AI Without the Hype: with Mitali Deypurkaystha
AI Without the Hype: with Mitali Deypurkaystha
Brand Design Masters Podcast
Episode 170
===
Philip VanDusen: [00:00:00] Mitali, I'm so glad to have you on the show today.
Mitali Deypurkaystha: Thank you for having me. I really appreciate it.
Philip VanDusen: So we're gonna talk about all things AI, and as I said in the introduction, you came and talked to my Bonfire mastermind community about AI, and Bonfire's full of a lot of creative professionals who are mid to late career, and like everybody, as we were talking about just before we hit record, there are the experts in AI, and then there's everybody else, because AI is so new and it's so overwhelming by what it can do, like literally every day we're hearing new things or new things are coming out, that we're all very overwhelmed about the complexity of it and the depth of it and the shiny, new objects.
Just right out of the gate, what would you say to people in terms of how to set their heads around what's happening in the AI landscape?
Mitali Deypurkaystha: Oh gosh, how do I answer that? Straight out of the gate, that's a huge question. First of all, I would say take a deep [00:01:00] breath. Understand that the vast majority of people are with you.
There's always this feeling that everyone else is getting ahead and you have to remember that's the world we live in. We live in a social media world where everyone's putting their best foot forward on social media, hiding their messy real selves, so we're always seeing, we go online, if you're on LinkedIn, if you're on Instagram or if you're on Facebook, wherever you are, it appears like everyone's ahead.
Then you're bombarded with all the marketing from the purveyors of various AI platforms and clones and chatbots and all the rest of it, and of course it's their job to make you feel FOMO. It's their job to say, "Oh my God, you're missing out. Your competitors are on this. You're not gonna be in business in 18 months' time if you don't get on."
That's their job, to sell you their wares. So I would say first step, just take a big breath and just understand that the vast majority of people on this planet are where you are. Only a tiny handful are, have actually moved forward in a way that's meaningful. I don't know how many people are aware, there was a big [00:02:00] report that came out from MIT last year I think it was August time, sometime in the summer last year.
They said over 90% of AI pilots in large businesses, 'cause they obviously had the money to move forward, over 90%, I don't remember the exact over 90%, that's what I remember, of AI pilots had failed. So that's these corporations spending millions, tens of millions maybe even more, on AI pilots which have failed.
So just take a deep breath. Don't start wrecking your head over this. Just understand that most people are where you're at. If you're old enough, if you're old enough like me, you'll remember the first digital revolution when, you had websites and y- the internet and- We all went through that then.
I remember the very first time opening a browser, and I didn't know what a browser was. And then what I'd done was I was working as a PA at the time, and I thought I'd destroyed the browser. All I'd done was I'd just hidden it, and it'd just gone to the bottom of my taskbar. [00:03:00] But what you have to remember, this is, would've been 1999, and I just thought, "Oh my God, I've only had it for half an hour, and I've broken it.
My boss is going to kill me." So now it's laughable. It's like it's just a br- now I have about 25 browsers open at the same time thinking, "God, I need to close some of these browsers," so it just goes to show how much we move on. I would say just take a deep breath. Second step I would say is a great way for...
This is what I work on with my clients. A great way to cut out the noise is to say forget about AI. Forget about technology altogether. What are the big bottlenecks in your business? Are you struggling with lead generation, or do you get lots of leads, but you're not able to convert most of them? Or you're able to convert lots of them, but then you're, you struggle actually servicing it all.
Maybe you're in a feast and famine cycle where you do a lot of lead generation, you do a lot of sales. Wonderful. Got lots of money in the bank, and then you have to stop it all. You've got to stop all [00:04:00] the lead generating sales 'cause you're now delivering on the services or products or whatever it is that you're providing.
So then the money starts depleting in your bank, and so you're now in a feast and famine cycle. Is that a bottleneck? Is that something that you would like... You w- would you like more level earnings each quarter, each year, however you're measuring it? Is it something to do with, do you have a recruitment issue?
You're not attracting the right people or you're not attracting enough of the right people. It's finding out where your bottlenecks are, then opening up, whether you do it on your own, if you've got time on your hands, lots of great resources online. Go... You can even use large language models. You can go to ChatGPT or Claude or Gemini and say, "Hey, I've got this problem.
What's out there at the moment?" If you do have some, if you do have any kind of budget for this transformation, then certainly you can shortcut it by using someone like me, an AI strategist, a consultant, and then we can do that on your behalf and find the best technology and AI solutions for you. That immediately removes the [00:05:00] overwhelm 'cause you're not thinking about tech, and I always struggle with anybody who's thinking tech first.
And I was saying, "Let's think human first. Let's think problem first. Let's think bottleneck first, and then let's add on the technology that only solves that problem." And that completely removes the overwhelm 'cause then, I love it because what, one, one of the key things that my clients say is they end up with a strategy map a full, they get key- a full key findings report that literally gives them a 6, 12 or 18-month plan on what they need to do, and you can just see that sigh of relief.
They're like- Oh, great. So now when that annoying person contacts me on LinkedIn wanting me to buy their whatever AI platform they've got, they can just filter it through the strategy plan. And if it doesn't fit their plan, it's "No, that's not... We don't need that. Thanks, but no thanks." And if something does, it's "Okay, let's meet with this person 'cause actually in this plan, we actually need this now or in two months' time or in six months' time," or whatever the strategy plan says.
And it [00:06:00] just gives you that clarity. So first step, take a deep breath, don't sweat it. Most people are where you're at the moment, despite what you might be seeing on social media. And the second thing is cut out all the tech and the AI and just think, like every good business owner does, you need to take a step back sometimes and go, "Hang on a minute.
What's working well? What's not working so well?" And then you can try and find technology and AI solutions that will plug the gaps that you've got in your business.
Philip VanDusen: So y- I go way back, too. So I was there with you at the dawn of the internet and d- and browsers and all of that. And in the creative professional community, we went from, plates and physical presses to desktop publishing, right?
