EP 169: D-ID Agents - The next era of customer experience?

This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.

What does the future of customer experience look like, especially with generative AI? I think it's going to change a lot. And yeah, sometimes I go on the show and I talk about my thoughts and my feelings. But today we're going to be talking about not just new technology, but even some technology that's getting announced today that I think is going to greatly shape the future of how we interact with brands, products and services that we love. So we're going to talk about that today and more on Everyday AI. This is your daily livestream podcast and free daily newsletter helping everyday people like you and me network not just learn what's going on in the world of generative AI, but how we can all actually leverage it to grow our companies and to grow our careers. And we're going to have some very practical and actionable use cases of that today. I'm very excited to have the head of Content and Creative Marketing for did, I'd say one of the most creative and unique players in the generative AI space. But before we get to that, let's start as we always do, with going over what's happening in the world of AI news. So the Biden administration is one step closer toward writing guidelines for AI policy. So the National Institute of Standards and Technology, the nist, announced yesterday that it is seeking input from AI companies and the public on generative AI risk management in reducing AI generated misinformation. So Biden's October Executive Order on AI prompted this effort to develop industry standards and enable America to continue leading and responsible development and use of artificial intelligence. Also, the use of external red teaming, which involves simulating potential risks via red teams, is being explored as a tool for identifying novel AI risks. Next, a new study is showing AI could add billions of dollars to the racial wealth gap in the US so a new report from McKinsey showed that generative AI could add up to $43 $43 billion annually to the racial wealth gap and automate half of jobs that don't require a four year degree. This also showed that obviously black workers are at a higher risk of job loss and face systematic biases in accessing housing, credit and jobs due to AI. So make sure to check out the the Daily Newsletter for reading more on that story. Also, last but not least, AI cannot be named an inventor. And that's according to a new ruling from a UK Court. This decision reflects an ongoing debate around the rights and protections of machines as AI continues to advance in various fields. The court reaffirms that only natural persons can be considered inventors under the UK Patents act of 1977. So the AI expert that's kind of at the heart of this case, we've talked about him on the show before. Stephen Thaler has made similar arguments in other countries with some successes in South Africa and Australia. But in his case, in this case, and he has been rejected and also by the US Supreme Court. So we always have way more going on in the world of AI news. So make sure if you haven't already, go to your everydayai.com and sign up for that free daily newsletter. But you're probably tuning in today to hear about the next era of customer experience. So I'm extremely excited to bring someone on that's really helping build that and shape that in the industry today. So please, without further ado, help me. Welcome to the show. Ron Friedman, the head of content and creative marketing at did. Ron, thank you so much for joining the show.

Pleasure to be here, Jordan. Thanks for, thanks for having me on. And let me just assure you and the viewers I'm an actual person, not an avatar. Although we'll be speaking a lot about avatars, I guess, during this conversation.

That's a great point, Ron, because, you know, in theory we probably could have done this avatar to avatar, right? And maybe that's the future. But before we even dive into that, but maybe just give everyone. For those that aren't familiar, can you just tell us a little bit about did what the product is, what it does. And for those of you joining the live stream, I'm going to be throwing up some different things from their website. But if you're listening on the podcast, don't worry, you can always come back and look at the show notes and watch. But Ron, maybe just walk us a little bit through what did is and what it does.

