Name: Best Practices for AI Prompting
Uploaded: 2026-04-28T18:20:07.377Z
Duration: 1 h 19 min 24 s
Description: Best Practices for AI Prompting

Transcript for "Best Practices for AI Prompting": Hey, everybody. Thank you for joining us today for our Everlaw customer webinar. My name is Zach Souza. I'd love to welcome you as our senior customer marketing manager here at Everlaw. I'm gonna be, guiding you through the beginning of part two of our AI and ediscovery series, focused on best practices for AI prompting. So this interactive session is gonna focus on how you effectively prompt any LLM based generative AI tool. So although we're gonna be demonstrating these within Everlaw AI coding suggestions, this is a skill that you can translate to any natural language, prompting that you might be doing with a generative AI out there. So in just a few minutes, I'm gonna be turning things over to our user education team. Before we get to the trainings section of this, I just have a few housekeeping items. The first thing I wanna mention is that today's session is gonna be fully recorded. So if you have, the need to revisit and rewatch any point of today's session, all you need to do is join using the same link you used to watch live today. It'll be available there on demand for your viewing at any point in the future. If you do run into audio or video issues as you're watching today, there is a light mode that you can toggle on below the stage video, and that is going to help solve the majority of issues you might see. So please try that if you have any technical difficulties. Onto chat features. So on the right hand side of your screen here, you can see we have our session chat open today. Drop in, say hi, tell us how you're doing. You know, have fun talking with other members of the audience today and people from the Everlaw side are gonna be on there too, so say hi. And now on that note, if you have questions specifically for the user education team, I do want to direct you to the q and a tab. So it's different than the session chat. It is where we can best support answering your questions and organize, anything that comes through to us. So if we if you want us to answer your questions today, please drop them in the q and a so that we can make sure we do our best to answer them. If your question does end up in the session chat, we can't promise that we'll be able to, address it in time. So use that q and a tab. Lastly, I wanna call up the docs tab. We've got a lot of supporting information today, so use this tab to access any of the supporting resources. You can download them for future reference. Well, we've also got links to a number of different things that we will touch on at the end of the session as well. So with all that, let's talk about AI prompting, and we're gonna kick off with a question. So today's first poll is gonna be in the context of using Everlaw AI, but we wanna get a sense for our audience's familiarity with coding suggestions. So how confident do you feel leveraging coding suggestions in review? I'm gonna keep this poll up, and it'll stay live for a couple of minutes. But at this point, I'm gonna turn, everything over to our user education team. They're gonna get us started. Thanks, Zach. Hello, and welcome to part two of our webinar series on AI, where we'll discuss best practices for AI prompting. For those of you who may have missed our last session, we covered predictive coding, which is a tool on Everlaw that points you towards more documents that are potentially relevant. Once predictive coding is set up, it will work in the background, learning from the review work that users are applying to generate a relevant score for every document on your project. So the higher the score, the more likely it is that your document is relevant. Predictive coding is included in your Everlaw subscription, so it's free of charge. If you'd like to watch the recording of the last webinar, which goes over how to leverage predictive coding across Everlaw, the on demand link is available in the docs tab. So in today's session, we are going to cover coding suggestions, which is a powerful tool designed to help you streamline your document review and get those insights faster. Coding suggestions goes a step further than predictive coding. It's not only going to suggest relevant documents, but it also will provide a rationale for its suggestion and, if available, a link to a potentially relevant area of the document. So note that codes on Everlaw are the same as tags on other platforms. Basically, a way of labeling or organizing your documents. So for example, you may code a document as responsive or privileged and so on. The coding suggestions tool will suggest whether any given code should or should not be applied to a document and why. So to get these suggestions up and running, you will need to provide descriptions for your codes before generating suggestions, which gives you full flexibility over the process. And the accuracy of the coding suggestions hinges on the descriptions that you provide. So today, we will dive deeper into the nuances of writing effective coding criteria. And note here on permissions, to actually configure coding suggestion prompts, you must be a project admin or have coding suggestion configure permissions. However, do note that all of the tips that we're sharing today are relevant to prompt writing across the Everlaw AI tools. So even if you don't have configure or admin permissions, you can apply these same techniques when asking via the document q and a, leveraging the writing assistant, or even asking a question via deep dive, which is a tool that we're gonna focus on in greater detail in our third session in this series. Finally, it's also important to keep in mind how your team will actually interact with these tools as there's two ways to generate coding suggestions based on your review goals. So you can use Everlaw AI including