BDD Program - Video & Script

Document Purpose

The purpose of this document is to provide our customers with a demo framework that can be used to present Reveal 11. This demo was created by the Reveal Strategic Sales Engineer team, and is used as a first touch demo to illustrate the differences or value-adds of the Reveal platform. This is simply a template demo, so please feel free to customize this presentation to your own needs. 

Video

Background

Hello everyone and welcome to the Reveal 11 demo.

To set the stage today, the purpose of this presentation is to provide our client with a demo framework that can be used to present Reveal 11 to their client. This is simply a template demo, so please feel free to customize this presentation to your own companies needs. With that in mind, thank you so much for joining us today and I will now begin the Reveal 11 demo.

Introduction

Hello everyone and thank you so much for joining us for the Reveal demo. What you were looking at today is Reveal 11, a transformative eDiscovery platform.

As I go through the platform today, I will highlight three things that differentiate us and make our clients more efficient.

1. Fast
   1. First, it is fast. To be efficient, you must be fast. While I use the system today, you will see how responsive and intuitive it is and searches are going to be running in sub-second time.
2. Review Enhancers (Tools within Toolkit)
   
   1. Second, we like to throw a bunch of tools into your toolkit - what we call review enhancers. We want all the tools in your toolkit to help you with the most difficult workflows. You will see as I go through this demo that I'm going to highlight things like AI Translation, Transcription, and Image labeling, which are very powerful technologies baked into our software that are going to help you with those very challenging workflows.
3. Effortless AI
   
   1. Finally, one of the biggest differentiators that we have is our artificial intelligence and what I'm going to refer to as Effortless AI. When you load data into our review platform, which is what we're looking at right here - all of the features and functions I'm going to highlight on the artificial intelligence are done automatically for you at the time of ingestion. So no need to be building out the cluster wheel, concept searching, or conceptual indexes. All of those features and functions that I will highlight today all get run in the background on your behalf, hence Effortless AI.

One important note before I dive in, for today I'm going to be presenting to you from the perspective of a System Administrator, meaning I'm going to have access to all the bells and whistles in our software. If I wanted to lock this down to where a user only saw the grid view and they had access to their batches in just one or two of these mass actions such as export or print, we could lock this down to where they would not see the Dashboard, the Clusters, or the Heat Map. That is all up to your discretion on how you set up your permissions within this environment.

 

Dashboard

Alright, let's dive into it. What we're looking at right now is called our case insights dashboard. This is where we can see a high-level overview of all of the data that we have loaded into our system after either processing or loading. As we go through this, I could quickly start filtering down using the different filters that are enabled within this dashboard.

Widgets

You'll notice that we are at 40,000 documents and maybe I want to filter on a date range I can click and drag and I am now down to a subset of documents on that date range. I could then choose documents based on the originals, the near dupes, or the exact dupes. I could filter on an extension type or maybe a custodian and as I continue to click on these filters, I am building out a more comprehensive query. I'm on that subset of documents and at any point I could flip over to the grid view. Now, the important thing about the grid view - this is your home base, right? This is where you're going to come in and you could start doing searching, tagging, document reviewing, or assigning batches. You're going to spend a great deal of time right here - I'll circle back on that in just a little bit.

One of the biggest differentiators that we have within our software is our artificial intelligence. Where I want to begin is by showing you a feature that we call Entity Analysis. I like to think of entities as the people, places, things or - the who, what, where  - within your data.

When I filtered down on this you can see all of the available entities that we currently have. Maybe I want to filter down on credit card numbers or maybe I want to look at phone numbers or for example Social Security numbers. I could click on that and I'm now looking at all of the Social Security numbers. Now an important note, this is unsupervised learning - our AI has run automatically in the background at the time of ingestion. So before any of your reviewers have looked at a single document, our AI is identifying these entities and saying here are the documents that contain a Social Security number.

This is an excellent tool for helping you find PII. I could grab this and say I want to view all documents and grab all of those that contain a Social Security number. I could export that out to CSV. We provide you all of that information right at the onset automatically.

One other entity I will highlight is one that is called entity product and it's the product will show you the high level themes or topics within your data. So at any point - maybe it's an investigation or maybe it's third party production data - you can come in and start getting a sense of what is within your dataset based on these topics or themes. 