My pleasure. Yeah. So did is a tech company. We've been in generative AI since 2017, before the term generated AI was even a thing. You know, kind of just, just that in itself. You know, I remember about probably a year and a half ago, you know, the concept was out right mid journey had just come out and the stable diffusion and ChatGPT was around the corner and it was like, what are we calling this technology? Synthetic media? Digital media? And then I remember the headline. The New York Times called it Generative AI's coming out party. And that was it. Once the Gray lady, as the New York Times is known, kind of gave the industry a name. That's it. We became a generative AI company. But in truth, that's something we've been doing since 2017. What we do, okay, those of you who are watching, you can see on the screen we've got images of people. And what our technology really does is take those images and through text to speech, turns them into videos of talking avatars. Again, those who are viewing the site, you can see there's several ways that we do it. There's several types of avatars. I'm guessing people who are following the AI space and familiar come across these. You've seen them in kind of mashup videos, popular culture. We've seen them definitely in learning and development areas, but now we're seeing it more and more in business use cases and particularly customer experience, which we're going to be talking about today. The studio was launched just over a year ago and basically what we allow or what we enable is anybody who wants go to studio.did, sign in. It's a free trial. You can test out the platform and you can create your own avatar or choose one of the existing ones, decide what you want it to say, and within a minute you've got a video of that digital person speaking the text you want. You know, some exciting, exciting aspects of it is that you know these avatars can speak any languages, right? We have like 120 different languages in, in the studio you can choose from. So you enter a text in Spanish, the avatar will speak Spanish, you enter a text in Arabic, the avatar will speak Arabic, etc. A wide variety of appearances. Right. So there's an existing bank of avatars where it's of kind characters of every age, race, skin color, nationality. But what's even cooler is that you can create your own avatar based on a single image. So in addition to the existing bank, which is very rich, you can upload an image. It can be of yourself, it can be of a friend of yours, it can be an image you created in an image generator. As long as the software can recognize a human face, we can take that and make it speak. Sorry, we get this often. We can't do animals unless they're very human looking animals. The algorithm just doesn't recognize the face patterns and we can't get them to talk.

Yeah, you know Ron, maybe as we get a little bit more into DID and agents, I do want to talk a little bit about because I'm sure people have this question on, like, okay, this sounds like a, like a fun technology, really cool. But I mean, there's actual, like, business use cases for using this technology. Can you talk a little bit and just. Even just about the technology at large? Right. Because obviously you have different products, different offerings at did, but what are some actual business use cases for technology like this?

Right? So at its very core, it's. It's taking an image and making it speak. I think the main pain point that we're trying to solve is production. Right. You're in media. Remember you saying that you used to be in journalism. Same here. We know what it costs to set up a production, right. You need camera, crew, lighting, sound, design, makeup, talent, people to deliver the message. If you're creating an online course or a marketing presentation or a product walkthrough, all these business needs for video. So we make that very easy to produce again within minutes and in multiple languages. So it can suit any audience. So we know that's one. It's just cutting costs. The ability to animate any image kind of adds to that because suddenly you don't need a human to do it. You can, you can have an alien deliver the presentation, you can have a robot deliver the presentation. Again, as long as the software can recognize the face, we can make it speak. So a lot of cost savings there and a lot of flexibility. Right. You can, within minutes, translate your presentation from English to Japanese. If you have clients there, you can change the setting. You can embed these videos kind of in any video content that you want. And that's just the video part before we've even started talking about the interactive part where agents come in.

Yeah, and maybe let's start to talk about this. And hey, as a reminder for those of you joining us live, live, what do you want to know, you know, from. From Ron, and just about the future of, of how we use, you know, generative AI avatars in business. This is a great opportunity to, to get those questions answered. But, Ron, maybe, maybe just talk us a little bit through about what an AI agent is, especially how you're. You're creating them at DID Because I think there's obviously differences. And, you know, even when you talk about AI agents, it can mean many different things. So in this use case with did, what is an agent?

All right, so I'm sure your audience is familiar with GPTs, custom GPTs. So it's very similar to that. Basically, when you go to the studio, you create an agent, you choose the appearance, and again, you can either choose from the existing ones, or upload an image of your own. You choose a name for it, you give it a selective voice and many, many voices, natural sounding voices, synthetic voices, you can go through for hours just listening. All right, that's exactly the one I want, but that's where it gets interesting. So you created the avatar that you're going to speak to, but now it's the time where you want to start educating, so to speak. So there's a process of prompting. You tell the avatar how it should function. So for example, if I want a technical support agent, then I'll tell it, you are a technical support agent for Acme Co. Or any company you like. This is the product that you're an expert on. This is the way you should answer. You should be friendly, you should be outgoing, you should be pleasant and cheerful. You should be inquisitive, kind of describe what it should talk about, what it shouldn't talk about. If you don't know the answer, what you should do a pretty basic short prompt. And then the next stage is you want to start feeding it knowledge. And that's where the similarities of GPTs come in, because you can upload PDF documents, you can upload text, you can upload PowerPoint presentations. We're still in beta mode, but soon you'll also be able to just include a URL and that provides the knowledge base that your agent is going to respond on. It's a very quick and easy four or five minute process. Click generate and that's it. You've got an agent that is an expert on the knowledge base that you've given and behaves as you requested it to. And that's it, you're off to the race. You can start chatting with it now. We all know what it's like to chat with ChatGPT or other large language models here. The difference is you're actually having a conversation with another digital person, right? So you speak to it as I'm speaking to you and then it answers you also in its voice and with the face moving kind of resembling as much as possible today, a face to face conversation with a different human.