generating individual coding suggestions at the document level at no additional cost. So for example, if you're looking at a single document and you want a coding suggestion or a document summary, you can generate that at no cost as long as Everlaw AI is turned on in your project. However, it's also often useful to run AI across many documents at once. So batch actions, batch AI actions, like generating coding suggestions for an entire results table and our deep dive tool, use Everlaw AI credits. These are a usage based cost managed through your organization's Everlaw AI contract. If you wanna see a full breakdown of which actions are included at no cost in which use credits, we've included a helpful article in the docs tab for you to reference. And if you're unsure of how this applies to your specific organization, please reach out to your Everlaw point of contact to learn more about pricing and activation. So now on to the fun stuff. So in today's webinar, we are going to watch and learn from two diligent Everlaw users as they try to effectively set up their coding suggestions for their case. So you, the audience, will decide what next steps they should take to successfully complete their work. So before we dive in, let's take a closer look at a bit of case background. So a secret recipe for a restaurant's signature sauce known as the golden ember sauce has been stolen from the head chef's private recipe book, and an internal investigation has been launched. Michelle must sift through various types of documents to uncover the truth, but she's running out of time. Can you help her configure coding suggestions and speed up her review? Alright. So now that we are up to speed on the case, let's meet our two legal professionals and see how the case is progressing. And I'm gonna leave you with one tip. It might be helpful to jump over to the docs tab and open the resource titled tips for writing and coding suggestions to help you address some of our polls throughout the session. Hey, Kayla. Thanks for meeting with me today. I'm in a bit of a bind with this Golden Ember case, and I'm really trying to lean on Everlaw's coding suggestions to speed things up. But they're not really giving me accurate results. Sure. Happy to help with this. So what's your current understanding of coding suggestions? I don't wanna give you any redundant information. So I think I get the general idea of coding suggestions that it's an Everlaw AI tool, that suggests relevant codes based on the prompts that you've written. And I especially like how it provides the rationale for its suggestions, but I'm having a bit of trouble getting accurate suggestions. Could you help me with this? Yep. You asked the right person. I actually use coding suggestions quite a bit on a big case a few months ago, and it made a huge difference. So I was able to get through documents much more efficiently. So I'd be happy to share some of my tips and tricks. Let me show you how I set up coding suggestions on my case. Okay. So I'll head over to the project settings by clicking this person with a gear right here and going into my project settings. So to update, coding suggestions, you're gonna need to be a project admin or have configure permission for coding suggestions, which you have on this case. Correct? Yeah. I'm a project administrator. Okay. Great. So, first, I set up my codes for my project from the codes tab. So codes fall under categories on Everlaw. So for example, I have my issues category and then several issue codes, and, those are the codes that I can actually apply to my documents. And this includes things like compliance warnings, funds misuse, things like that. So I can add more categories and codes from here if I needed to. Then I wrote my descriptions for these. So I'll go over to the Everlaw AI tab and go into my coding suggestions tab. And so from here, I wrote descriptions for each coding category and each individual code. So these along with your larger case description are what the AI tools are gonna use to determine if it thinks a code should or shouldn't be applied to a given document. So essentially, you're confirming the definitions of these codes here. With coding suggestions, it really comes down to how you prompt the AI. It's almost going to be an art form. The more detail and nuance that you provide, the better your results are going to be. Let me back up here and move to the Case Description tab just so that we can provide an overview of the entire case before we start specifying codes and category definitions. I'll click on this Case Description tab right here. So everlaw.ai is going to use the case description to understand the larger case, both when applying coding suggestions and when responding to questions that you ask of Deep Dive. So, we'll talk about that again in our Deep Dive webinar. So you can see an example of what I what I wrote for my case on the screen right here. So, within this case overview, you're gonna wanna summarize the main issues, arguments, key entities, their roles, and any other important context that applies broadly across all of your coding suggestions. So once that's complete, we can move back to the coding suggestions tab, and then we can select, configure next to any, coding categories that we'd like to configure our coding suggestions for. So I'll just go ahead and click edit in here. So this is where I already wrote out my, descriptions. So this is where I would for descriptions, I'll wanna include information that's relevant to most codes within that specific category, but not necessarily to other coding categories. And this is where you can define your relevant terms, entities, that are specific to that category. And then finally, we have the, code descriptions, and this is where you'll list the specific criteria for applying that particular code. So you can describe the information, attributes, or qualities that are found in document text that would