Rev Streamlines Legal Evidence Review With AI (Full Transcript)

[00:00:03] Speaker 1: Thanks, everyone, for joining. Glad to go through Rev with you guys. My name is James. I've been at Rev for a little under a year, and I work with attorneys and law enforcement to make sure that the software is a good investment for you guys. I'm going to start sharing my screen, and then we'll have a Q&A session at the end. I'll run through a demo of Rev from start to finish, all the way from uploading files to interrogating those files to preparing documents. Okay, so this is the homepage, and the first thing I want to direct everyone to is this big square in the middle, because this is where you'll actually bring files into Rev. If you'd like to set up a storage integration, this is where you can set it up. I have mine set up to Google Drive, so you can see my folder structure there. Most folks set up a storage cloud integration, and other folks also enable a Clio integration as well, so they can see their cases. This is an example of what a Clio integration might look like. But if you're not setting up any type of integration, then you will have to drag and drop those files from your machine. Literally drag and drop it here, or browse the files within your machine. You can also drop a zip in here as well, and any hyperlinks that you have here. So once a file gets into Rev, let me touch on file types. For video, these are all the different file types here. I have yet to see a file type that we haven't been able to work with, and these are the audio file types, documents, and images. So once you drop a file in, the first thing you'll get is the transcript. And for about an hour-long file, it should take about a minute to load the transcript. Now this file is preloaded, but you can see I have my transcript here. It is in Speaker 1, Speaker 2 format, and it is verbatim. With the transcript, if I want to go ahead and share it, I can just share right here. Copying the link, or just putting in the email right there. I can also download it into PDF, Word doc, or TXT format. The way I have mine set up is inline, with timestamps and speaker names, and I'm keeping the filler words in there, but you have a lot of options here prior to downloading. This is an AI transcript, so what that means is it is 96% accurate. But if I would like to make a correction in here, then I can actually come in and make that correction. So you'll see I've highlighted Mr. Woods there, and I can make any corrections that I need to make. Those folks don't have to make too many corrections, because it is pretty accurate. If you do want to take it up to 100% accuracy, you can order the human transcript right from the homepage right here, and that would be a human on our end transcribing it. Along with the transcript, you're also going to get templates. So because this is a DUI stop, the first template that comes up is my DUI defense analysis. It's giving me the reasonable suspicion for the stop. It is going to cite everything back to the source file. So you'll notice this quote here, and then you'll notice the citation here, so that I can go. It's giving me the probable cause for the arrest. It's going through in much detail the field sobriety tests. Three different tests there with citations to all of them. The Miranda rights communication, how the officers acted, what were the conditions of the field sobriety test, and then any observations and inconsistencies. This template right here is meant to be all-encompassing of what is in this video here. However, if I want to make any edits, I can edit them right here. Or if I really like this and I actually want to share it to somebody as a PDF or Word doc, I could save it to my files and then export it as such. You can also just copy it and paste it. So copying to my clipboard, pasting it onto an email or a doc. I have all of that at my disposal. You can actually also create your own templates. So you'll notice that I have over 137. I will show you how to create your templates in the AI Template Builder. If I also want to ask any questions about what's in this file, I have this nice search bar here, very similar to a GPT or a Cloud, where I can just ask anything that I want or just take me to a specific spot. So I want to say, take me to the explanation of why he crashed. And right away, I'm getting the output here and it's saying based on the transcript. Now, that's important because what this means is it's not searching in Google. It's not hallucinating into other files on the internet or even other files within Rev. It's only looking at this one video. So whatever I point the AI to, that's what it's looking at. It's not going to look at anything else. And then right here, we have the quote. Yeah, I was looking down at my phone and I was going to change the stations. And the next thing you know, boom. Now, if I click on this 251, it's