Yeah, technology is wild, y'.

All.

Like I'm, I'm so excited to, to be able to experience this a little more already. A great question from Mauricio coming in. So asking can people use this did agent to act as an example, as an influencer or content creator, to take over an Instagram account, or maybe even just saying like, can, can people use this to create kind of an avatar for their brand? That can, yeah. Act for, for them on social media.

Absolutely. Jordan, I urge you, take all the transcripts from your last 170 shows, feed them into DID Agent, upload an image of yourself. You can also obviously clone voice and you know, prompt it to answer as Jordan Wilson with your knowledge base. And I'm not saying it's going to be a replacement, you're one of a kind of. But you can put that up on your website and have your users ask you questions, engage with you. And again, the better the quality of the prompt is and the better the knowledge base that you're giving it, the more it's going to resemble an actual person. So influencers we're talking about, service providers can be a digital sales representative, can be an insurance agent, it can be a, a fashion advisor. And again, the sky's the limit. It's up to your imagination what you feed into it in terms of the knowledge base and how you want it to behave. You know, there's the really. The sky's the limit and you know, it opens up a lot of exciting opportunities.

You know, Ron, I'm, I'm looking forward to what happens next, right, because when you think about the, the concept of, you know, chatting with someone that's not human, right, it goes back a long ways, right. Like intercom I think, came out in, you know, 2011. So you know, there's been chat bots and things like that for, for a decade now. Is this where the industry as a whole may be headed? So instead of even, you know, talking with, you know, smart AI bots in the future, will we just be talking with AI avatars in video in real time?

Well, you know, kind of goes to a larger question and the way we like to frame it is, you know, we as human beings are literally hardwired to respond to other faces, to see people's facial expressions, body length. That's, that's the way, you know, since we came down from the trees, right, We've, well, we've interacted with other people face to face conversations. But when we're interacting or engaging with technology, up until now we've always been doing it on the technology's terms, right? So if it used to be, you know, we talk about the, the evolution from textual user interface to, which is, as a child of the 80s, I remember, you know, those great green screens where all you had is a command line and you're typing in and that's the way you communicate. Then we evolved, I guess in the 80s 90s to GUI the graphical user interface, which is basically what we're using now. It's mouse and keyboard and windows and drag and drop and scroll. But that's still communicating or interacting with computers, machines, anything digital on their terms or on its terms. What we believe is that the technology AI, large language models, natural language processing, natural language understanding really opens the door to a new way to interface with these devices and one that's more suited to us as people. So we see tui, textual user interface, gui, graphical user interface into nui, that's a natural user interface. And agents is kind of the first step towards that. We really believe that you're going to be encountering this type of interface everywhere, whether it's on your phone, your small screens, whether it's your big screen in the living room, whether it's kiosks at the supermarket. You'll just be speaking to digital people, digital humans, avatars, everywhere you go.

Yeah, I like, I like this from Ben saying it sounds cool, like an idea, like an ideal version of, you know, Clippy from, you know, 20 years ago from, from my.

Yeah, we can't, we can't animate inanimate objects. But no. And yep, there's a question. When, when is it going to be? When is it going to be appropriate? Because, you know, there are instances where it's natural and then you want to be speaking to another person, bouncing off their expressions, reactions, stuff like that. But sometimes, no, you just want it in your, in your ear or you want it, you know, you want to type or you want. So it's not suitable for all solutions, but we definitely think it's a step forward. And think when you're connecting that avatar to different systems in the background, it's your operating systems, it's your shopping experience, really. Again, we're a platform company. We're providing the basic infrastructure to do that. But like we saw with the product Creative values view, it's really difficult to predict how users are gonna take advantage of it, what kind of creative ideas they're gonna have with it. And, you know, we're here looking at what everybody is, is working, building. It's an exciting time. Yeah.