Emotional Intelligence

Next, when we ingest documents, we do what is called Emotional Intelligence or Sentiment Analysis. We run natural language processing against the text within your document and we can find things such as positivity, negativity, pressure, opportunity, and this is unsupervised learning. Before any of your reviewers have even looked at a document, our AI is coming in and saying here are the high scoring documents that are showcasing examples of negativity.

I could click a button and filter down and use this as a prioritization exercise. I can quickly filter over and look at everything within the review grid so I can grab the results. I can pop over the review grid and here is where we could start using this as a way to jumpstart our review to find those key documents that we're interested in. Again, all of this is available right as we go in, so we could see we are getting all of the scoring on the emotional intelligence right at the time of ingestion.

AI Models

One of the biggest differentiators that I do believe we have is called our pre-built AI model library we can see in this widget right here.  I'll flip over to our model library here and show you. Our data science team has built over 25 pre-built models that are available at the onset of a project. For an internal HR investigation or employment matter, you can see we've got bullying and toxic behavior model, we've got comments on appearance, we've got hate and discrimination model. One very valuable utilization of this is to find junk documents that we could look for out of office or advertisements and promotion and any of these models can be added into your project at the onset.

For an internal investigation, we could use one of these models to start jumpstarting our review. On the flip side, one feature that our clients love is the ability to deprioritize documents and they can do that with some of these pre-built models that we have that will help find junk such as out of office e-mail or advertisements and promotions. These are excellent prioritization and de-prioritization tools. 

I could come in at the onset of this project and say find me all of the junk documents before I've even reviewed one document. I could click on this. I'm now looking at the subset of the high scoring documents that are junk - I could flip over to the grid and maybe I want to tag all of these documents as non-relevant. Maybe I want to assign them out as a batch or I want to folder them and come back to them later on. We give you the ability very easy - to start filtering down utilizing these prebuilt AI models from the onset of your project.

The pre-built models are excellent, but one of the key features I do want to highlight is that it is extremely easy to build out your own custom AI models. As an example, and I'll touch on this a little later in the demo, maybe you want to build out an anti-trust model. Well, as your team is going through and tagging positive and negative examples of anti-trust, we are going to start training the AI model. At the conclusion of your first project, it is that simple as taking that model you've built - you can say I want to import that model and at the start of your next project it can be brought in, and it can say here is your anti-trust model - it is going to automatically start scoring based on the expertise of your team. As you continue to use it from project to project, it is going to get a larger sample set, it is going to get more refined and it can start to very, very quickly identify those key documents you care about right at the onset of your project. Once again, Effortless AI. 

Flipping over to the grid view - this is really your home base within our software.

From here if I wanted to grab all these documents and do a mass tag, I could throw everything into a folder. I could mass tag these documents.  If I wanted to start going in and reviewing documents, we could do that. You'll notice the folder structure is on the left and our search toolbar is across the top.  This should look very familiar to you.

For more, see: Lawyers Guide to AI Models

Concept Search

I want to showcase a couple of built-in features that we have that are very powerful when it comes to searching. First, when I come in I can type in a word here and you'll notice I could easily do a keyword search. Now you'll notice I could do a keyword search and this is going to go and start finding all the documents that contain the word raptor. 

What we also have is what is called concept searching. When we ingest documents into our system, our AI runs against them and it starts grouping documents based on conceptual similarity. So, if I run a search on Raptor, the concept search is going to pull back the top ten conceptually similar terms and phrases to Raptor.

The way I like to think of this, a keyword search is very good at hitting on the term that you just searched on.

• A concept search is incredibly powerful at helping you find new information about your data that you may not know. This is extremely powerful when dealing with investigation.
• When dealing with third party production data - this can start to show you new key insights into your data that you may not have known.

For examples, I run this search on Raptor and I can basically see - here are my top ten conceptual terms. I may not know what Catalytica it is. I can quickly come in and say I want to make that a required concept. I could hit the search button and I am now down on that subset of documents. You could quickly start viewing these documents from here and tag them. Any of the workflows you've built, all can be done very easily.

So again, the concept search is a very powerful tool to help you start finding new information about your data that you may not know. From within the concept search I mentioned, here are the top 10 conceptually similar terms and phrases to what I searched on - on the right we have essentially concept 11 through 100.

