AI in Design with Diana Deibel, Chief Design Officer at Grand Studio
26:14 Share

Welcome to the AI Chat Podcast. I'm Jaden Schaefer. Today on the podcast, we have the pleasure of being joined by Diana Dybul, who is the Chief Design Officer at Grand Studio, where she specializes in crafting different tools. And she's worked on a bunch of different AI-enabled tools for industries ranging from healthcare to finance. She is the co-author of the influential book, Conversations with Things, which kind of goes into the nuances of conversational AI. She has a rich background in AI,

qualitative UX research and a focus on healthcare applications. So Diane is a leading voice in the AI and design community, and we're super happy to have her on. Welcome to the show today, Diana. Thank you. Thanks so much for inviting me. I'm excited to be here.

Super excited to have you on. I kind of wanted to kick this off and ask you a little bit about your background and your journey. So right now, you've worked in a number of different AI tools. Did you ever think like, oh man, someday I'm really interested in AI and kind of work on some tools and things in that space? Or are you kind of more just focused on design, what you want to do there, and this kind of just evolves as the world changes?

This episode is brought to you by FedEx. FedEx knows running a small business is hard enough without the hassle of shipping. That's why there's FedEx OneRate. With FedEx OneRate, you can ship your holiday packages cheaper than the post office for as low as $14.50 for small boxes. Visit FedEx.com slash OneRate for details. Exclusions apply. Valid through January 19, 2025. FedEx OneRate. Two-day retail shipping. One flat rate.

What up, y'all? It's Joe Budden here to talk about PrizePix. PrizePix is the best place to win real money while watching football. You can get up to 100 times your money. PrizePix will give you $50 instantly when you play your first $5 lineup. You don't even need to win to receive the $50 bonus. It's guaranteed. Just download the PrizePix app and use code SPOTIFY. That's code SPOTIFY on PrizePix to get $50 instantly when you play a $5 lineup. PrizePix. Run your game.

Must be present in certain states. Visit pricepicks.com for restrictions and details. Yeah, I neither. My degree is in playwriting. So I come very much from like a storytelling, investigative journalism, dialogue centric type of mindset. So I worked in entertainment for a while before I even got into the design world.

And it was very much like through that path of script writing, which is a very easy and for any conversation designer. And from there, it was just sort of like all the pieces kind of clicking of like, oh, well, the documentary journalism stuff that I've done, that is incredible.

very closely related to UX research and the dialogue script writing stuff that I've done very closely related to Louis design. And then the problem solving and the logic was something that like my brain kind of did already, but getting a little bit more formalized in that with the people that I worked with was awesome. And then it just kind of like all pieced together.

That's super cool. So walk me through a little bit of that, right? So you're writing plays. Do you have a friend to introduce you? Did you just have interest? What kind of brought you into the space you're in now? So the first design job that I had was for a health tech company called Emmy. And they got bought out later by Walters Klor. So if you're familiar with that organization, it's the same one. And they were originally looking for a healthcare writer. That's how they build the role.

And they just wanted somebody that their perfect person did not have a medical background because what they were trying to do was create patient education that was online, interactive patient education for people who could not understand their doctors, which is like 99% of us. Even like other doctors can't understand other doctors. So just trying to like translate it into plain language. So it kind of helped to not have that.

background, but to have a good like ability with language and ability to sort of like parse bigger topics. So I wound up getting the job. I loved it. And right off the bat, there was some sort of like baked in logic flows of if the patient wants to explore this topic, how do we

how does that change when they come back around so that you're not surfacing the same thing every time and that it feels a little bit more dynamic and conversational to where they're at. So that was really kind of like how I got in. And then from there, it was deeper dives into omni-channel and multimodal designing and doing more of like on-site field research in the UX space and kind of like

piecing all of that together. Then like now what I do at Grand Studio, we're far more like digital design and multi-service plus digital product plus conversation design. Kind of like we have the big toolbox. We just pull out whatever's needed for that particular problem.

Okay, very cool. So I understand you've done a number of design things specifically kind of in the AI space. I'm wondering, or, you know, working with some AI tools. I'm wondering if you could talk a little bit about some of those. Yeah. So for most of my work, I'm under NDA, so I can't necessarily name drop clients. So I'll start with one that I can. This one was one that I did for Emmy and it was working on a team project.