And so suddenly everyone had these tools at their disposal, and everyone thought they were a graphic designer. And then they s- you know, you thought, "Oh, my dad can make his own poster for his business." And of course, he tried with, Microsoft Paint and all that sort of stuff, and very quickly the populace r- realizes that [00:07:00] graphic design's a little more complicated than that.
And the funny thing is that with all the layoffs that we're seeing right now at a lar- a lot of e- even enterprise companies, but particularly I'm hearing a lot of graphic designers and creative professionals who are getting laid off because their bo- their bosses are just saying, we're gonna get AI to do it.
Get AI to do it." And it's like we're in that phase right now where they're seeing AI as this thing that they know everyone uses, and it solves everyone's problems, but they don't understand the complexities of implementing it or, as you said, how it's solving a bottleneck or who's gonna run it, who's going to make it work to do the thing they want it to do.
They're executives. They don't think in the weeds. They're just going, "Throw AI at it. It's gonna solve everything." And it doesn't. One of the things that I know that you're a specialist in is data and how companies use data or what data has become inside of companies and how AI might be able to get them [00:08:00] out of that kind of quagmire.
Can you talk a little bit about that? What is the functioning about AI have these days that can transform businesses, and what does it take for you as a consultant coming into a company that ha- is swimming in a lot of data that they're either not using to good effect in their business or what does that sort of thing look like in terms of AI?
Mitali Deypurkaystha: Oh, that's a... God, that's a multi-layered question there, so the
Philip VanDusen: first- I'm good at asking really big, deep questions
Mitali Deypurkaystha: You are, I love it, but it's like it's very layered. So regarding let's talk first around the people side of things. I think we've already seen a lot of... So certainly major businesses that have had a lot of egg on their face recently where they've laid off loads of people and then had to rehire those people, or new people.
Yeah, we're
Philip VanDusen: seeing that now, yeah. '
Mitali Deypurkaystha: Cause the realization what's happening at the moment, and this is quite sad really, what's happening is a lot of these businesses are falling for the marketing [00:09:00] hype, so what I was talking about just before. You have to remember a lot of these AI businesses they're in the job of hyping themselves up because they need to impress investors.
They need to get more investors. They're doing various rounds of investment, so therefore they need to keep hyping. I just saw something last week where it, I'm pretty sure it was OpenAI. Te- OpenAI is the biggest hype merchants, but they're all in it, to be honest. We hear more about OpenAI 'cause it's the biggest company.
No one had ever even heard of OpenAI until November 2022, and now that's all we hear. But they've just released their latest image model, and this is now able to create magazine covers. I don't know if you've, if you actually saw that, Bill. I just
Philip VanDusen: used it for the first time yesterday, and it freaked me out.
Oh, okay. It's amazing. Yes. It's absolutely amazing.
Mitali Deypurkaystha: It's fantastic. You just give it a few prompts and it creates them. So obviously there's this outcry, this whole like, oh my God, what about magazine layout professionals? How... What are they going to do? But [00:10:00] what was really interesting is I read an article from one of these professionals, and it said what, you get it, but then there's no way to tweak it because it's a solid image.
You can't then move it into Adobe InDesign or anything to move it around. So what ends up happening is it's a great way to show off to investors. That's what... A- and apparently that's what they did. Somebody showed it off, and it became news, and the investors are all clapping and going, "My God, this is amazing.
Let's throw a few more billion at this company." And it goes on. It's this game that they're playing. And then we're the ones who get caught up in the crosshairs because we're going, "Oh my God." And then all these professionals are thinking, "Oh my God what value am I going to bring in if now all people have to do is put a couple of prompts in and they get..."
'Cause that's what they did. They put literally a prompt in. But what they didn't say is that, yeah, but that's just the first iteration. Now we need to tweak it. You can't tweak it, so now you have to prompt, keep prompting it. And what I found, I was playing with it to get the magazine cover that [00:11:00] I wanted, I ended up spending about an hour and a half- And still not exactly what I wanted it to be, honest, 'cause I kept having to tweak it.
It would be wonderful if I could then take it off into a design platform and, but that's not how large language models work. They just create this image. So you have to remember that there's a lot of hype there. So I would say to, to anybody who is in, who's a design professional, a branding expert, it must be very uncomfortable at the moment, especially if you're working in a company or maybe you've got some high-profile clients.
They're saying I probably won't need you." I'll say with large language models that you've gotta treat them almost like people. Not in terms of please don't start talking to them as if they're a real person, 'cause we've had whole huge problems with that. We even had a horrible incident last year where one young man killed himself, that he was talking to ChatGPT and who ended up agreeing that him killing himself was a great idea.
OpenAI is now being sued by the family [00:12:00] quite ri- quite rightly so that's going through the courts right now in the UK. So when I say treat them like people, don't do that. Don't treat them like your neighbor or a family member, not like that. But I mean in terms of you get in what you put out. So if you've ever been with somebody and you've had a really in-depth conversation with that person, think of how much more that person is willing to do for you because they feel seen and heard by you.
I do a lot of networking, in-person networking, 'cause I love speaking to people, and one thing I learned the hard way is I used to, a number of years ago, work the room. So I would go through, have 10-minute conversations with everybody. I'm thinking, "Oh my God, look at me. I'm amazing. I'm talking to everybody here."
And then I realized I was creating these very shallow connections. So now what I do is I'll just speak to three or four people, and there might be other people there that were, but I will see them again. I live in the northeast of England. It's a reasonably small enough... The business community is reasonably small enough, I will see these people again at some point.
But having those really in-depth [00:13:00] conversations with people, that just opens up doors. Either they wanna work with me, or they know somebody who wants to work with me, or they wanna introduce me to a pot- a potential affiliate or strategic partner, and I've noticed I get more by going in deep rather than going in wide.
It's exactly the same with these large language models. What I can get out of a large language model when it comes to creating a magazine cover would be okay, but I'm not a design professional. Someone who's been in the game for the last 10 years, they prompt those large language models, they can get something infinitely better.