going to take me straight to the point in the video and the transcript. And you can see it right there. I can hear it. I know you guys can't hear it. And so that's how the search bar works. Templates are great to get an overarching view with citations. And then the search bar is just great for, hey, take me to this. I want to find this specific thing. And then you'll get it there. With this output as well, you're also able to copy and paste it. So if you want to archive that for your records. So you'll notice that this 251 was cited, and that's what it's playing here. If I want to clip this, I just click that little scissors. By the way, those scissors will follow me throughout the transcript. And let's say I want to archive this with the audio, the video, and that piece of the transcript. I can just go ahead and save that clip right here. And that's going to log the clip. And you'll see I have a running log of clips here already. And with these clips, a lot I can do with them. So I can share the clip. I can copy the hyperlink and share it with others. I can also download it to my machine or to a USB drive if I want to play it for somebody or play it in a trial setting as an exhibit. Can go ahead and just download this clip onto a USB drive and just play it for the group. Rev also has a mobile app. So any clips that I have will be available on the mobile app. If I just want to take my device to trial and just play the clip right there so everyone can hear it, I can do that as well. First thing I would recommend is renaming the clip. That way, you'll keep your clips organized. Perfect. So everything we looked at here was in the individual file. I will now gravitate over to looking at multiple files at once. By the way, you can have different editor settings here when you're interrogating this file. So that's where that will live right there. I'm going to come back to the homepage. Before I jump into the multi-file analysis, I do want to touch on the AI template builder here because I did touch on templates, how useful those can be. And I mentioned that you can build your own. And this is actually where you would build your own templates right here within the template builder. Click on new template, and we'll give it a name, a description, and a tag. And then the blocks are actually going to be how I want to prompt for this template. And then that'll be logged both in my individual file templates and then also in my multi-file. So let's now jump into multi-file. That'll live here in the Rev Insights. So a multi-file is exactly what it sounds like. I'm looking at multiple files at once versus just an individual file. Now I have a use of force case here. You can see I've already done quite a bit of work. And I've got 14 files over here logged on the right. The generated files, these are clips that I've taken from the files, and these are also documents that I've prepared from the output. First thing I started with in this file was give me the case chronology. Just as a first crack, I want a chronology of the events. You can see it laid it out for me. I've got a nice table and then subject lines with bullet points. And what I like to do with that is actually go into this document. So it'll generate at the end of each output. I can draft the document, which is going to be PDF or Word. And there it is. So now I have a very clean template of the case chronology, which I can share with folks as a PDF or a Word doc. I can also just download it to my machine. Also can make any changes that I want to make. So I have a very nice Word, Microsoft Word style ribbon here where I can make changes. And you'll notice that when I click into this, my cursor is right here. So if I'd like to get rid of something or add something of my own, I can absolutely do that within the doc editor. So it's giving me the timeline of the events, even if the files that I have are not in order in chronological order, AI will put them in order. And everything is nicely cited right here. So this bystander right here, where it says, ain't nobody's had an ambulance, not on time. If I want to verify that, click on this 38, it's going to pull up that point in the video. Among this whole stack of videos, it's going to take me straight to that one. And then, of course, I can clip that for my records. I know I asked something very overarching, like give me a timeline of the events. You can also ask very simple questions here. Like, if I know that something happened, for example, I know that in this case, there was a point where the defendant started having a seizure. I just don't know where that is. And I want to focus on that part of the video. I can just say, take me to the seizure. And it looks like the seizure was captured on a few body cams, not just one. And what I like that the AI is doing is it's separating out the different body cams with quotes from each body cam so that from