We put together what effectively was like an AI brain that then sort of captured data from patients who had diabetes. And the intent here was either people that were just diagnosed with diabetes, so they didn't have like a ton of information and maybe needed a little bit more support, or people that had been diagnosed previously, but like for some reason or other had slipped and had just been in the hospital with elevated blood sugar. So-

Kind of like two different paths already of like people that know a little bit more and people that don't. And something about diabetes in general, this is kind of prevalent throughout health care. But I think specifically when we've got conditions that are chronic, that are tied, at least in the media and like public opinion to lifestyle issues.

You have a lot of baked in judgment assumptions from the recipient of any communications. Okay. So the reason I think that this AI approach worked well in this instance is because you had to take the person out of it. So you don't have like a doctor in front of you or a nurse in front of you or diabetes educator who's looking at you and being like, well, how are your sugars today?

Did you get out and go for a walk? Which like even if you ask that in the nicest possible neutral way, it can still feel judgy because of the history of people judging once for that. Gotcha. So putting that in the AI context just took the human judgment piece right out of it. And it was like, I am a robot. I am collecting some information. That's it. If you need like something else, I can connect you to somebody who can get you something else.

But like storage of the information was helpful to deliver the appropriate follow ups. So like if somebody is in an elevated sugar state, well, then we can get them to a human to triage that and to give them like more in-depth support. We would also remember that for the next time. So when we call, we could kind of follow up on, hey, I know last time we talked, you were in this state. How are you doing now?

And kind of do one of those check-ins that isn't necessarily needed or appropriate if the person's like kind of killing it and doing fine. Then it's more of like a transactional, hey, just like checking in. How are we doing today? Yeah, yeah, yeah. That's super cool. And I see like that's one area that I really see like AI excelling at. You know, a lot of people, of course, have the concern that's like, oh, no, AI is going to come. It's going to displace a lot of people, a lot of jobs.

And, you know, I think those are definitely like conversations that are like great to have and look at and stuff. But at the same time, I see so many incredible areas, like you mentioned, right, where maybe someone doesn't want to talk to a human about it, right? Maybe they just want to talk to something that doesn't, you know, feel as judgmental or whatever. Or even, you know, I was recently talking to someone on the podcast that

They're working on like autonomous helicopters to help deliver medicine to like conflict zones. And they're like, they're like, you know, some people like complain and say, hey, we're like displacing pilots jobs. But they're like, the pilot's life is at risk to, you know, fly this medicine into these conflict zones. Like,

you know, you know, for us, we're super thrilled that we have the ability to kind of so there's so many different areas that I think I have the ability to kind of make these really positive impacts. And I think they're a lot less controversial to be like, yes, of course, we should have more things like this, if it's really improving people's quality of life and making it making people's lives better. So something I would love to ask you about with with this one that you worked on and some others, what were some of the more challenging aspects of kind of designing this whole thing for you?

Yeah, I think in every instance, like as a designer, I mean, that's really like the hat that I wear. It is oftentimes challenging to work with folks who are not, I think, as in the weeds in like the design world as designers are. Like that's our job. Yeah. And I think sometimes there can be this block sometimes.

of we designers are taught to have empathy for the users and sometimes that one like other people aren't necessarily don't have that beaten into them the way that designers have that beaten into them so like that's just some sometimes needs to be learned or it's somebody innately has it or they don't and so um that can kind of be like a translation issue but also

designers tend to sometimes be a little holier than thou around that and like forget that the empathy also needs to be had for your stakeholders. And it's not just like, oh, I'm going to go to bat for the user and I'm going to defend them. And yes, that's great. That's important. That's our role. But things get done much better when you also have empathy for the people that you're working with. And so particularly I think in these situations,

in these spaces where design has higher stakes in healthcare and finance, where you have folks that for very good reason have to be black and white, there is less like nuance because someone's life or their livelihood is at stake. Yeah. Yeah. Then that often feels like the biggest challenge is just working through the, okay, I know that like clinically it has to be like this number of

for us to make a decision one way or the other. But if the person doesn't know it, which is sometimes very true for things like weight, blood pressure, sugars, like anything that you might be trying to collect sort of by a self-service method in a healthcare setting, if somebody doesn't know it, you have to have some accommodation for it, for the system to work and for the person to continue to use it and trust it.