So I would say I feel bad for you because what's happening is if you're an independent, you've probably got clients who are now saying, "Oh, I can just do this on AI." If you're working in a business, you could be worried about your job. I would say the best thing to do is start having those convers- gentle conversations.
Please don't say you're stupid for thinking that." Don't do that. But having those gentle conversations, just say- Hey, just be [00:14:00] aware with these large language models, it, you still need a human per- person's input in there, and the more, the better the input, the better you get out of it. So really this is not an excuse to be removing people from their work.
It's to empower them to get things done at a higher level, at a faster output, making them more productive. That's what it's all about, not getting rid of roles. The only roles that really should be going are roles that I believe human beings shouldn't be in. Like for example, or anything that's robotic, automated, anything that doesn't involve creativity.
Human beings shouldn't even be in those roles. And then there's obviously the argument what are these people going to do? This is where I get a little bit, acting out there. I'm I'm a member of the Green Party here in the UK. We're one of only two political parties in the UK that are pushing for universal basic income.
We believe that every human being [00:15:00] around the world, but sorry, our jurisdiction is the UK. We have man-made countries, so what can we do? I'd love to speak for the world, but I can't. But definitely in the UK, every human being here should be given a mandatory universal living wage, not just like a small amount just to keep them going.
Not enough to buy a yacht or anything crazy like that, something that's nice, livable. And then they can choose if they want to work, if they're in a creative field where it's required, you need that human creativity, then great. Otherwise, they can go into voluntary work or they can bring up children and, be full-time parents or do whatever they want inside the community.
Create art, get into academics, that sort of thing. So that's what I believe in. But if you're a creative, then really you're needed now more than ever because what we're seeing with these large language models is when creative people go on there, regardless of whatever it is, whether you're a branding expert, whether you're a videographer, whatever it is that you're doing, your output from these models is, I was gonna say 10 times better.
It's not even 10 times. I would say probably close to [00:16:00] 100 times better to what the average person can get out who isn't a creative. So you shouldn't be fearful for your job, but I understand why you are 'cause other people are being blinkered by the AI hype. What happened last week with, oh, now we don't need anybody to design front covers of magazines anymore, which is absolute nonsense.
We absolutely do need them
Philip VanDusen: But then as a designer, I'll tell you, then as a designer, the CEO, whoever, the creative, the CMO comes in and says, "Oh, we need to change that copy, that, to say this," and you're like we can't."
Mitali Deypurkaystha: It's edited.
Philip VanDusen: Uneditable. I'm a- I'll tell you a very quick story. I used to do some work for one of the largest grocers in the United States, and they were developing a private label brand, and this was right at the beginning of crowdsourcing, right?
So 99designs and all those companies where you could like- Yes ... put, could put $100 up there and ask 500 people to contribute designs. So the head of marketing said, "Oh, we're developing this private label thing. Instead of paying our agency," who [00:17:00] was us, "$75,000 to develop a brand identity, we'll just put it up and crowdsource it for 2,500 bucks."
So they did. They wrote the brief. They uploaded it for 2,500 bucks. They got back something like 800 submissions. It took them about six weeks to get down to 60, and then they called us and they said, "Hey, can you come in and help us sort through these submissions that we've gotten?"
That's cheeky. And we're like, "Yeah, sure." Yeah, talk about cheeky. Exactly. That's
Mitali Deypurkaystha: cheeky.
Philip VanDusen: And so we go in, we choose five, and they're like can you take these five and just adjust them for us and tweak them for us?" So we're like, "We gotta get some money out of this," so we said, "Sure, yeah."
So we went back and we did that, came to submit them to them, and they're like, "You know what? We don't really like any of these. Why don't you guys just design the identity for us?" So we'd burned through eight weeks of time, and now we had much less time to do it, and we ended up getting the project and having to do it anyway.
Yes. So it's, that has a lot of echoes to what [00:18:00] I think both of us are seeing right now. Okay, so my question was actually about data, and I took you down a rabbit hole. Okay, so let's come back to data.
Mitali Deypurkaystha: Yes.
Philip VanDusen: Okay, so businesses swimming in data, what can AI do for them?
Mitali Deypurkaystha: A, a lot, depending on how the data is structured, how it's labeled, how it's cleaned.
That's one of the biggest issues that I'm seeing. So when I got into this, it'll be just over a year ago now I thought most my role was going to be strategizing the AI and then getting it implemented. Happy days. That's what I was meant to do. What I didn't realize is, a- and this, by the way, is exponential depending on the size of the business.
The bigger the business, the more data problems they have. What they tend to with smaller businesses, if you've got a business that's only got five or six people, the data can still be in disarray, but it's not a huge amount of data, so it doesn't take too much to collate it all. They also, even if they don't have basic things like file naming conventions, maybe people are just naming files whatever they feel like, because it's [00:19:00] only a small team, they can figure out what they're...
'Cause they're in close proximity with each other. So they can even if someone else has named a file in a way that they wouldn't name the file, they know that person well enough to go, what I was talking about before, you know when you go deep, when you have those deep relationships with people?
Yeah, it's "Oh yeah, that's how they always name their files, but I know what this file's about." Then that means when I come in and I can say you guys don't even have a file naming convention. You don't, you store things in different places when they should be stored in one place."
They can go, "Okay, j- just give us a couple of days. We can sort this out." And between them they can sort it out. And I'm like, wonderful. Once you start getting to bigger businesses, the problem just gets exponentially bigger, 'cause some of these people are literally not even speaking to each other on day-to-day basis, or even a week-to-week basis 'cause they're in different departments.
And I also, AI is sold to us, again, AI hype, it's sold to us as this magic bullet. The reality is the AI is a bit like having the sleekest, most beautiful sports car. But without the data, i.e. the [00:20:00] petrol, you're going nowhere. You're just looking great.