suggest applying this code. You can also include information that might appear directly in the document text, but it's still important for evaluating that code. Alright. So that's generally what you need to do to set up your coding suggestions. I also have a resource with some tips and tricks on prompt writing that you might find helpful. So let me pull that up. And as a reminder, this is gonna be available in the docs tab, within Goldcast. It's called the tips for writing coding suggestions. So this resource really helped me as I created my coding suggestion descriptions. I think we can use this as we walk through updating your current descriptions. So I'll go ahead and send this to you so that you have this on hand as well, Michelle. Awesome. Thank you. That that was really helpful, and I I can see this this resource is really gonna help me improve my prompts with these tips. Should I share my screen and show you what I've got so far? Yeah. Definitely. Let's see what you're working with. So here I've got in the review window looking at a document. And when I click on the sparkle icon over here on the left to actually see my coding suggestions, I see that it's being suggested as a yes for quantity discrepancies. But when I actually read the document, this is not responsive to that at all. Okay. Yeah. I see how that's an issue. So my theory is that we'll need to revise the descriptions that you wrote for your coding suggestions. So let's start out by talking about some tips for writing a case description. Can you pull up that resource that I shared with you? Yeah. There we go. Let's go ahead and scroll to the part that talks about tips for writing a case description. Awesome. So taking a look at those first few suggestions, here are some tips for writing an effective case description. So first, you want to summarize the overall case, so you should include the main issues and arguments at play. You're going to want to define the key entities and their roles and any additional information that's important context for making coding decisions. And then finally, this is a good space to provide any important information about the case that's likely relevant across multiple coding categories. So overall, you want to be general in the case description. Less is more. You don't want your case description to be too long. So with all of these tips in mind, let's take a look at your case description. You should be able to see it, in the review window actually by clicking the View Configuration button. Oh, do I need to click this button here? Yes. Okay. Okay. So here's my case description. A restaurant seeks to prove its signature sauce recipe was stolen from within, pointing to suspicious staff behavior, late night kitchen access, and inventory discrepancies as evidence of internal misconduct. Okay. Let's take a moment to analyze the current descriptions to see what they should do next. Based on Michelle's current description, what do you think she should change to increase the accuracy of her coding suggestions? To help you answer this question, pull up the writing coding suggestions tips resource from your docs tab, and then go ahead and open the polls tab to weigh in. We'll give everyone a couple of moments to respond to that poll. Okay. Let's go ahead and see what they decide. Okay. I see a few things here that I really like. So you briefly summarize the issues in the case, like suspicious behavior and kitchen access, and you provided information that's relevant across multiple coding suggestions, such as internal misconduct. But there's something you're missing. You didn't mention the key entities involved in the case and their roles. Okay. That's really helpful. Thank you. So I think I can maybe go ahead and update that. Let me show you what I've got. I'm just gonna paste this in. Okay. So with this new description, I went ahead and specified the name of the restaurant, and I also added the names of the employees who are our prime suspects. And I also specify the name of the restaurant that we're invest or excuse me. I also specify the name of the recipe that we're investigating as well. Great. I really like how you define the key entities and roles while keeping that description really simple. Now let's talk about some helpful tips for your coding categories. Can you go ahead and bring up that resource one more time? All right. Great. So when you're writing coding categories, you're going to want to include information that's relevant for coding decisions to be made for most codes under this category, but not relevant to other coding categories. Next, you'll want to list alternative names for entities. So for example, I might explain that, you know, Everlaw was formerly known as Easy ESI or Company ABC consists of a joint venture between AB Incorporated and C Corporation. So this is gonna capture more instances of those entities. And then finally, make sure that you're defining technical terms. So you'll want to define any technical, chart terms or jargon, acronyms. For example, you know, don't just if you say DOI, make sure that you, describe that that means date of incident. Oh, okay. I think I understand. So let me go find that category description for the code that was giving me trouble. So if I go back to my review window, then this is the category description for my inventory issues coding category. This category is for identifying issues related to GES in ingredient inventory. This includes POs and IARs related to the KIs. So let's take a look at the coding category description for inventory issues. What do you think needs updating based on the tips that we just went over? What could Michelle do to improve it? Go ahead and open up that polls tab and submit your response. We'll give everyone just a couple of moments to do so. Okay. So let's go ahead and see what they decide. Okay. For starters, I really like the way that you included relevant information for this