And so that's, I think, where the conversations sometimes get really tense. Meaning like if someone's putting their personal like medical data in, but they don't know like what their blood pressure is at or. Yeah, like in the diabetes instance, like we might call and be like, what's your what's your sugar level today? And somebody might say like, oh, it's I don't know. It was seven point three yesterday. So probably seven point three.

A clinician might want, well, no, I need to know what it is today, right now, so that we can check this and make sure it's accurate. And I think from the design side, you say, like, I hear you. I understand why you need this. But like, we also should account for this situation where somebody may

Maybe they like are on vacation and they don't have their tools with them. So they want it to be like exactly to the T. And it's kind of tricky, like you mentioned, too, because with areas like health care and finance, there's regulations that also kind of oversee all this stuff. So, yeah, I totally get that it's tricky. But at the same time, yeah, that's so interesting kind of being the work between especially in an industry like that is, yeah, the designer, you're definitely kind of put between those two places trying to find the best solution that works for both of them, which I'm sure is

is a challenge for sure. Something that I'd be really curious about, I know you've worked, you mentioned finance. What are some areas that are exciting to you in AI and finance that people are designing and working on today? So a couple of the recent projects that Grand Studio has worked on, I have not personally been involved in these specific ones, but

As the person like just kind of peeking in, I think they're really fun. One is just so in finance doing kind of like when we have big piles of data. So if you think about like a trader, they can make very quick, accurate decisions sifting through massive amounts of data that are just like flying at them and changing in every moment.

This episode is brought to you by Amazon. The holidays are here, and you know what that means? It's time to get your friends and family the gifts they deserve. Take the stress out of shopping with Amazon's great deals and low prices on a huge range of items from toys to tech and much more. Whoever you're gifting for, Amazon has great prices on everything you need this holiday season. Shop Black Friday week starting November 21st.

This is the sound of your ride home with dad after he caught you vaping. Awkward, isn't it? Most vapes contain seriously addictive levels of nicotine and disappointment. Know the real cost of vapes. Brought to you by the FDA.

We've designed a dashboard to be able to basically pinpoint the second that it's like, hey, here's the thing that you said you're interested in. Here's another thing that you said you're interested in so that you're focused. You don't have to look at everything. You just have to look at the thing that matters in the moment that it matters. And you can kind of set those parameters for the AI to deliver that to you. And then they can make way better decisions about what they're buying and selling in that moment.

Yeah, that's super cool. And then... Okay. Oh, sorry. Go ahead. Oh, no. Oh, well, something I was going to ask you about is like with this healthcare side and also with this finance side, something that I have always found challenging because I also have an AI startup and we're working and designing and building it. And, you know, we're a small lean team at the beginning. So we have a head of design, but I also have like...

I've done like design work in the past, designed apps and stuff. So like I'm kind of in in it's probably like my designer's worst nightmare, to be honest, because like I'm all you know, I have an opinion on everything. Right. And like I've built things. I'm like, no, we should do it like this. But like, you know, he's obviously much more talented than myself. But in any case, something that I know we've struggled with and I think a lot of listeners, whether you're working on the corporate side or whether they're working on their own like AI tools they're building or even just people that are using AI tools.

How do you find the balance between making the design very simple, intuitive, clean, easy to use, but also like getting all of the features in that you want? I think like sometimes I've seen people on completely different sides of the spectrum with this where it's like super simple, but it doesn't feel like there's a lot there and others where it's like,

There's a ton of buttons and things to do. I have no idea what to do. So I'm kind of overwhelmed and like back away. How do you as a designer, how do you kind of balance that? I'm sure you have a lot of projects. Yeah, I think I mean, for us, the key is user research. So before even getting into these are the features we're going to build, these are the requirements and sort of like setting that all up, doing some research with people who either are existing users are going to be representative of the people that might be using this.

To understand mental models. How do they think about this stuff? What is the sort of organizational mode that they have in their mind? What is the information that is super important to them? So when we're structuring hierarchy, we make sure that we've surfaced the right things in the right order where I would naturally follow it. And then also understanding what their workflow is. So we've designed stuff for electrical engineers, software for smart grids, and we've

honestly, I look at it and my mind just gets shot because my brain doesn't think like that. But I do not need to use it. It doesn't matter if it blows my brain. It works for the electrical engineers. And I think that feels like the real crux. You can't get to it without doing the research. But when you understand what somebody needs, you can deliver on that need. And then it could be

super busy it could be super empty but it's intuitive because it's meaningful i love that i think that's that's a good key because definitely um you know talking with my cto will sit down he'll show me like his code portals and all the things he's like working on and stuff and to me i'm like oh my gosh i don't know what's going on this looks overwhelming um

So, yeah, I think you're right. It's like the you really have to focus on the user, what they need and something that might seem overwhelming to me. I'm like, this is a terrible design. But for an actual user, this is exactly what they want. And it's the best way to do something, you know.