Philip VanDusen: What a
Mitali Deypurkaystha: lot of AI does, looks great, doesn't really do a huge amount.
It's you know when I saw this this demonstration of the magazine cover, you can't tweak it. Who's gonna use it? And that's what this this gentleman who was criticizing it, that's what he was saying. It's who is this then for? Because it was not actually for magazine creators.
Who is it for? And then he, that realization that he came to, this is for nobody apart from impressing investors so they can put more money into the bus- that's literally who it's for. Because who else would be... I know you're gonna get, a load of idiots on LinkedIn or Facebook of the you heard it here first, yeah, expect in the next few weeks to get...
Do you remember last year when it was all the action figures, everyone was, were turning themselves into an action figure- Yes ... ChatGPT. The next one is going to be, you've heard it here first, expect to find it flooded on your feeds, people have created magazine covers of themselves. Some people will do it as a laugh "Look at me, I'm on the cover of a [00:21:00] magazine."
But in reality, what, who is this actually for? Nobody. So this is the reality o- of AI. It looks good, but it can't actually do any good without the data. And I'm going into business now, which are larger businesses, 10, 15, 20 million plus, and I'm having to spend weeks, sometimes month, months just structuring and cleaning their data, because I know if I put AI on top of...
I use the analogy of a sports car and petrol, but actually I need a better analogy because if you don't have any petrol, the sports car doesn't go. It just sits there looking pretty Which is annoying, but at least it's not causing any harm. I need to find a better analogy because with AI, if you put that on top of unstructured, uncleaned, disarrayed data, it will go and it'll sound-- it'll tell you all the things that you want to hear, but it's on corrupted [00:22:00] data, and it's giving you completely the wrong-- well, there's no insights.
It's just making stuff up. And then we blame the AI, "Oh, it's hallucinating," but really it was the data. So I need a better analogy because like I said, a sports car doesn't go without petrol. It's not gonna go the wrong way or go backwards or anything like that. Whereas AI is actually more dangerous in that it will go, but it'll steer in the completely wrong direction, and you can make some serious costly mistakes, and also reputational mistakes, which equals costly mistakes as well.
Philip VanDusen: Okay. So you reminded me of a story I heard recently on social media, I take it, but someone had developed some, a couple simple agents essentially to clean their inbox and their email, and it ended up deleting all of their emails- ... and they couldn't get them back, and this was, a decade, two decades of emails.
And so that-- let's talk about agents a little bit and what agents are and what they can do. 'Cause when I, literally three months ago, I didn't really understand [00:23:00] what agents were or agents that could work independently as opposed to directed agents.
Talk about what agents are and how they operate.
Mitali Deypurkaystha: Okay. First of all, I just wanna say don't get bamboozled by agents. Again, it's another buzzword in the industry right now, and everyone's "Oh, I need agents." I was just speaking to someone. I was speaking at an event just yesterday on a panel of AI experts in the northeast of England.
And a gentleman came up to me, and he runs a business. He runs a fitness business, and he was sold AI bots, I guess were ag- agentic bots two years ago that were meant to just take over his business. Like it can basically, it can ba-- they had one AI bot agent that can go and do all the marketing and posting for him on LinkedIn and Facebook.
He's mainly focused on Facebook and Instagram, being in the fitness industry. He also had w- a chatbot agent that is able to, converse with people on his website and then book in calls with him so he can close them. [00:24:00] He had a whole plethora of them, and he threw it all out within three or four weeks.
Within a month, he was like, they were hallucinating. They were... Basically what this person had done, this person had sold him the dream, essentially. They'd just put in-- I think what they must have do- I'm only guessing here, but I did ask, I did a little bit of, discovery with him and said what data did they take from you?"
He's "Oh the stuff that's on my website and on my Facebook and Instagram bio." And it's do you normally do calls on Zoom with potential clients? And it's yeah. Do they ask for transcripts of that? No. Do you do podcasts? It's "Oh, yeah, I'm on podcasts all the time."
Fitness podcasts and shows like that. Did they ask for to see the recordings of those? No, I didn't say. So basically they, I assume what they must have done is gone to a large language model and just asked it, what are the typical questions people ask of a fitness guru? And basically, of course, it's so general, of course it's going to hallucinate.
Whereas the way I [00:25:00] would've done it is we need at least three months worth of transcripts, podcasts. I'm working with one gentleman at the moment, and he's done, I think he's done 50 podcasts or video podcasts as well, which is amazing, so I'm now going to clone him as well, 'cause we're gonna have a clone of him on his website, but the data has to be there, otherwise it's just, again, going back to the magazine thing, 'cause it's fresh in my mind, it's just happened. Who's it actually for then? If it doesn't actually work, it just looks good and could cause a lot of harm. So to answer your question regarding agents, think of I always think of almost like a multi-step level.
So at the base level let's go with the email analogy, because you just talked about your friend who, bless him, lost so many. That would drive you crazy. I've got emails going back to 2007, 2008, so I definitely don't wanna lose those. If you think of at the basic level, if you, just on your own with no technology, you open...
you have technology to open, you have emails, but that's it. You have to sift through everything. [00:26:00] At the ba- at the basic level, you could have something, and some people already have this, I use this, I've got Google Workspace, so I have Gemini that follows me around. And when I try to create a reply for an email, it's already sat there and thought about what the reply should be.
It's read through that email. If it's a whole string of emails, it's read through all of it and summarized it and giving you a possible way to reply. That is the first step. That, I would say, is, I would call it tools. You're using very clever AI tools that can summarize, and they can do things while you're there with it.
The agent is the next step up. For example, I've got a wonderful piece of software called SaneBox, and it does keep me sane. Oh my God, it really does. Wonder whoever named it, they probably are making a lot of money. But SaneBox are a brilliant company, and what they do is it slowly learns which emails are important and which emails are less important and starts segregating them.
So the most [00:27:00] important emails it'll put into your main folder, but emails that are important that you might want to read, things like maybe newsletters that you've signed up to or any notifications, it'll put you into a separate folder saying, and it's called Later, as in, when you've got time, have a glance at these.