category and not for other categories, like the focus on ingredients and how you listed alternate names for the recipe. But it looks like you didn't define your technical terms and acronyms, so that would be helpful. Oh, okay. I I think I see what you mean. Let me see if I can go ahead and edit that. So I'm gonna click edit, and let me copy in the updated description. Okay. So for my updated description, I went ahead and defined GES as golden embersauce. I also defined POs as purchase orders and IARs as inventory audit reports, and I defined KIs as key ingredients. And then I went ahead and listed all of the ingredients for good measure. And then finally, I defined the terms that I used, in the the in the definition of inventory audit reports. So things like shrinkage and discrepancies, because those are actually very specific to the restaurant industry. Yeah. I think this looks great and good catch on that ingredient list. Alright. Now let's take a look at some of the tips for writing the actual coding descriptions. So if you head back to that resource So first, you're gonna want to be mindful of when your language may be ambiguous or subjective. So you should define words or expressions that might be subjective. For example, instead of just writing that you're looking for misleading statements, you should explain that a misleading statement is characterized by a projected profit. Next, in general, you're gonna want to be affirmative with examples and instructions. So avoid saying don't look for internal communications between employees of Company A and instead say focus only on external communication involving an employee of company a and at least one member of another organization. So you you wanna be affirmative with what you wanna find. And then finally, if appropriate, describe the sentiment or tone that suggests a specific code that a specific code applies in that circumstance. So you can include sentiment adjectives in your descriptions like angry, upset, negative, positive, or optimistic. Okay. Yeah. These are great tips. Let me take a look at the description for the code in the review window. So, quantity discrepancies was the code that was giving me trouble, and, this was the one that was giving me those inaccurate suggestions. So if you could provide some feedback on this description, I'd really appreciate it. Here's what I have so far. So this code is for when items are missing or item counts are off. Communication may have a tone of concern or alarm. This includes documents specifically mentioning missing ingredients, items, or significant discrepancies. Alright, folks. Let's take a look at Michelle's quantity discrepancies code description. What could she do to improve it? Again, go ahead and head over to that polls tab to submit your response, and we'll give everyone a couple moments to do so again. Okay. Let's go ahead and jump back to Kayla and Michelle to see what they do next. Okay. It looks like you did describe the sentiment and tone associated with the code, when you describe the tone of the documents that you're looking for, which is great. You're also providing examples of information that might be found in relevant documents, but you will need to revise the code description to be a little bit more specific, including examples of what actually constitutes significant discrepancies. So this is gonna give the AI a much clearer criteria to work with. For example, instead of documents indicating disruptive or unusual behavior, you might try identify documents containing language that explicitly describes actions by employees that disrupt the workplace. Adding examples like this can be incredibly helpful. Wow. Yeah. That's a fantastic suggestion and really specific. I definitely see how being more explicit and giving examples will help the AI understand what I'm looking for. Let me make those improvements right now. I'm going to click edit again and go down to the quantities discrepancy code. Let me paste in what I've got. Okay. So, I defined the terms more clearly. So I specified that this is related to those key ingredients, and then I listed them again just to keep it consistent with the case description. And then finally, I also defined what we are looking for in terms of what a discrepancy, actually is in these documents. That looks great. Okay. So I'm gonna hit save. Now that I've updated the descriptions, what should I do next? Great question. So now that you've refined your criteria, you should rerun the suggestions on that problematic document and other sample documents to see if the changes made a difference. Okay. So I'll go back to the, Everlaw Assistant and run the coding suggestions again. But I do notice the the soft yes button has changed next to quantity description discrepancies now. It's it's got like a dotted line around it. Do you know what's going on there? Yeah. Any existing suggestions that are out of date, you edited the description, so now they're out of date. They're gonna have a gray dotted line around them to indicate that. Oh, okay. So I think I understand. So I'll go ahead then and hit generate and see what happens. Okay. Great. Look at that. Much better. It's now saying no for quantity discrepancies. So does that mean I'm ready to go ahead and start running coding suggestions across all of my documents? So not quite yet. It's all about testing and iterating to ensure that your criteria performs well on a representative sample of your data. So you wanna verify that the updated suggestions better match your manually applied codes and that there are no regressions. Okay. I I think I I see how that can make sense. So how do I get a sample of documents? Yeah. So go ahead and head over to your homepage and go ahead and click the view all documents button in the upper right hand corner. And so now we're looking at all of our documents, and so we want a sample of all of our documents. So go ahead and click on the settings