So that's super interesting. Diana, tell us a little bit about your book. I kind of mentioned it in your intro, but tell us a little bit about your book. Yeah. So my book co-authored with Rebecca Ebenholt is called Conversations with Things. And it is basically a practical guidebook for folks who are designing conversational AI. So whether that's a chatbot or a VUI or multimodal.

understanding a bit more about everything from like the principles and best practices of how to do stuff as well as like just getting in the weeds on things that are weirdly hard to find like documentation and part of that is because it's a little bit still wild west there isn't one like figma for everybody to use yeah so people use like a lot of different methods to do it so just kind of like going through some of those practical things that are a little bit harder to find

examples of to teach people how to do this. And then, of course, there's like process and ethics and other things to be considering kind of like big picture as you're moving through the design. That's super cool. What's one maybe insight or valuable thing that you feel like people could take away from your book? Like give us like a teaser for people that might be interested in going in and getting it. And I'll also I'll mention I have a discount code if anyone is interested in the book that I will leave in the in the show notes as well.

So, I mean, I love multimodal stuff and I often find that it's hard to find good information on it. So for me, that chapter is one of the more exciting chapters. And I think especially as we think about like where AI is right now and where it's going.

Again, it's a tool in the toolbox of all the other tools that we can use. And the most successful solutions tend to be mashups of a couple different things. It might be an app that also has a chatbot integrated with it, or it might be a process that supports a digital tool or the other way around.

So knowing how to design for multimodal, multichannel experiences where you're connecting all the dots and still tuning into what does this mean for conversational AI and how does that serve the broader picture? That's what I really get excited about. So that's my favorite chapter. That's super cool. Very, very cool. Something I would love to ask you your opinion on is who do you think could use a little bit of improvement?

Ooh, okay. I was actually, I'm going to not like call out this chatbot, but like I was surprised the other day. I have a couple different credit cards and apps and I was going through, because we're going to be traveling and I was like, oh, I wonder if I need to like let anybody know that we're traveling.

and I went through all the chatbot experiences, and the one that stood out to me as like, "Really?" was Eno, the capital one. Oh, really? Yes, because Eno was so good for so long. It blew my mind that I was like, "What's going on here?" The thing that caught me, it's not like it's awful or anything, but it's so programmatic.

I mean, like you can't it doesn't really feel conversational, especially in this day and age where we've got a little bit more access to large language models. Yeah. And it was repetitive with the choices that it was offering. So it wasn't like taking into consideration. It gave me like four choices. I tapped one of them. It responded. I was like, do you need any other help? And presented the same options again.

Oh, interesting. Okay. Well, that seems like a very basic thing that could be served, even if I get transitioning to any new platform is a beast. So fair enough. But that piece is what caught me was that one's an easy fix. Yeah. Very, very interesting. It's kind of crazy when you see these big, huge platforms, banks or other ones, and it's like they have some fundamental issue like that. Yeah.

I don't know, something about banks and their customer support is, you know, everyone's always got something to say about it, but this game might take it to another level. Well, I will say like customer support, I'm not going to throw banks entirely at fault here, but customer support in general is where everybody really wants to trim the fat, right? Like to just scale it down as much as possible, automate it as much as possible because people are costly. And unfortunately, customer service isn't really valued

in the way that one might hope. So that is not seen as money that is well spent. And so I think- Right. You might think like sales, like get more money, bring more in, less of the like, keep what we have. Exactly. Yeah. So I think I see a lot of the, especially when you were talking about like displacement of jobs, that's where I think we see the most of that is people trying to clear out customer service centers and replace it solely with a bot.