For more, see: Searching in Reveal 11 - Concepts

Brain Explorer

One additional feature, the Brain Explorer. What the Brain Explorer does is it will take the top ten terms and phrases that are conceptually related to Raptor, the ones we just saw that were in that list view, and we can now start to visualize them.  Now these are the top 10 terms on the left and again concept 11 through 100 on the right.

Let's say I want to dig in further on Catalytica - what I could then do is come in and when I click on Catalytica, it is going to pull up and show me - here are the top three concepts that are related to Catalytica. This they very powerful tool at helping to provide a visual mapping of how all of these concepts are interconnected to one another.

Now another term is Enconto Falls - maybe I want to know more about that and I can click in and start to see the top concepts that are conceptually related to Enconto Falls. As the Brain Explorer provides the visual mapping of how the concepts are interrelated to one another - I could add to the search, I could then filter down on a subset of documents utilizing the power of the Brain Explorer and the concept search.

For more, see: A New Way of Visualizing Concept Searches

Cluster Wheel

Another integrated AI technology that we have is called the Cluster Wheel.

When we ingest documents into our tool our AI starts to group them based on contextual similarity and what you'll notice here is that with the different colors - these are the different clusters that the AI has identified.

I look at the cluster wheel as having two very powerful features. One it starts to cluster those rich key documents you care about and another is it will also cluster junk and allow you to remove it. 

You could quickly grab the junk cluster and say let's tag all of these as non-responsive. Again, a tool to help the prioritization and de-prioritization of document that can all be done automatically on your behalf. 

Maybe I'm interested in this blue cluster. I can click in, I see terms like natural gas or physical, and we can start seeing all the sub-clusters right here. When I click into one of these clusters, you will notice the cluster identify how many documents are within it and it will also show me key terms being discussed.  

Once I have filtered down on a sub-cluster of interest, I could start reviewing the documents. From here I could start to tag the documents, folder them, or any of the workflows you follow. All of this is done where our AI is identified and group documents that are contextually similar to one another.

One of the best features I do highlight when I talk about the Cluster Wheel - let's say you start reviewing these documents and you find that this document right here is an extremely powerful example document or maybe it's a hot document - a smoking gun. What we are able to do is we can click this little symbol you see right here in the thumbnail preview and when we click this, it is going to show us exactly where in the Cluster Wheel the document resides. 

Now you may ask yourself, why does that matter? Well, if I know that this document is a hot document an example document, something I want to know more about? If it resides here in the Cluster Wheel that means that the most contextually similar documents to the one I just found all reside right on this line. This is essentially a find more like this button.

I always like to highlight that feature because it is an extremely easy way to start finding documents that are contextual related to one another. Extremely powerful tool that can really help you start to narrow down on information in a very short period of time.

For more, see: 11 Reasons Lawyers love the Cluster Wheel

Heatmap

On top of all the wonderful I technology that I've shown you today, I want to highlight a feature called the Heatmap. 

The Heatmap - essentially I like to to think of a pivot table - where we can chose any two fields and plot one on the X-axis and one on the Y-axis. By default, I've got custodian and extension populated but I could come in and say I want the custodian to be in my rows and I want issue tags to be in my columns. What I'm doing right now is running a quick table to say - here are the custodians in my case and here are my issues tags. 

I can now start to see a high level overview or breakdown of all the tagging that has happened so far by custodian. As an example, I could see Richard Sanders and notice he was not involved with LJM, but on the other hand Louise Kitchen seems to be highly involved with LJM or Jedi. I could take any of these results that we see here, click on a few cells and I could add those to a search. Or I could come in here and I could take all of these results and export them to a CSV format. 

There is some very nice functionality around the Heatmap and it really allows you to go in and find those key documents you're interested in.  

For more, see: A New Data Visualization Tool

Searching

On top of all the wonderful AI technology that I've shown you today, we have built in some very powerful search capabilities that I want to highlight for you. When we started to create Reveal 11, we worked very hard on adding in some excellent quality of life features with our search as well as make sure that our search is very fast.

So when I come in, I want to showcase some features to you. First, I'm going to run a search on Raptor and I'm going to run a second search on Raptor. I want to show you all a feature. One of the key differentiators that we have in our search engine is we are able to group documents down - such as families, threads, or duplicates on an individual conditional level.