And hey, as a reminder, if you're joining us midway through the conversation, we have Ron Friedman, the head of content and creative marketing at did, I'd say one of the most creative, creative and unique AI companies out there. Ron, I have a question. So when we, like, look at the actual application of this, you know, broadly, are there cons or concerns? You know, and, and I'm not talking like you know, technical obstacles, you know, that, that maybe DID is facing as a company. But, you know, when we look at the, the larger issue of, you know, AI agents and, you know, being able to interact with something that feels human, I mean, are there any things that the industry has to keep an eye on in terms of ethics, in terms of use cases, like, what are those areas that we should all still be keeping in mind and not just racing to implement this anywhere and everywhere we can?

Right? Well, yeah, of course there's ethical concerns and I think everybody who's in the AI space is aware of that or should be, at least. You know, you started off the conversation, the news today with new regulations coming out of the Biden administration in Europe. And I think the industry or the technology is so far ahead of the regulation that we, in some ways we have to police ourselves or put the brakes on ourselves because the technology is so powerful and look like every tool, you know, the product or the result is as good as the person using it. But that's, you know, that's not a cop out or a line. We do take proactive steps to prevent misuse. You know, we have guardrails on, preventing uploading images of celebrities or politicians because we don't want to be used for nefarious purposes to fool anybody or you can imagine some of the unwanted consequences of these capabilities. And we're making sure that our users are abiding by our pretty strict terms and conditions. So it's not to misuse it. But, you know, anything that comes from either kind of industry organization alliances, which we're part of the government, I think is really important. I think people should be aware when they're engaging with AI. So, for example, in did every video that comes out has a watermark. People complain about it, say, hey, you know, I want to put out a great video. And you've got that DID logo or you got an AI watermark on the bottom of the screen and kind of, I don't want it for us, that's always going to be in place. We, you know, it's important for us to know that the videos that are coming out of our studio are branded or are marked as AI So that again, it's not misleading. There's a bunch of, a bunch of steps that we're taking to kind of prevent misuse. But you know, things that come out from governments, from industry alliances are very welcome as well.

Yeah, that's, that's a great point because we talked about on the, on the show yesterday, I think there was some A piece of AI news that the former Pakistan leader there was in prison and used, you know, an AI version of, of himself to have a, have a rally, so to speak. So I'm glad you bring up that point because yeah, I think people are always thinking about when it comes to AI, you know, things like deep fakes or things like, you know, we're coming.

Into an election year in the States.

Exactly, exactly. Yeah, yeah, yeah. You know, maybe Ron, I, because I think that there's so many like very practical business use cases for this technology.

Right.

Like we talked about earlier, you know, explainer videos, you know, training, a way to make training a little more personable and maybe even having, you know, the person that's normally leading training, you know, leading it, you know, corporate communications. But maybe what are some more creative or out of the box business use cases that, that you all have seen so far that maybe you weren't necessarily building AI agents or avatars for that reason, but you're like, oh, that's, that's pretty creative. That, that's really helpful.

Well, look, the product is very new. We launched it a few weeks ago. So industries are, tend to, tend to, especially big enterprise, tend to be slower to adopt. But I can give you an example where you know, we have a client that's food company, they specialize in Mexican cuisine and they, you know, they took an image of the grandmother of the CEO and folks can try it out. It's called Abuela AI and basically they trained that model on recipes and culinary expertise all around Mexican or Latin American cuisine. And that's it. You can ask Abuela, that agent, anything you want. What's. What. What beverages go well? There we go. What beverages go well with, you know, my, my whatever my recipe is or you know what. Yeah, let's. Okay, so for, for everybody who's not viewing Jordan is at an Abuela now and he's asking what beverages go well with my Mexican. I don't know.

Yeah, I don't even know what to put there. Maybe with my spicy taco recipe.

Right. Okay.