Doesn't need a reply Then you can train it to be like, "I really don't wanna hear from this person." It goes into, I love this, the SaneBox black hole. Love the name. I love putting things into the black hole, like never darken my doorstep again, and I put it into the black hole folder and it will never come into my...
but it automatically knows that Holly does not want to hear from this person, or does not wanna hear from this company, and it goes straight to the black hole, never to be seen again. And it's wonderful. That's a, it's doing stuff without me always there. In fact, it's doing stuff now. There's emails coming into my inbox right now and it's sorting them.
That's a basic level agent. Next level agent would be it's sorting, but then it's also starts creating replies. So [00:28:00] I've created an agent based off SaneBox, so all the people that need replies, it creates a reply and it sits in my inbox folder in drafts, and it gets better and better. When I first started it, most of it needed tweaking.
It's now got to a stage where I would say 65 to 70% of it doesn't need tweaking. It's already ready to go. Some of it needs a bit of tweaking. It's eh, I think that needs to be a little bit more direct or maybe not as abrupt, needs to be more friendly or, and I need to tweak it a little bit.
That's the next level up. A level up from that, and that's something I'm looking into, but this again going back to data, my I'm very good at structuring other people's data. Please do not ever look at my Google Drive. It's an absolute mess. I've had it since like about 2009 or 2010, and I keep thinking I need to go and clean that drive 'cause it's just ev- all, everything everywhere.
But once I do clean it and structure it correctly, then I'll be able to go [00:29:00] to the next level where not only will it create replies, it can go and fetch, it can go and look at my Google Drive and actually fetch attachments if it thinks it's appropriate. So it can go "Oh, you wrote, you created a PDF on this," or, "You created an infographic on this.
Let me attach that," 'cause that'll be... But then not sending it, getting me to verify that everything's okay. So you're bu- does that make sense? You're building layers, almost going back to thinking of them like human, but I always say it with a caveat. Please don't start talking to it as if it's your pa- don't fall in love with it.
Don't treat it like a therapist. You need human beings for all that kind of stuff, so I keep saying this caveat. But apart from that, treat it the same way you would do with a human being. So you have a, an assistant on the first day. You'd give them very manual things to do, and you'll keep checking up on them.
And then you'd expect, if you expect that assistant to grow in their role and become your personal assistant, and you expect that person to be able to do things off their own bat and not be asking permission or asking you to check for every single thing that they do. [00:30:00] That's how agents work. They've got, you start giving them some autonomy and that's...
I always say that, if you think of, a lot of people I think with agents they now think of 007 And like secret agents and don't think of them like that. Think of them more like a travel agent. Imagine if you went to see a travel agent and you said, and they ask you questions about where you'd like to go or what kind of vibe you want to have on your holiday.
Is it a beach holiday? Are you going skiing? Is it a city venture, and imagine if you then told them exactly which, w- which site to open up to start, looking at different flights, which comparison site. Or they'd be like, "Excuse me that's my job. You tell me the outcome and let me figure out the rest," because that's...
And that's what an agent is. It's basically a person, or in this case we're not talking about technology, where you would give certain amounts of agency to do things on your behalf. That's essentially what it is. And hopefully you've [00:31:00] now breathed a sigh of relief and thought, "Oh, is that all it is?"
It's you have to remember, people just hype it up, especially AI marketists being like, "Oh my God," and then bamboozle you with all this tech. That's all agents are, is just next step up from large, a large language model like ChatGPT or Claude or Gemini. More intelligent than the most intelligent person that's ever lived on this planet, but it does just sit there idly and do nothing until you prompt it.
Whereas an agent will start doing things on your behalf even when you're not, even when you're away, as long as you've set it up to go and do things on your behalf. That's all an agent is.
Philip VanDusen: I try to... Initially I began to think of it like Zapier on, on steroids. Like I use Zapier to, like when someone signs up or buys a course, then I get an email.
It does this, it adds the person to my email list over on Kit. It just it aligns and connects all these different aspects of my business, but it does it so I don't have to, purposefully move that stuff through the sales cycle. [00:32:00] But this is like layering a lang- a language model on top of that and giving it a level of intelligence that Zapier doesn't have.
And also sometimes setting up the technology of the steps in Zapier is really laborious and can be very tech he- heavy. So anyway, that's how I think about it. I want to pivot just a little bit and talk about LLMs. One of the things that and I don't know, if this crosses into your area of expertise, but LLMs are now becoming the new Google search, right?
Everyone is doing their searches for their businesses or doing searches around topics in LLMs, and LLMs are now starting to recommend products. They're starting to recommend different websites, different sorts of content. Yes. And so for discoverability, LLMs are becoming much more important. There's AEO, ask engine optimizations.
There's GEO, generative engine optimization. There's different, acronyms that are being assigned to it. [00:33:00] But so- In terms of your expertise in AI, how would a business or even an individual contributor or consultant, how do you actually try to figure out how LLMs would recommend you as a consultant to businesses who might be searching for help with AI?
Mitali Deypurkaystha: That's a really good question. That's something actually I've not gone into, although I've been told I have turned up on people's searches, which is good, so I'm doing something well inadvertently. But I haven't even looked into that. I get most of my clients from recommendations, referrals. I do a lot of networking.
I'm a very... One thing one of my mentors taught me, the gentleman who, for Industry Rockstar Academy, that's where I'm certified Kay Minkus, and one of the key things that he talks about with AI is the more we get into the AI world, the more people skills become important because w- the people who can actually speak to another person, it's already a dying art.
You [00:34:00] have... there's I believe the under 30s don't even pick up their phone. I know. Just if it rings, they don't know what to do with it,
Philip VanDusen: it rings- Or they break out in a cold sweat and they get- Yeah ... an anxiety attack. Yeah.
Mitali Deypurkaystha: Exactly. They just, the idea of just one, one of the key things that I do that works wonderfully for me is that if I email and I email a couple of times, and if I've not heard back from you, I just call them.
