the next thing to do is to develop those skills that we talked about earlier to really really get over that fear of using that AI. But as a Gen Z or a member of the Gen Z celebrate uh uh generation, I don't think they have that problem. I think the problem that they have is persuading the managing partners and the senior managers that they report to to actually invest in AI or to spend money on it. Absolutely. You know, I totally agree with you. Yeah. Yeah. Yeah. It's a it's a It's always a cost and the decision if someone is going to drive this is going to be the um the people who like it whether they are Jenz's or old guys like me. But uh it is on us. It is it's in our hands. I mean, if you want to uh join the troops of the smart lawyers of the future, you're going to have to uh jump on that boat for sure and sail. And sail. Thank you for that. I think I've got one piece of advice for your listeners, which is that a very wise lawyer once told me that it's very, very difficult to convince a room full of millionaires that they've been doing everything wrong up until now. And so pick your battles. Pick the partners that you really want to impress upon who should adopt AI and invest in AI on behalf of the firm. If you work for a corporation, talk to your general counsel or the head of risk and compliance to understand how far along they are in their AI strategy as well. And I think you'll be pleasantly surprised at what they say to you once you find the enablers in your organization who would want you to advise them about how they should be using AI as a member of the Gen Z DE uh generation as well. And I think this is where the synergy comes in. This is where the skills that Gen Z's and millennials have outweigh the skills that we bring as Generation X's. Um because they're so intuitive with technology. They're so geared toward technology and they're so good at using it that we now need to seek advice from them. Absolutely. And so my advice to all of them is to say, look, don't ask what your bosses can do for you. Ask what you can do for them concerning AI. I think they'll want to listen to you. I think they'll want to hear from you. And just like we've been talking today, they'll want to find out about your opinions concerning AI as well. It's it's really amazing. A topic I will never stop talking about this. Just for your information, this episode we did not prepare. We did not script. It's just coming from our genuine interest in technology, our genuine interest in artificial intelligence. This is the future. You have to embrace it. Whether you um are working with someone who's enthusiastic about it or not, it doesn't matter. If you like it and if you're interested in it, pursue this knowledge and I advise everybody who is interested to understand more about technology, the use of AI and AI agents to reach out to Khaled, try to u learn more about it and understand it and embrace it and maybe you can again jump on that boat and sail. Thank you very much for coming here and thanks for the all the amazing knowledge and for uh giving us your time and your expertise. Uh it was amazing, fun and knowledgeable. It been an honor. Thank you so much. Thank you Khaled. So this is your do from Monday legal. See you next week.