button. You should see that sample option on the left. You're going to narrow this set down to a sample so that you're only looking at 25 documents. Go ahead and type in 25. And one thing I'll just note here, you may also want to change the grouping to none, to see just these 25 documents, or you can group by attachments, to see the full families together. Note that this may bring in more than 25 documents, but you're looking at that full attachment family. But you can just leave this as is. That's fine. And go ahead and hit save. And what we're gonna do is we're gonna add these documents to a binder for the first sample test. Okay. So I have the documents in my, sample set. So I'll go ahead and open that binder from the home page. Great. So you have two options for running coding suggestions on this sample set. You can perform a batch action to create suggestions for all of these documents at once. Note that batch actions do use AI credits, so there is a cost associated with them. And to do that, you can click on Everlaw AI and then coding suggestions, And you can generate coding suggestions for these documents for all of the codes that you've written descriptions for. Alternatively, you can open these documents one by one and generate suggestions from the review window, which comes at no charge as these are single document actions. Okay. Well, I know we have some AI credits, so I think I'm gonna batch run this just to save some time. I do want to see the coding suggestions column, so I'm going to go ahead and click on this add coding suggestions column button so that I can see that in my results table. I'm going to click generate and see what happens. Great. So we'll go ahead and rerun those suggestions. And then once that's done, we'll take a look at the first document and manually review it before peeking at the coding suggestion. Okay. Let me pull up this first document. So looking at this document, this does not look relevant to quantity discrepancies. So if I were reviewing it myself, I would say that's a no on quantity discrepancies, and I wouldn't apply that code. So let me see what the coding suggestions decided. Oh, look. There we go. Over here on quantity discrepancies, it's also saying no. Perfect. So you'll continue to do this for the rest of the documents in your sample set. You manually review and then check the coding suggestion. And if there's an error, take note of it and update your coding descriptions again. Then you can rerun the coding suggestions on that same sample set to test your updates. So if that all looks good, you'll run the same test on a brand new sample of 25 documents. Wow. Okay. Well, that seems like a lot of steps. I'm not sure if I'm gonna remember all of that. Fair enough. If you wanna follow this iteration workflow, you can take a look at the, workflow for coding suggestions for review, and that is in the docs tab. That's gonna outline all of the steps for you. Okay. Great. Yeah. Once I'm done with this, I'm definitely gonna want to refer back to that. So after I finish that iterative process, can I just start running coding suggestions across all of my documents? That's right. But, you know, if you really wanna speed things up even more, you could also set up a predictive coding model to learn from your coding and then help prioritize the next set of documents to run coding suggestions on. I hadn't really thought of that. I know I set up a predictive, coding model last month after I watched the webinar on predictive coding for review. But how do I use that with coding suggestions? They aren't really the same thing, are they? Yeah. You're right. They're not the same thing, but they can definitely complement each other. So predictive coding is built on supervised machine learning and continuously learns from reviewer decisions to predict relevance across the entire case. And then coding suggestions, on the other hand, leverages pretrained large language models, and your specific prompts to suggest codes for documents. So the key here is that predictive coding can help you prioritize which documents you send the coding suggestions. For instance, if your predictive coding model has identified documents that are highly likely to be relevant, you can run the coding suggestions specifically on that prioritized set. So this means you're focusing your AI efforts on documents most likely to re to yield important information. So this is significantly gonna accelerate your review. And remember, predictive coding is free of charge. It's included in your Everlost subscription. Okay. So that means that I can use my predictive coding model to narrow down my focus for my coding suggestions. That's really clever. That's gonna take me a a excuse me. That's gonna save me a ton of time, and some money too. How can I do that? So let's go back to your homepage one more time and click on that view all documents button. And we'll wanna add the predictive coding model to your results table view. So you can go ahead and click view and then add or remove columns and then find that predictive coding model that you created. Alright. So now that we have this column, you can see the prediction score for all of your documents. So, from here, you could sort your documents if you want or you can filter your view. So let's go ahead and filter your view to documents with a prediction score of 50 or higher, and you can now batch generate coding suggestions for these documents with the credits that you purchased. So you can do it one by one free of charge, or you can do it on the batch level for credits. And now you're just using coding suggestions on documents that the PC model thinks are likely to be relevant instead of running them across your entire dataset. Okay. That's really helpful. Let me just recap just to make sure that I've got it handled for next time. I need to create my coding categories, use that tip sheet to create coding suggestions, to configure my