Yeah. And it's like, yeah, that one's tricky because it's like definitely a human could give you a better experience, I feel like, for a lot of times. But at the same time, the AI does like a fairly good job of automating a lot of the process. And as it

as it gets better and better, like inevitably AI will be capable of doing a lot of those jobs. And so, yeah, that one's that one's a tricky one. It's because it's less it's less of even like, is it good or bad to replace people with AI? It's just like it's more cost effective. So love it or hate it, like companies will do it. You know, so that's kind of like what I don't like. Is there like regulations? Is there like what do we do about that? How do we look at it? Because it's, you know, it's coming. And as you mentioned, that's kind of where everyone's

first focuses on on clearing things out and doing stuff yeah we do like as part of our process we do like an impact analysis of whenever we're like designing an ai what does this mean for the people who are in the roles currently and like how are you are we just straight up letting them go is there a way to reuse some of like this intrinsic knowledge that they have and maybe position them to do something else that is higher value to the company and that way we're not necessarily like

clearing everybody out. We're just repurposing them for something that does generate more direct revenue for the company. That's awesome. Yeah. And I think like that is a that's a big aspect is, you

AI makes companies definitely run a lot more efficiently. You can be a lot more efficient. But at the same time, you know, if people's role isn't needed because an AI is doing it. Yeah. Find a new role for them. Find a new place for them in the company to to kind of grow and excel. And all of a sudden your company is able to produce at so much higher of a level. And, you know, it is interesting. I was listening to a podcast recently and they were talking about

they were talking to like some founder and they're like, hey, like, you know, what are your hiring plans for the next year? And he's like, I'm not planning on hiring anyone new because AI is essentially like leveling up my whole company. All of my developers are now, you know, twice as productive. All of my, you know, juniors are now acting like senior developers, whatever. And so it wasn't like he was like, they found this productivity and they're just like, OK, great. We can let go half the workforce. They're like, no, like,

everyone in the company is now leveling up. We may not need to hire anyone for a moment while we like figure out what to do with all this extra productivity, but the whole company as a whole really started excelling more. And I think that's a better mentality than like, great, we have some productivity. Now let's just cut everyone to remain in the same place for a little bit less, you know, when you could double what your outputs are. Yeah, it's definitely like the half glass empty, glass full kind of viewing of things. Yeah.

Yeah. So something as we wrap up this podcast, something that I would love to ask you about is what is one piece of advice you feel like you could give to people that are currently designing in the AI space or, you know, trying to make their AI tools, their AI products, services within their companies, you know, more user friendly, more intuitive, you know, give us the design perspective. I would say it's spending some time with people that are going to use it, whoever that end user is, whether it's like internal employees or outside customers.

And understanding, like, again, those big things of what are their mental models? What are they what are their needs? What are they trying to do? Because a lot of times we see like, especially chatbots, but like, I've heard generic financial advice all my life. Like, don't buy fancy coffee every day.

I like my daily dose of vanilla latte. Luckily, having an account with CefQ means I know how to manage my money in a way that works for me. And their mobile app makes it easy to keep track of my spending and budget for my coffee treat. Now, I'm doing things my way. CefQ. Not a bank. Better. Learn more at cefcu.com. Insured by NCUA. This episode is brought to you by Shopify.

Do you have a point of sale system you can trust or is it a real POS? You need Shopify for retail. From accepting payments to managing inventory, Shopify POS has everything you need to sell in person. Go to shopify.com slash system, all lowercase, to take your retail business to the next level today. That's shopify.com slash system.

Any sort of conversational AI gets thrown out as like, yes, we'll just put this up and it'll solve the problem. But it's not necessarily the mechanism that people want to interact. Like sometimes it's just a better layout on your website. You don't need a chat bot. And getting to understand sort of like what people are actually trying to do and the ways in which they want to do it.

will help you decide like, is this even a thing? Is AI even the right thing for us to be doing here? Not to like knock it out because there are really great use cases for it. But I think people are more receptive when we as like an industry are a little bit more selective about where we are applying it. And then it becomes actually useful and people are like, great, a chatbot as opposed to, Jesus, a chatbot.

Right, right. Yeah, I think that's a great piece of advice. That is very applicable. Diana, thank you so much for coming on the podcast today, sharing your insights, your perspectives and everything that you're seeing. If people want to get in contact with you or your design agency, what's the best way for them to go about doing that? Yeah, we are at grandstudio.com and they can email me at hello at grandstudio.com and I check that. So feel free to reach out there.

Awesome. And I will, to the listener, I will leave a link to that in the description of the podcast. Once again, thank you so much for coming on. To the listener, thank you so much for tuning in to the AI Chat podcast. Make sure to rate us wherever you get your podcasts and have a wonderful rest of your day. Thank you.