there, I can actually focus on which one I want. If I really love this output, of course, I can draft the document. And now I have it in a nice, clean document right here. This is dispatch right here, OEMC1. And then these are body cams here, BWC3, BWC1, BWC6, 4, on and on and on. And if I want to verify one of those, I can just click on this number 9, and it's going to take me straight to that body cam where I can... Again, if this is really important to me and I want to capture this moment where he's having the seizure, I can clip this. And if I want to drag a little bit more to capture a little bit more of that, I can drag it there or I can drag it here as well. And what it's going to bring with it is the video, the audio, and that piece of the transcript. By the way, in the doc editor, if I want to... These citations, they're helpful for me, but if I want to share this with somebody, I may not want them... The distraction of all these citations, I can just turn those off. Turn them on, turn them off. We touched a little bit about templates here. Templates are very helpful as well on the multi-file as well, because if some of our attendant customers want to build out, say, a motion to suppress or some kind of a brief, this is where they can actually have that living, and they can leverage their templates to get them started on the motion. So if I want to say motion here, draft a motion to suppress outline, and then I'll just say go, and now the AI is working to draft a motion to suppress within the document editor. Of course, I can add my touches, I can make some changes. While that's generating, I do want to call out that Rev does a really good job of looking at not only audio video, but you'll notice in this case, this is a truck accident, and I've got hundreds and hundreds of pages here, none of them are videos, and it does just as well to find what I'm looking for, draft documents, cite everything back to a source file here. Among a large stack of files that would otherwise take me hours and hours, if not days, to read through, this can give me an overarching view of what's going on within these files, and now I can have a better idea of what's going on and focus on certain files here. So one of the things I asked it to do is list all the parties and their relationships, and that's what it did. It gave me the plaintiff, counter-defendant, different defendants here, it's giving me a little bit of a backstory on each person, and then the eyewitnesses, and then some non-party individuals, all the way down to counsel and insurance here. Also handles pictures really well, so I've got a lot of pictures here, and I asked the AI to just give me a summary of everything, and if I hover over, I can see a summary, so this is telling me a street photograph of heavy Midtown Manhattan traffic. Some folks like to look at screenshots of text messages, and they're wondering if the AI can actually read the screenshot, and so that's actually what I prompted here. So I have a batch of pictures, I'm looking for a screenshot within a text that just has one word in it, in this case, Taiwan, and it took me straight to that screenshot, where I know it's very tiny, but you can see way down in the bottom, she says, hi, Roger, my name is Anna, and I'm from Taiwan. So even within a large back and forth text message, if I'm looking for just one word within a screenshot of back and forth conversation, the AI does really well to find it. And this is actually a second conversation here. Okay, so here's my motion to suppress outline, going back to the use of force, and this is something that I created in the template builder, so you will create your own. And now here's my motion to suppress, right here. We recommend attorneys to look over this, make changes as necessary, it is a document editor, so we are guiding you to make those changes, to move things around, but what we're trying to do is speed up your workflow and get you maybe 60% there, so that you as the expert can come in and get yourself 100% there before sharing it out and printing it as a PDF. To start a new chat, you would just start here, and now it's going to guide you to upload files that are within Rev. So that's really Rev from end to end. We touched on bringing files in or setting up an integration, we touched on interrogating an individual files, having access to the transcript, using the search here to ask some off-the-cuff questions, leveraging templates, building your own templates, and then looking at multiple files within Rev Insights, and leveraging templates to draft paperwork and motions from there. So I know that was an overarching view. If you do come on board, we will have an onboarding team to help guide you through building templates, best practices, setting up integrations. But I will stop here. I know I've been talking for quite a bit. I want to take some questions from the group.