And you can see their surprise like, "Oh, hi, Natali. Yes, I did receive your emails. I'm so sorry, I've just been very busy." I said, "No, that's why I'm, that's why I'm chasing up. No, I'm just calling to see." And then amazing how many deals I've closed just by ringing up the person. But it's something that most people have lost that art of just the idea.
They'd rather just "Oh, I'll just, I'll email them again. I'll just give them a gentle nudge." The old just checking up on you email, which is actually very annoying. I hate it when people do that to me. Say, "What are you checking up on?" Do you know what I mean? So just a gentle nudge.
You've seen those emails. Just a gentle nudge.
Philip VanDusen: I've sent a thousand of them.
Mitali Deypurkaystha: Just ring them, honestly. And [00:35:00] I've got another tip. The best time to ring them, Friday mid-afternoon. Everyone's in an awesome mood 'cause the weekend's coming. They're usually winding down, and you just ring them and they're like...
And it's amazing how many people are just like, Natali, thank you so much for chasing those. I've been seeing your emails. I've just been flat out." I'm like, "No, not a problem. It's just how things are." And they're like let's move this forward," and I've been able to just get the deal across the line.
Just by ringing them up on a Friday afternoon when they're in the best mood, and they're just so surprised that someone dared to actually call them- I know ... 'cause it just doesn't happen. They didn't ex-
Philip VanDusen: I just did this two days ago. I came... I was in Egypt for about 10 days, and I was on vacation, and I've had this big client project that kinda stalled and has a tendency to go in fits and starts.
And this very senior executive that I'm working with gets very busy, right? She's probably overwhelmed with emails, too, and so she would go dark for, three, four weeks at a time, which is a problem when you're trying to get a project through the steps and phases. And so I came back from Egypt, and you're like...
I'm like, "I'm not sending another 'Hey, just [00:36:00] trying to follow up' email. I'm just gonna ring her up. I'm not gonna text her first. I'm just gonna call." And of course, she picks up, and we had this wonderful conversation. It's very musy, very light, and she's "Oh yeah, let's get this thing going again."
And it was so easy, and I was just like, "Oh my God, this is I just discovered the new internet." It's like... and also the other thing, Mitali, is that it's in branding and in business we talk a lot about differentiation. How do you differentiate yourself amongst all of your competitors?
Mitali Deypurkaystha: Yeah.
Philip VanDusen: If anyone is listening right now who's an independent consultant who wants to differentiate themselves in their industry, get on the phone.
Oh, God. Literally get on the phone and have a conversation with somebody. Because I totally agree with you that the human side of things is going to become even more important as we move into the age of AI because things are becoming so much more automated. The other thing is that you can't tell whether you're getting a response or over 50% of the posts on LinkedIn are by AI [00:37:00] now.
Yeah. So- Yeah ... it's like the human connection is going to be the only assurance that we have that we're speaking to a human and not some sort of an AI agent or bot. Okay, I wanna talk about something that you started, which I think is brilliant, and I wanna hear about where it came from for you. And so that is you're the co-creator of the Human First Responsible AI Pledge, which is a UK-based pledge, which is a fairly simple pledge that executives or companies can take to set themselves up to be responsible around AI.
Where did that initiative come from, and what was your- What was your, motivation? What did you see that made you feel like this needed to happen?
Mitali Deypurkaystha: Thank you for this. This is something that's very close to my heart, and I... i'm a co-creator. I've... There's two more AI consultants based in the North East of England.
We keep seeing each other at networking events. Again, just goes to show, phone calls and in-person [00:38:00] is gonna become even more important in the a- age of AI. You being able to converse with people in person and have those in-depth conversations, those real human conversations. We've only met a handful of times, and yet we became very fast friends because we were in person with each other, or we were calling each other, or on Zoom with each other.
And something that we were, all three of us were seeing independently with the clients we were working with. So one, th- the gentleman, there's three of us, there's two of us women and one gentleman. Pascal is a digital transformation expert. He's been doing it for a very long time.
Pretty much helping, years ago, helping people just get online, get their first website up and that sort of thing, and obviously now moving into AI as well. Really focuses on the marketing side, so he really goes in and helps businesses from the marketing point of view and now how to use AI to market your business more effectively.
Then there's Yulia, who has a lot of experience with social enterprise and also working with organizations like the universities and colleges. [00:39:00] So she's got a very different viewpoint. And then I work with larger bus- I work... I started off working with founder-led businesses. Still working with some founder-led businesses, but now I'm moving into working with larger 10, 15, 20 million pound turnover businesses, and I'm seeing something different entirely as well.
So we're coming from three very unique angles, and yet what we're all seeing is the issue with data. So we've already discussed at length that if you don't have data that's structured well and labeled well and cleaned well, then y- d- you don't have a Lamborghini. You have a Lamborghini that's going backwards and gonna crash into somebody and cause all sorts of mayhem for you.
But there's something else we were noticing with data, and that is even when it's structured correctly, there could be biases in the data that are causing issues. Let me give you an example. So one of my clients based in Manchester, she runs a recruitment firm. And she comes from very extremely humble background to the point where we actually bonded on the fact that we both didn't [00:40:00] have indoor toilets till we were six or seven years old.
We had to go outside for our toilet. I'm giving away my age now, aren't I? But that's how poor both our families were in the beginning. We both come from poor immigrant families. And we had to work our way up a lot of immigrant families do. And we bonded on that, and she has a very successful recruitment firm now in Manchester in the North West of England.
And she was using, 'cause, you know- Like everybody, you want to use AI to be even more productive. She was using an AI model that is able to sift through CVs and find the best 10 candidates. And so she no longer had to have someone go through it by hand themselves. They can, they can use an AI model to do that.
What I was what I was able to discover for her on her behalf, because she asked me to come in and look at where there may be issues with what she's doing, is that it was automatically, this model was automatically discounting anybody who was sending in a CV written by the free version of ChatGPT, [00:41:00] regardless of how good they were.
