Welcome to ASAE's Elements Demo Days. I'm Larry Covert. I'm ASAE's Senior IT Director. I'm sitting in for Reggie Henry today. We're kicking off the day with EnigmaChat. And Thomas Wong is going to give us a great overview of that product. So I'm going to hand it over to him now and look forward to a great presentation. Thanks, Thomas. Alrighty. So let me start sharing my screen. All right. And so, Larry, can you let me know if you can see my screen? Looks good. All right, fantastic. So welcome, everybody, and thank you for spending your early morning session with me today. So I'm the first one in your long day, I guess, today. And today, I would like to introduce a slightly new or different type of learning management system than what are typically expected from a learning management system. So the product I'm going to introduce is called EnigmaChat. So just a quick introduction of who we are. So my name's Tom, and I'm with 360 Factor. And we have a learning management system called Oasis LMS. And we had it since 2010. And it's a product that is very similar to what probably a lot of other learning management system you will see today. So if you're interested in that, please let me know. I would like to focus on a new product that we're launching called EnigmaChat. And today's session, we're going to focus on really why we decided to launch a different product and how this product compare with a typical learning management system. So just to get everybody started on why we even ventured down a slightly different path than what has been tried and true in a learning management system. So about a year and a half ago, I think everybody is just as amazed as everybody at 360 when chatGPT launched their almost magical chatbot. So it's a chatbot that basically can answer everything and anything under the sun in basically a split second. So we believe that we are seeing a beginning of a slow but sure shift in how your learners are going to look for information. So slowly, you will see the user will no longer use, I won't say completely, but oftentimes they are going to start their search for questions by going directly to chatGPT instead of going to Google Search. And I can see this personally just within the two kids that I have who needs to do homeworks. And they definitely start using chatGPT as their go-to when they have questions on how to answer some of the homework questions instead of doing Google Search. And I always remind myself that whatever my kids are doing is what the new generations of learner will be doing 10 to 20 years from now. So you have to prepare for just a whole paradigm shift of how the user go about learning about things. And my suspicion actually is pretty much confirmed by a lot of surveys out there. One of the survey indicates that in the future, 42% of the learners were likely to search for their answers using an AI-powered chatbot instead of using a search engine. But then the question is, so exactly how Enigma is going to come into play? So to answer the question, I want to lay out how a traditional learning management system works and how we think it might change in the future. So if you put your mind in a learner, how they go about learn something right now. So they will probably go to Google and then they ask a question. And I'm using a hypothetical question about someone who need to ask about a prescription for a diabetes patient. So they will basically type something in Google. And the Google, to no one's surprise, would display a bunch of ads before it give you a list of links that will take you to possible resources. So once again, step one, you do a search. Step two is you have to mentally decide which link to click on. And the way you click on the link is step three, where you see the page that contains the information that you're looking for or not. You sort of have to go through that page to decide whether that page has the information that you're seeking. So translate that into a user journey within a learning management system, you can easily see how it's almost a one-to-one relationship. So inside the learning management system, the user will basically go to your LMS and they do a quick search with a question or keywords. And then a list of your education product will come back. So that'll be step one. And step two is where the user will basically look at all the possible product that they could purchase if you sell your courses. And then they will have to pay for that course. And once they're inside the course, they will be able to consume your content. So most likely, your content will be in the format of a video or a PDF. Sometimes the video might be 20 minutes long. Sometimes your video might be 60 minutes long, depending on how much you chop up a particular recording. And some of the PDF can go as long as 20 pages or 200 pages. So I think I'm not the first one to admit that sometimes the content are so long that I really didn't have the energy to read through all that content. But the bottom line, though, is the learner will consume all the content from your product and then potentially will claim some credit for continuing education if your organization offers credit. So this is a pretty standard user journey that we see is probably still going to be a try and true model moving forward. And that's really where our Oasis LMS, along with the other LMS out there, will continue to deliver on this model. Now, if we pivot a little bit into how at least I see how my kids are learning now, I want to translate that flow into a chatbot. So with a chat GPT, you can ask the exact same question. And notice the chat GPT will not waste any of your time. It will just go directly to the answer that will answer your question. So it doesn't really give you a bunch of links for you to decide which one to go to. It's not going to wait for you to synthesize the content on all the pages to decide what page makes sense, what page is relevant. It will basically synthesize everything and just give you the result, which I totally understand why my kids' generation prefers chat GPT because it's much more direct to get the