[00:19:09] Speaker 2: Hey, James, I do have one that's pretty good. So if I find an inconsistency between something like a statement and body cam, how could I show that side by side?

[00:19:22] Speaker 1: Inconsistency between a statement and a body cam.

[00:19:26] Speaker 2: But you have two pieces, yeah, that you want to show side by side.

[00:19:28] Speaker 1: Yeah. So the way that would work is it would list out the inconsistency. So for example, in my truck accident, and I'll show this here in a second, but in my truck accident, we have a lot of conflicting inconsistencies. And so it'll list out this person said this, and then this other person said that. And so it will cite the source. So whether it's a video or a statement or an interview, it will cite it, and then it will cite the second source. To show them side by side, the best thing that comes to mind for me is just to draft the doc editor. And you'll have those side by side with citations on both. So I'll go back to sharing just to show you what I'm talking about. So we'll expand this, and we'll come back to my truck accident. And we'll say, what are the inconsistencies among the witness testimonies? Okay. So right away, I'm getting some output here. I love how it's so organized, and it's not giving me just a word salad. It's splitting up things into topics for me. So we'll start from the top. So the inconsistencies among witness testimonies up here at the top is what I prompted for. And looks like there was an inconsistency on who struck whom right here. So Demby is a person, and he said, hit me from behind. Whereas Boresons says he never saw the truck before the impact. And he says, I thought you guys were in the other lane. And then it's citing the source here. So both are cited here, different witness testimonies. And when I draft the document, it would have both of them. If I want to then, in my doc editor, put a title there, that this testimony versus that testimony, just that way it's more side by side, then I can do that as well. Looks like there was other inconsistencies on tire conditions, who called 911, or whether they were even called. So I'll go ahead and just draft this as a document. And there it is. And then within here, I can actually, with my cursor in here, organize and move things around. Maybe I don't care about any of the other inconsistencies around 911 or the tire conditions. Maybe I'm just focusing on the direction of who struck whom. And this would be a perfect place to get rid of those other ones that I don't care about and expound on this one right here. But now I have things cited so that I can go back to the source file, listen to it, or read it for myself. Great question.

[00:22:40] Speaker 2: Thanks, James. I have two more. First, can you demonstrate how to group into folders body-worn camera from the same case in My Files?

[00:22:51] Speaker 1: Yes.

[00:22:52] Speaker 2: And then I have one other one after that.

[00:22:55] Speaker 1: So My Files is here. Group them into folders?

[00:23:02] Speaker 2: Yes. Correct.

[00:23:04] Speaker 1: OK. So we'll just pretend that these are body cams here. And I'll just go ahead and select them. And then we'll go to Move. And then if I have a folder already created, mine is the James Gregory one, I can just move them there, or I can just create a brand new folder called Body Cams, and then maybe specify the case.

[00:23:38] Speaker 2: Thank you.

[00:23:39] Speaker 1: And there we go.

[00:23:41] Speaker 2: Oh, you're not sleeping yet?

[00:23:43] Speaker 1: Oh, I'm so sorry. OK, let me redo that. OK, so My Files here. And let's just pretend these are body cams. And we'll click Move. And then we'll select either a pre-existing folder, or we'll just create a brand new one. BWCs. Case 1, 2, 3. And then I'll move them there. We'll hit Save. And now you just saw them leave My Files and go into that specific folder.

[00:24:22] Speaker 2: OK, great. Thank you.

[00:24:25] Speaker 1: And what was the second one?

[00:24:26] Speaker 2: Yeah, the second one is, are you able to disclose what AI platform you use on the backend? Is it a best fit among more than one LLM, or do you principally use one LLM?

[00:24:38] Speaker 1: We use more than one LLM. In the Template Builder, you can actually... So let's say I want to leverage this template here. Let's see. You can... Give me one second. So we do use more than one. We use Haiku. We use Sonnet. We use a few from Gemini. And I believe it's in the Template Builder. I know that there's a place where if you would like to select one over the other, you can do that. We try to default to the most powerful one, the best case fit. So yes, to answer the question, we do use more than one LLM, and you will have access to more than one LLM, and you can actually choose between them.

[00:25:29] Speaker 2: Thank you, James. That's all the questions I've got here in the Q&A. Perfect.

[00:25:36] Speaker 1: Great to meet everyone, and looking forward to speaking with you as well. You can put some time on my calendar. If you guys want to do a one-on-one demo, I'd be happy to do that. I always recommend if you do want to schedule a demo with me, you can actually share some of your own files with me, and we can show how those would work within Rev.

[00:25:55] Speaker 2: Thank you.

[00:25:57] Speaker 1: Thank you.