So what that meant, and I explained to her, is that if you think about it, the people who can't afford the $20 a month are probably the people who were in your situation when you were growing up, and that really struck. It made her all... She had to choke back tears. She's "I don't want to do that. I want everyone to have a fair go.
In fact, I want to give these people more of a go because they deserve it. They need a leg up." So this is a lady who didn't even know. She wasn't doing it on purpose. She wasn't being classist. We have a horrible class system in the UK, which I love the fact that the US isn't bothered about class, but in the UK it's horrible.
She comes from a very humble working-class background, and yet unwittingly, she was now discriminating against the very class that she came from. I've gone into businesses where there are biases w- in terms of race, in terms of gender. These are lovely people. They are not wanting to be racist. They're not wanting to [00:42:00] be misogynistic.
They're not wanting any of these things, and when I point them out, they're horrified. They just had no idea that their data was skewing in particular ways. So the reason for the Human First Responsible AI pledge was we don't have a law any specific AI laws in the UK. Which in one sense is good, because sometimes laws can be draconian, and it can hamper innovation.
But the flip side of that is no law, and there's always gonna be some annoying human beings out there that will just go, "Oh, we can do whatever we like." And that has happened, where people haven't really been thinking about the consequences on other people's mental, emotional, and physical wellbeing.
So the pledge is a way for businesses to come together, and it's a very simple 10-point pledge. And if you can just go through that and just think "Yeah, I can do that." And there's things that are not even to do with technology. For example, one, one, one pledge is, it's N for neighborhood.
We've basically done an [00:43:00] acronym- and it's human. So there's two... It's 10 pledges, two for each letter. So two for H- yeah, we're getting b- you can tell I'm a former copywriter. I have fun with that, had to put an acronym in and all the rest of it, had to do that. But for the N, for the final letter in human one of it is neighborhood.
One of the pledges is neighborhood, and that is what can you do to make sure that AI is a tide that lifts all boats? What we have at the moment is that people are able to have, able to get pro versions of Claude. They're doing really well. But even if you're using a free version, the fact that you even have a computer and you're online all the time, there are people still in, y- beggar's belief, being a first world country, the UK and it's the same in the US, first world country, and yet there are people who can't even afford their own laptop.
They don't have internet access in their homes. They have to go to a library or somewhere to get intermittent, internet access. These people are falling even [00:44:00] further behind because of AI. So one thing that I do is once a I literally, I'm looking at it right now, next to me, right opposite my office there's a school, and right next to school there's a little community center.
So one Friday every month I go there, I take a few of my laptops and anybody who wants to come, I can teach them how to use large language models, and they get to use... 'cause some of these people don't even have laptops. So it's what can you do? And I als- say it as, and before you think, "Oh I'm not sure if I have time to do that in my business," yes, you do.
Some of the biggest I look at posts on LinkedIn a- and places like Facebook and Instagram, and the ones that get the most traction are the ones where you're not going on about how amazing your business is. Not everyone's doing that. It's when you're talking about things you're doing for charity, for the community, voluntary work that you're doing.
That gets so much traction, and then you start attracting the right kind of clients, the right kind of talent as well, [00:45:00] especially with Gen Z. One of the biggest things is they want to work with businesses that have real purpose, and they're not just in it just to make a pot load of money, so it's a real shift in how the generations think, and the younger generation, it's all about, they want purpose.
So e- even if you don't, even if you're like I can't be bothered with charity," do it for selfish reasons. Do it because it's gonna actually add to your bottom line to be thinking in terms of community and values and environment and that sort of thing. So yeah, it, it is y- but to be honest Anybody, you can go to the website.
I'm sure you'll put the links in there for everybody. You can go to the website and s- it's just a 10-point pledge, and what happens is you sign up to it, and you get an email with it as a PDF. You can print it. You can, put it as a screenshot, or whatever you want to do.
And then any time you're implementing anything new with AI, just go through and just... and if you can agree to everything, you go, "Okay, this is [00:46:00] a good use of AI." And if there's any points you think, "Ah, I'm not sure," this is where you need to stop, down tools, and fix that before you implement AI, 'cause otherwise it could be harming your business or inadvertently harming groups of people that you didn't even mean to harm.
Philip VanDusen: Thank you for doing that, A, because I think that this is sorely needed, and particularly because of this, this every CEO and leader of a company is drinking the AI Kool-Aid and not really understand, understanding where it can take them or their company, and to catch it before it catches fire essentially, to help set that groundwork or clear the kindling so the fire doesn't blaze too out of control.
I think this is really needed and and thank you to you and your co-founders for doing that. All
Mitali Deypurkaystha: right. Let me give the l- I'm sure you will put the link, but you just go to
Philip VanDusen: human- What's the link?
Mitali Deypurkaystha: Human First Responsible AI Pledge. So it's all one word, humanfirstresponsibleaipledge.org.
It's a nonprofit because we're [00:47:00] not... We don't wanna make any money from it. We just want awareness. That's one of the key things that we've found, that people don't mean to... maybe I'm seeing the world in rose-tinted glasses. There are sadly people who are, do mean to be racist or misogynistic or homophobic, but I really do believe it's a tiny number of people.
Most people are not doing that on purpose, but inadvertently they're doing that. So it's just open- just getting rid of those blinkers, and that's what this pledge does. It makes you look at your data and then go, "Hang on a minute. Maybe..." Like for example, a, another client of mine didn't realize that sometimes he was putting job posts on LinkedIn, but he was writing them, he's a very like a masculine guy.
And I was like, and he was saying, "I need more women in my, in this." And I looked at his job post. It's no wonder no women are answering to this. You c- I could smell the testosterone from here on the sort of screen. No wonder we're not... and he had no idea. He loves women. He's got three daughters, no sons.