All right, let's, let's see this live. So, so this is a really cool use case. So yeah, if you're listening on the podcast on the left hand side there is Wayla there, right. And then I'll be able to hit enter and presumably, you know, I guess it's going to talk to me, right?

Yep. Interesting for a second.

We'll see. Yeah, we'll see. So yeah, I'm asking what beverages go well with my spicy taco recipe Offer.

Spicy taco for a refreshing contrast. Try a traditional Mexican beverage like a cold horchata or a crisp, citrusy margarita.

I love it. Yeah. And hopefully everyone can hear that out there. But that was a really cool, you know, showcase in real time of, of something that you could use, you know, for a Mexican restaurant. If people want to really build community. Right. Or to give their customers just a unique experience, that's a great use case there.

Are you still running in circles trying to figure out how to actually grow.

Your business with AI?

Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Gen AI. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us BE because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for ChatGPT training for thousands or just need help building your front end AI strategy, you can partner with us too.

Just like some of the biggest companies in the world do.

Go to your everydayai.com partner to get in contact with our team or your you can just click on the partner section of our website. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on Genai.

Thanks. Yeah, yeah, no, and I'm, you know, I'm sure we're going to be seeing this type of use case and many more to come. Again, a lot of the times it's just putting a product out there and letting the business world or also individual users kind of see the value and take it and adopt it to their own purposes. And like I said, this is an exciting time. We're eager to see what people make from it. Like I said, we're still in beta mode, so closed, limited release. But pretty soon I anticipate we're going to be seeing a lot of these agents everywhere.

Yeah, and speaking of, you know, business use case, I'm looking at the time here and I'm like, hey, we can technically still still break a little news by a couple of minutes here, Ron, because yeah, I do believe you have a tool that you're going to be announcing in literally two minutes. So what's kind of launching today specifically with emails?

All right, so we're DID and partner companies called Living AI created a plugin, an integration with DID's technology that relies on DID's API for sending personalized video emails. So if you're a business, you want to send an email to 100,000 different people, but you want it to be personalized. You want it to not just be, you know, a single generic video. You want, hey, Jordan, this message is for you. Or, you know, if you're in X company, this is the perfect solution. So now with AI, you can create these videos, personalized videos you can send by email and you know, very simply can integrate with any, any email service provider, send it out to as many recipients as you want, and it'll have that customization built in. Yeah, that's fresh off the presses for you.

I'm glad, I'm glad we got it in just, just a minute before the press release. Press release was released, right. Like the old journalist, right? We're like, all right, let's, let's, let's get this news out there, right? I like this comment from, from Megan because this is something I wasn't even necessarily, necessarily thinking of, but just a different way to learn. So Megan saying this could be great for young education, talking to historical figures. You're learning about. Yeah, maybe Ron talk. Can, can you talk a little bit about that? Because, you know, like you said, you can't upload photos of, you know, of a famous celebrity and something like that. But I'm guessing you can, you know, create, you know, maybe in mid journey or something like that, kind of generic era photos, you know, certain eras in history and then be able to learn what are some use cases for just learning by talking to an agent.

So education is one of our favorite use cases. So right now with the studio even, you can create videos, you can have ChatGPT, generate a script talking about Charles Darwin talking about evolution or Albert Einstein talking about relativity or a fictional character. Right. I want Bilbo Baggins to tell me the story of the Hobbit through his own eyes. So you generate a script. You can generate an image of any one of those characters. And again, within seconds, we can have Bilbo delivering his life story in his own image first person. But when you combine that with the capabilities of agents, it becomes even more interesting because you can take those same characters. I don't know, let's stick with the Hobbit. You can upload the book or upload anything that was written, documented about that character. Again, create an avatar based on their design. And there you have it. Bilbo, you can, can go back and forth for as long as you want with Bilbo Baggins or Charles Darwin or historical figures, fictional characters. So a great, great tool for education history. But, you know, you can create an agent of a physics professor, a law professor, a finance guru, and you can go back and forth. It's kind of a force multiplier there.