answer in front of your learner instead of sending them down a wild goose chase. So if you translate that into how a chat-based LMS is going to work, so you can see that the user will continue to ask the question to your platform. But this time, they will see the answer right away. And ideally, the answer will be purely based on your content. So we're trying to eliminate hallucination. We're trying to eliminate any answers that are based on non-authoritative content that's on the internet. So the answer will be purely based on your content. And if the user want, they will be able to click on any reference links. And then they will be able to see any additional resources where the answer was based on. So these additional references might be things that's publicly available on your website, or it might be gated in a certain part of your learning management system that you can monetize by sending the user to. But the idea is now we flip the model of a learning management system where we're basically giving the answer to the user as soon as they ask the question. And then we can lead them to additional content in your LMS if they choose to. So at this point, if I have convinced you that a chat-based model is a much more direct way of doing education, you might start asking, well, then where does Enigma comes in? So just a couple of things. First is that ChatGPT or any other AI-powered chatbot is based on everything on the web, which may or may not be authoritative, whereas we want to create a chatbot that is using fenced data, which means it's only using your data. I'm not going to go out to the internet. So this is really, really good. If you have enough content that can be harnessed to answer basically any question within your domain. So if you don't have a lot of content, obviously this model is not going to work very well because all the questions will be met with, I don't know. But if you have enough content to cover every sort of knowledge space, knowledge within your space, then a fenced data approach is perfect. Also, you can't really monetize of ChatGPT, whereas with Enigma, you could monetize by either making this as a member-only benefit or provide the answer for free. But the content that's referenced by the answer will require a paywall. In the demo later today, I will actually remove all the friction so that we are not going to try to monetize on the user journey. But just keep in mind, it is possible to monetize. The third problem we're trying to solve over ChatGPT is that on the left-hand side, you can see that when the answer is provided to the user, there is no additional reference provided. So what we're trying to do is we're going to make sure not only is user given the answer to their question, but we're also going to provide additional references. Not only are we going to give additional references, we also want to prevent giving user a link to the beginning of a 200-page document or the beginning of a 60-minute video, because we believe that's not really helping the user. It's like dropping the user at the lobby of a hotel and say, OK, you try to figure out which room you're staying. So we are trying to say, not only am I going to give user a link to the resource, I'm also going to tell them exactly which chapter, which section, or which minute within that video addresses that the answer is based on. So this way, we're basically completely remove all the friction where the user need to find the education goat nugget that they're looking for. We're just going to go directly to the right position in your document or the right timestamp in your video. So that's really the true power of how we can add a layer of functionality over ChatGPT. So just a recap before we jump into the demo. So Enigma is a custom chatbot that can authenticate against your AMS or CRM for you to monetize. You will answer questions based on your content instead of the internet. You will use the same technology as ChatGPT so that you get the same sort of magicness that the people are getting from ChatGPT. And all the responses will basically are linked to your content. And the last one is optional, is you can track what the users are asking to help your education team figure out what people are confused about. And if, let's say, they ask a lot of questions on a particular subject that you don't have a lot of content over, that will trigger something in your mind to say, hey, you know what, we probably should develop some webinars and such for that particular topic. So hopefully, I spend the last about 10, 15 minutes to describe the motivation of why we decided to embark on this Enigma product. So now what I'm going to do is I'm going to show everyone the product in action. All right. So I'm going to open the, there we go. And Larry, help me out. Can you guys see the browser that I just launched? Yes, you can. All right, fantastic. So the demo I'm going to show today is we're enabling, I'm going to go to the site. There we go. So it's a pilot program that we have set up with ACC, American College of Cardiologists. And what we're trying to do is we're trying to accomplish two things. First is, everyone sees here, this is their guideline page. And their guideline essentially goes all the way to 2008. So that's over the 16 years of guidelines. And the problem they have with other guidelines, I won't say the problem, but one of the friction is that, I'm just going to click on one of these to show you. Their guidelines are very extensive. And you can see I can keep scrolling, and this thing scrolls forever. As a matter of fact, it's faster for me to use a scroll bar on the right-hand side to scroll it. So I don't know if this guideline is common within your organization. But just imagine if your content are in this kind of a web-based format or a PDF format that has multiple pages. It's not very conducive for your learners when you say, hey, the question you're asking is answered within this page. Because they will have to say, well, which page on this 200-page document is my question answered? So that's one problem we're trying to solve. Here's a second problem