This means I could say on this keyword of Raptor, I want to include the family and I could say and not included the word  raptor. If I wanted to bring in the responsiveness tag, I could grab that and say let's bring in all the responsiveness. Now I see that I've built a search running a family grouping on the keyword Raptor and also not the word raptor and maybe on a certain condition I want to bring in threads. Why is this important? 

In many other technologies you can only group an entire search - on one grouping criteria such as family and then you have to build out a nested search that references that search over and over again. We can build all of these searches on an individual conditional level to stipulate I want families on this and I want threads on this criteria. 

One thing I want to highlight, we will always show you real time. How many hits am I going to get if I wanted to then include let's say family members unresponsive, we'll notice that this number is going to update real time.

Very, very powerful and helping you to get a sense of am I on the right track with what I'm doing so far.

For more, see: RQL/Elastic Query or Speed of Searching

Term Lists

Another very powerful technology that we have in here is the term list. The term list allows you to come in and if I wanted to paste 50 terms I could or start building this myself and quickly add in terms as I see fit. When I flip this over to our table view - our clients are really liking this feature because it allows them to start getting a sense of how many hits they are going to get on their terms in real time - and is very interactive. 

I quickly type in monitor and I see with a click of a button that I've got about 1,400 doc hits and 3,500 family members. If I click execute on weather, I can see how many document hits and families there are. maybe monitor is a little bit more inclusive than I want - I could come in and I could say let's do monitor within two of gas. I can quickly filter on that and now I am down to that subset of 40 doc hits and 120 or so family members. From here I could continue to add new terms if I wanted. 

One of the things we did when we architected our search engine is that we created a query language that allows us to build out a query that contains both document text and metadata - simultaneously. 

So, this query I'm about to show you right here, this is highlighting that I am running a search that is saying: Bring me back the author field that contains Vince or Sally and I do not want a file extension of Excel or PDF and I want that combined with this complex search of profit or derivatives, etc. Keep in mind this is a very complex search and we're running this across 40,000 documents and when I hit execute, we see how quickly the search engine comes back with the results. At any point, I could now take these terms that I've added and I could save them as a word list report where I could folder each term individually or I could export them out as CSV. Or when I click add to search, we now see - here's how many results I'm going to get on this search. Maybe I want to include the family and in real time the search engine will tell you how many hits you are going to prior to actually running the search. 

These are some excellent quality of life features that we see in our search engine which is extremely fast. You can start building out these term lists in real time at the onset of a case and you can come in here at start grouping all of the conditions of your search on an individual conditional level like we see here. Very intuitive, very easy to use. 

AI Tagging

I've shown the Effortless AI. I've shown our ability to quickly use the feed within our environment. Now I want to show you what we call some of our Review Enhancers.

So, when I come in and I'm looking at some of these key documents, we have integrated in some very powerful technology. The first thing I will highlight as we have integrated in CAL (Continuous Active Learning) technology into our software and it is an extremely easy to use.

You'll notice in the top right corner here I've got a responsiveness field and then the AI tag and the number 14. What that is saying is that we have enabled this responsiveness tag, to be enabled for AI. Once you're reviewers have gone through and tagged just 10 documents, we start feeding that over to the AI engine. The AI engine starts learning based on the decisions that your reviewers have made and it kicks out a probability score of 1 to 100 of the likelihood of that document being responsive or maybe Chewco, Jedi, Ljm, and Raptor.

Couple of notes about this - you'll notice we support multiple AI tags on this document we're looking at right now. We actually have five AI tags right and we can support these on a multi choice tab as well.

Additionally, the score that you see here that shows the probability, that can be hidden based on permissions. So if you do not want your reviewers to be seeing what the score that the AI is tagged, that absolutely can be locked down to the appropriate people. We don't want to be influencing anyone's decisions as they're going through the process, but what is very nice as you go through this is we will then come and populate a document metadata field with these scores.

So, I see Reveal AI Score responsiveness, I see Reveal AI Score Chewco. Maybe I would want to come in and say let's find all the documents that have a responsiveness score of 90 to 100. I can quickly add that into a search and I get those subset of documents and away we go. Now where this is really valuable, we have seen people build out auto-batching rules related to batching out any document that comes back AI Responsiveness 80 and above - assign that out automatically to family.