coding suggestions and my case descriptions. I need to test my coding suggestions on a small sample of documents and update them as necessary. And then finally, use predictive coding to help me prioritize documents to run the coding suggestions across. Yep. That all sounds right to me. It might seem like a lot of steps at first, but you've summarized that perfectly. Awesome. Well, thank you so much for your help today, Kayla. Yeah. No problem. Once you get into that rhythm of refining your prompts and prioritizing your documents, you're gonna be flying through review. Good luck with the case. Alright. So let's wrap things up with one last poll. Now that we've made our way through today's webinar, how comfortable are you with AI prompting on Everlaw now? Again, head over to that polls tab for one last time and submit your response there. We'll give everyone a couple moments to do so. And while folks are responding, I just wanted to thank everyone who joined us today. I know that we've covered a lot of ground on coding suggestions. So please use the resources that we provided in the docs tab as you create your coding suggestions. And as a reminder for everyone, these prompting tips can be used at any time you're prompting or asking a question of Everlaw AI. Think of asking a question of an individual document in the review window, leveraging writing assistant, and especially with deep dive, which we're going to cover in our next webinar. And as you're getting more familiar using these tools, please don't hesitate to reach out to our support team with any questions that you may have. And I'll go ahead and pass it back to Zach to wrap us up. Thank you, team. Give our speakers some thanks in the chat. They worked really hard in getting this session together for you all, and we hope you learned how to be just a little bit more effective in using those AI prompts. I know it's something that helps me whenever, you know, what I'm trying to use. And get some extra mileage out of how I'm leveraging, AI in my day to day job. So hopefully it helps you as well. Before we close, I have a couple of announcements. First, the user education team is always looking for feedback, so we're gonna make their training survey live. You should be able to click on it, on your screen now. That URL is gonna stay up for a little bit. Please take a minute and let us know how you enjoyed today's session or if you have any feedback for us and the team, we always really appreciate that. Next is actually something new we wanna share and make part of some of these webinars. We have so many of you turn out to this. We don't wanna miss the chance to remind you what's new on the platform. We've got monthly versus and with April's recent release, we've got a number of improvements to help your team review even faster. So starting with some database and organization admin, improvements, you or admins can now upload folders in their original compressed state exactly as they arrive. So this simplifies large data uploads and helps save you time from the start. If your team is very collaborative or you are a project leader admin, you would also benefit from the new global object access permission. With this global object access access, project admins can grant designated user groups, view, edit, or full access to all work product objects in a project. No manual sharing is required there. And last but not least, for story builder, you no longer need to input fax by hand into the facts timeline. You can actually import via CSV or Excel files to create new timelines or bulk add facts to existing ones. So this can turn your existing tables or exported fact timelines into structured shared timelines. So all of these are great. If you want a more detailed overview, we have a link to our latest release notes in our knowledge center in the docs tab, so don't miss that. Next, I want to give a just a real very quick reminder that we have our third and final part in this AI and ediscovery series coming up next month on May 26. It's gonna be all in on Everlaw AI deep dive, and we're actually gonna be structuring it in a fun choose your own adventure workshop. So we're excited for that, and we hope you'll have fun with this one. You'll get to experience how you can use Everlaw AI's deep dive at many phases of a project, including servicing insights, finding and, asking the right questions, and then reviewing large datasets. So that'll be fun. Make sure you're there. If you haven't already registered, we've got the in the docs tab as well. Lastly, and this is brand new, if you're interested in even more AI knowledge in a hands on environment, you can sign up for Everlaw AI Lab. So what is an AI Lab? This is a one hour in person on hands workshop where the, Everlaw team, experts from our team can actually come to your office, set up we can set this up in a lunch and learn style format, but, they will deliver a workshop that actually gives you access to practice projects, and they're gonna review with you all sorts of Everlaw AI workflows, including deep dive, review assistant, writing assistant, whatever you're really interested in learning from our Everlaw AI suite of tools. This one hour session can be delivered on-site for your team. So So we're really excited to share this. We hope that this is a good forum after seeing and learning about AI prompting today. If you want to, learn in person with our team as well, you can schedule an AI lab. Look for the link in the docs tab to inquire about scheduling, and members from our team will get back to you about visiting your office and seeing you in person. So we're really excited about that, and that's the last thing I wanted to share with you guys today. So there was a lot to go over. We hope you enjoyed the session, learned, and and could take what we talked about in AI prompting into your next project. And with that, we will see you in May at our next webinar. Thank you so much.