we're trying to solve. So this is one of ACC's product. I just want to give everyone a sneak peek of Oasis, even though this is not a demo of Oasis. So here's a product that has a lot of video content, probably hundreds, if not thousands of videos. So the user will be able to play the video. So I'm not gonna play the whole thing, but you can see it's a 20 minute video, right? So 20 minute video is already sort of sliced and diced into smaller chunks, but your learners still probably don't want to sit through a 20 minute video to find where in the video is their question answered. So those are the two problem we're trying to solve, right? Is not only are we trying to sort of digest all the guidelines to help the user answer the question based on the guidelines, we're also trying to digest all the videos, all the content within your LMS to answer the question based on all the content you have in the LMS. So what I'm gonna do is I'm gonna come over here. So this is the Enigma chat, and it's a very similar user interface as any sort of chat GPT or Bing chat interface that your user probably is already familiar with. So you can see that earlier today, I was trying to do a few dry runs with my basic high school biology questions, which is not gonna be very serious questions. But what I will do is I will actually open up just a few questions that are more real question within our cardiovascular field, right? So I'm just gonna copy this question, and I'm gonna start a new chat. So for the first chat, I'm gonna ask just the guidelines, right? So keep in mind that I could check both knowledge space, and it will basically ask for both knowledge space, but I want to sort of show everyone how it works if I just check one. So as you can see, I asked a question, and the chatbot basically answer the question based on all the guidelines that was published. Now keep in mind, the guideline is, we're talking about 16 years of guidelines that has been digested into the chatbot, and the chatbot was able to just quickly answer them. So that's the first step, is I'm answering user's question based on the fenced data. The second thing is I'm just gonna click on here to show you that not only do I give users the response, I will also give user additional reference where they can click to see sort of how the answer was based on, right? So I'm just gonna click on this one. So in this particular case, I'm just gonna click on the link to go to the actual guideline that answers this question. But notice the first thing is that I'm going to get dropped at the URL, and this URL has quite a bit of content. So the second problem we're trying to solve is, a second problem we have solved is sending user to additional reference, but we don't want to drop off the user just at the lobby of the hotel, so to speak, right? We want to sort of lead the user to the right floor in the right room, right? So we're gonna say, okay, go to this section. I'm just gonna quickly copy the section, and I'm just gonna find it. So that's the section, and then I'm gonna find the subsection. There we go. So now the user will understand that the content over here actually is what the content in this section is where the answer was based on. So you can see how it is a much, much more direct way to get the user sort of find the needle in the haystack by getting to exactly where within 16 years of guidelines, I can drop you at the exact place, right? And keep in mind what Enigma does is you actually sit through all the guidelines, and we don't just pick the best match. What we actually do is we pick a few top matches, and we sort of digest all the best matches to create the answer, and then we basically let the user know, here's a few best matches. So the user still have the last say in exactly what the answer is by going visiting all the top matches for the reference. So this is sort of how OASIS help the user find content that is on a website. Now I'm gonna do another new chat, and this time I'm gonna ask a different question. Oh, hold on. I'm gonna ask this question right here, and this time I'm gonna ask just within all the video contents I have. So just give it a second. Keep in mind what this is doing is Enigma is going through thousands of hours of content, video content, going through the closed caption, and then basically answer the question based on what was actually said in a few videos within your vast library of video content. And now what I'm gonna do is I'm just gonna click on this to show you here is the video from your vast library of content that answers the user's question. Not only that, I'm going to click on the play to show you that not only do I display the video that answers the question, I actually take the user to 16 minute and the 34 second mark. And that is where in the video the speaker is answering the question that the user have asked. So keep in mind, this is a sort of a very much a different way of a learning management system where we're not asking the user to sit through the entire video to then decide, okay, which part of the video actually answers my question. We are taking the user directly to where the exact time, exact minute and second within the speech. But with that said, context is everything, right? So the user can definitely sort of a scroll back in time to see what the speaker was talking about right before. And then they can definitely go further to say, okay, what did he say in the next sort of a time slot, right? But the idea is we still try our best to get the user to the exact location where their question is being answered. And then they can do a search and so on and so forth, or they can just scroll back in the video, okay? So with that said, I'm just going to jump back to my slide really quick. So here's a few screenshots of the technology that I was showing. And, oh, that is actually the end of my slide. So I will stop sharing, all right? So actually I have eight minutes left. So I would like to just open it up to anyone who have any questions. So, okay, all right. So, okay, so the website, just so everyone sees, you can definitely reach out to me if