So, it is very easy to go in start building in some nice workflows related to the CAL workflows that we have enabled within this environment.

Review Enhancers

Review - Foreign Language

Next, I want to highlight what we call Review Enhancers and within our technology we have a built in translation engine. When we process data, we will find the composition of the languages contained within that document.

For example, on this document we see both English and Japanese.

• What that means is I can then run and I can say - Let's assume these four documents are in Japanese, I could grab them all and I could say I want to translate these documents.
• I simply choose my source language. In this case I know it would be Japanese. I would then say what do I want my destination language to be in this case, English.
• With the click of a button, I now see the native file that contains that Japanese language, and I can say I want to translate that into the translation field. And there we go.

This is great, right? No need to do an export, go to a third party, and reimport. We can do all of this directly within the tool itself. We don't now need to bring in any reviewers that speak a different language. It all can be contained directly within our tool.

One additional note, we support over 100 different languages, which means you include Chinese or we can handle Farsi - any of the languages that you see on this list, we can actually translate the UI of the software into that language as well.

So if we ever had a foreign language review where someone was based in Germany and they needed their reviewers to be able to review in German, we could flip the UI into that as well.

Very nice to have no more need to do any export to a third party. Reimport all can be done directly within the tool.

Review – Image Labelling

The second feature I'll highlight is what we call image labeling.

Image labeling, the analogy I always use to describe this is think of a Google Captcha where you go through, and it says click on all the bicycles or the stoplights that you see. Well, what you're doing when you click on that, is you're training an AI neural network to learn what a bike or a stoplight looks like.

What we can do is utilize that AI technology and run that across your images and essentially the AI will tag your images with labels, thus making them searchable. So as an example, when we look at this image, I see a woman, I see tech, I see a passport - here is what the AI identified on those images. Now you can see it has labelled: passport, driver’s license, document, or text.

At the onset of a project, you could come in and say I want to run a search on the image labeled field that we see right here and bring me back all of the images that contain a passport. This allows you to start quickly searching across your images, which can save a ton of time.

We've seen a bunch of different use cases for that. It is extremely powerful in construction cases - helping find bridges, pipeline, black mold, soil. It is also very good at finding things like passports, drivers licenses, or handwriting.

Another very used feature with this is looking at this on logos and trademarks. So, you get a lot of those e-mail footer, junk images that get brought in. You could do a search on logos and trademarks, grab all of them, throw them into a folder, deprioritized them, tag them non-responsive – up to you. Very powerful at helping you search across your images to really start narrowing down on the documents quickly.

For more, see: AI Image Recognition or Image Recognition during Legal Review

Review - Transcription

And the final feature that I will highlight, we have a built-in transcription engine. Any audio or video file that you bring into the technology - we can come over and I'll just show you an example, we can hit transcribe.

We can choose from over 30 different languages and it does an excellent job of dialect detection - you can see Australian, British, Indian and Irish. We can choose the desired language and I would then run the transcription against this video. It will pull the text out of that video, index it, and make this searchable.

When I hit play on this, you won't hear it because it's been my headset right now, but it's going to highlight exactly what's being said. It will highlight that, and we could quickly build a search to say I want to find a key term. Let's say the key term we're looking for related to this is subpoena. I could build a quick search so they bring me back all the video files that contain the word subpoena and you'll notice I could quickly filter down to that key term I'm interested in. I could click on subpoena and watch the video, it is going to jump directly to that location.

If you're ever dealing with a case that deals with a lot of voicemails, phone calls, body-cam footage, there's been extremely powerful tool that can really help you start pulling the text out of it, make it all searchable, and allow you to jump directly to that location when the document, which can save your reviewers a lot of time.

Conclusion

With that, this has been the overview of Reveal 11.

As a reminder, we are differentiating a lot of ways, but three key features: 1) We are extremely fast 2) We provide you some excellent Review Enhancers to help with the tools in your toolkit 3) We provide you with Effortless AI where you are going to have access to extremely powerful AI technology at the onset of the project, which can really help you cull down on those documents and find the key rich data at the very beginning of your investigation, litigation, or legal matter.

Thank you for joining us today, I really appreciate your time. Thank you.