Yeah, Megan, that's, that's a great comment. Because, you know, one thing that I think generative AI is extremely undervalued in is the ability to help you learn better. Just kind of personalized micro learning. And I think that's a great example of that. One more, one more here from the audience. You know, Andy, just kind of joking around, you know, saying, hey, Jordan, it sounds like you can make a vacation or you can finally take a vacation. But, you know, I'm, I'm wondering, Ron, from a business side standpoint, is this something that, in theory, right, because everything we do now, it seems like, you know, we're in so many zoom meetings and emails. But when you can create an, an agent and tie it into all of your data, isn't something like this kind of possible? Right, because you can give it training where you could have an AI agent version of yourself. You know, especially if you're constantly talking to teams, if you're presenting, it could do a big bulk of your work for you. Correct.

Sorry, Jordan, I seem to be connection.

No worries, no worries. Yeah, we'll give, we'll give you a second to see if we can't reconnect. Yeah. So this is at least for me and Ron, Feel free to jump back in if you're able to. But I think that this is such a fun use case, right, to be able to train yourself even, right. You know, with everything that, that you talk about. You know, we have, even for us, right, like we have so many hours of YouTube videos, we have, you know, hundreds of blog posts, you know, hundreds of newsletters that I've even personally written. So to be able to upload all of that and to make a, you know, essentially a large language model, but that can speak in real time, that can look like myself, right, Ron, it's seems like, you know, if you're, if, if the connection is back here, Ron, it seems like that's such a great business use case to, to, to be able to kind of replicate yourselves or maybe you weren't able to give people personal attention before, you know, kind of the example of, you know, emails at scale. You know, I'm even thinking of, of, you know, can that be done for your own Internal employees. Right. Like if, if normally you can't give all your employees one on one attention, is that a good use case for agents moving forward in the future?

Look, I'm not sure for one on one attention how you'd feel if your CEO or your manager would pawn you off to an agent. I think it's more about gathering information, Professor. Right. They can't, they can't speak to every single student in their 500 person classroom. You know, so they can upload their course curriculum and kind of answer a lot of the basic questions. When's the assignment due, how's the grading work, stuff like that, you know, for a utilitarian purpose. But you know, when it comes to external stuff, external communication, external messaging, I think it's a huge, huge time saver. I think people are going to want to, you know, take the time, think about. You can have a conversation with a customer service agent or technical support agent and they're not on the clock, you know, they don't have to worry about, you know, getting to the next call. They can patiently listen to your queries. You can kind of introduce topics that you couldn't before. They're available 24 7. They're always trained to be pleasant and polite and patient. I think that's probably the first places we're going to be seeing these.

I love that. And Ron, as we wrap up here, because we've talked about so many different things in today's show, you know, creative use cases, future products that are just getting announced today. But maybe specifically, you know, what's the one thing that you want people to know and understand about this AI agent technology moving forward and specifically how it's going to impact customer experience. What's one thing that you think is.

Important for people to know? Look, I think like everything in the AI world, it's super valuable to experiment. You know, I urge people to go to the studio, try it out for themselves. There's a waiting list right now, but we're going to be opening it up, see how to build one, what kind of information it needs and then carry out a conversation with them. Yeah, in this age of AI, you know, you say it often on the show, there's a new product coming out every day. There's, you know, kind of a world shattering, mind blowing innovation taking place at such a pace. I think it's important that people try it out, see if it works for them. Imagine the use cases or the solutions they can build with it. And a lot of the times it's just taking the plunge, combining it with other tools, something we didn't mention. But you know, that's something that should be happening and letting it loose on the world. I think there shouldn't be much holding us back now.

I love that. And hey, yeah, there is so much that we couldn't even get to. But hey, that's why we have the newsletter. So, Ron, thank you so much for joining us on the Everyday AI show. We appreciate your time.

Have a good one.

All right. And hey everyone, as a reminder, there was so much we couldn't even get to. New products being released, all that. If you haven't already, please go to your everyday AI dot com. Sign up for that free daily newsletter. We'll have a lot more about some of what DID is working on their current products, what's coming out even today. We appreciate that. We hope to see you back tomorrow and every day with more Everyday AI. Thanks, y'.

All.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going for a little more AI magic. Visit your everyday AI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.