you have additional information at the, I'm just going to share my screen really quick again. So give me one second. So my email is at thomas.wang at 360factor.com. So you can definitely reach out to me directly for any additional information. If you also want to check out our product, our Oasis is at the oasislms.com. And the Enigma chat is at the enigma-chat.com. Keep in mind right now, Enigma chat is in what we refer to as a beta testing. The reason we want to do a lot of beta testing is because we are working with existing clients who we have existing relationship with because they're using our Oasis to really do two things. Number one is other content already live in our Oasis. So it's very easy for us to just digest all that content to populate the knowledge for the chatbot. That's number one. And number two is just because we already have an existing relationship, we want to leverage that to test the chatbot because in order to make sure that we launch this product successfully, we want to make sure that the answer that's coming back is accurate. And every knowledge space is going to have a little bit nuance to it on how their learners are going to expect their answers to be. So I'll give you an example, right? So one thing we discovered is that in medical knowledge space, there is this idea of level of evidence or whether it's a strong evidence or weak evidence. So when you answer the question, it is important to cite whether the answer is have a strong evidence or weak evidence. Those are some things that we as a technology platform does not actually know when we initially launched this product. But during the pilot phase, we are learning from our client on what works, what doesn't work and how to fine tune this product so that we're going through this pilot phase to make sure that when we eventually launch, the chatbot actually will answer the way that we want to. So that's actually what we're currently working on right now. So, and to answer Keira's question, no, you do not need Oasis to run Enigma chat. It's just easier for us to get existing Oasis LMS client launched on Enigma chat because we already have their data, which is populated much easier. If you don't use Oasis, we just have to load your content manually instead of point and click and sync it over. So, okay. Gabrielle, thank you so much for sharing my email there. And to answer Alice's question about what it takes to set it up. So I know that I don't have a lot of time to set it up, but what I will do is I'm just going to show everyone the back admin site so that you can get a sense what you will require to set it up. So give me one second. I'm just going to log in as a support. So give me one second. So you can see that in this particular instance, right? So I have two knowledge-based F guidelines and I also have ACC SAP, right? And I just click on guidelines. You can see here is everything that has been loaded. So each line is what's referred to as like a knowledge, like a vector or like a chunk. So basically the idea is you have to import the data. So if you have your content already, we can basically import it using Excel file. And then what happens is you import the data and then Enigma would digest your data into a computational format so that you can do somatic matching, so that it can be used to answer users' questions. So keep in mind the system right now, we basically allow you to import the data, but we use Excel to import your data. And then if your data is in a web format, we'll basically, we'll take the data from your website, load it into Excel and then use Excel to load into Enigma. So that is the process right now to get your knowledge into Enigma to digest. So to answer Deborah's question, you do need transcript for the video in order for the chatbot to understand your video. Most of the LMS these days automatically generate transcript including our Oasis. But if you don't have a transcript for your videos, let us know, there are many different ways to get transcript of your video. So, and Alice, okay, so let me just make sure. So yes, you can import in bulk using Excel for sure. And Janice said, will this feedback loop be part of the final solution? Yes, so this is the product still in beta. So we're still trying to balance the user's privacy versus what analytic we want to extract from the system. But the goal is to make sure that what the user is asking and what the chat engine is responding and how the user rate the response are all gonna create a feedback loop so that your subject matter expert can review to say, hey, you know what? Some of the answers are getting a lot of downvotes. Maybe it's because the content we have is a little bit more out of date or a little bit more controversial than we thought they are and that you can sort of start revise your content. And what's nice about Enigma is you can definitely take a chunk of your knowledge and then move it out of the chatbot and you can shift some new content in and then the chatbot will basically operate with whatever is in the chatbot and it doesn't have a memory of what used to be in the chatbot. So to answer Samantha's question. Okay, so to answer Samantha's question. So the idea here is that you can either make the chat a member benefit or it can be a paid product but the overall chatbot will be sort of a, you know, either a paid product or a free product. Then the reference the user links to is gonna be, could potentially be a paid product. So those are the two type of monetization model that we are providing within the platform is either make the chatbot paid or a free member benefit and, or make the content where you link the user to as a part of the paid benefit. So, and to answer Alice's question, there is no limit to amount of data that you can host. So. Great. Well, that's our time, Tom. Thank you so much. We appreciate it. We hope everyone enjoyed the presentation. I sure did. So we hope everyone will join us in about 10 minutes for the crowd wisdom presentation.

EnigmaChat

chatbot

learning management systems

curated content

customizable

monetization

continuous improvement