He's "I adore women. My [00:48:00] life is full of women." But he didn't realize the way he was coming across was wonderful to men, so all these men were attracted, but women were like, "Oh God, this doesn't... Yeah, I don't think I want to." Sounds like a
Philip VanDusen: macho culture here. I can smell
Mitali Deypurkaystha: it, yes. And it isn't. I've been to his offices.
There's nothing macho about it. It's a lovely, inclusive, welcoming culture, but he didn't know he was doing that. And then as soon as he knew it, it was like, "Oh my God, I had no idea." And then he changed it, and now he's attracting women into his business and he's so happy. So it's just, people are not, and that, that's a guy who absolutely doesn't have a misogynistic bone in his body, and yet inadvertently being misogynistic without realizing it.
Philip VanDusen: Yeah.
Mitali Deypurkaystha: So therefore it's just, it just allows you just to open the blinkers. So yeah, it's humanfirstresponsibleaipledge.org. You'll find it, you can sign it, you can download the PDF, and then yeah, just disseminate it out to everybody in your business or even beyond your businesses in your network of businesses, and just get it out to as many people as possible, 'cause we have to.
When there's [00:49:00] no law, like I said, it's a good thing on one hand, but it's a double-edged sword. It's a good thing it doesn't hamper innovation, but the flip side of it is it can and we've already seen in the news, used irresponsibly. So the more people we get this out to, the more people can start thinking about how we use it responsibly.
Philip VanDusen: So when you came to s- speak to the Bonfire community I'll tie up with this, you offered everybody a free cheat guide of probably the top 20 or 30 cool AI tools that they could try so they wouldn't feel overwhelmed, but that would actually be something that they could start to approach the subject with.
Are there three or four, favorite tools? You mentioned SaneBox, and it sounds, I'm gonna try that one out. Yes. Are there other ones that you think are valuable that are just like a nice little nugget you could leave them with and something they could try or s- play with that could be helpful in their business?
Mitali Deypurkaystha: Yes. The problem is, like I made that list and I pr- I could probably update that list now 'cause things move so fast. I probably need [00:50:00] to create a new list every month really. But if I say one tool that I really feel that I'm getting the absolute most from, that has to be Claude. It really is.
First of all, from an ethical point of view, the fact that they were willing to w- walk away from a government that was essentially saying- Huge ... "We want your models," without any caveat- Yes ... without guardrails, and they went, "Yeah, no." I've gotta respect them for that. But outside of that, even if Karat, you're like, "I'm not really bothered about the politics side of things, don't talk to me about politics, Mitali.
Let's just leave that at the door." Just from what... it's the most creative large language model. If you're a copywriter, if you're a bit of a wordsmith, you'll find Claude far more elegant than any of the other large language models, especially ChatGPT. But also things like Claude Design has just released last week.
You're now able to just create slides and visuals and inc- infographics in Claude, and crucially, [00:51:00] this is how different they think from OpenAI, you can then open it in Canva and tweak it and do things with it. Which OpenAI is not even thinking of. They're just showing off really. I'm not saying, I still do use ChatGPT here and there, but I've just found Claude more eloquent and more designed for business users.
I will say, if you're just a person who's just using AI, yeah, go and use ChatGPT. But if you're a business owner, even as a solopreneur, you could be like a s- just a brand expert or designer on your own, you're gonna get so much out of Claude. Claude Code, stop it. A lot of people are scared of that.
It's I can't code. I can't code." That's the whole point. Go and start messing on it and start asking it, "Oh can you create a landing page for me?" And start tweaking and working with it. It's amazing what it can do and the more you use it, the more it learns about you and starts to understand your personality and what you're looking for.
So it gets to a stage where I always, you know, whenever I do workshops, one of the key workshops that I do is helping people engineer prompts. But [00:52:00] then it's quite funny because then I'll go onto my laptop and I'll give it just two sentences and people look up, "Hang on a minute, you've made me write two paragraphs and you've just put two sentences in and managed to get even better than me."
And I'm having to explain to them it's because it's learned so much about me, just like a human being. Think of like someone that you know is close to you and all you have to say is two words and they know exactly what mood you're in, exactly what you need. Whereas a stranger, you're gonna have to give them a few paragraphs to, for them to understand where your head's at.
It's the same thing. So you start off with these large language models really, putting that foundation work in the same way you do with a human being. Really letting them know what your thoughts are, what your outcomes are. If you're a business owner, who are your clients, what keeps them up at night, what keeps you up at night?
Really getting that deep work in to get to the stage where you can just give them a one sentence prompt and they get what you want, so I, I don't wanna, I definitely, I could, I can give you 10 or [00:53:00] more different tools, but I'd rather just stick to get into Claude, really start using it, even the free version.
If you can afford the pro version, even better, and really just start playing with it. They've also, which I've not tested so much yet, they've got a agentic browser as well. That's a whole other conversation, but I love agentic browsers because it's marrying large language models into a browser so I can use an agentic browser to actually do things for me while I'm on LinkedIn.
So one of the key things that I do every morning is I've got so many connections on LinkedIn. I can spend the whole day just reading all the posts in the last 24 hours and I'll get nothing done in my business. So what I do is f- one of the first things I do in the morning is I open it up on my agentic browser.
Claude has an agentic browser and then I just say, "Can you go through my feed and find 10 people who are my ideal client profile who've posted in the last 24 hours and I need to make a comment on their post?" And it goes off, finds the [00:54:00] people Then creates a comment, then says, "Is this comment okay?" If it's okay, great, go and stick it in.
If it's not, can you please tweak it a little bit? Tweaks it until I'm happy, and then posts it. And I get so many people who then send me a message saying, "Oh, thank you for leaving a n- a nice thoughtful comment on my post." I've been meaning to speak to you,
Philip VanDusen: and here we go again, that human connection.
I've been meaning to speak to you. I'm so sorry. We did start a conversation a
Mitali Deypurkaystha: couple of months ago, but that kind of petered out. I'm really sorry. Let's start it up again. And it's that human connection. So go and check out Claude. I've yet to find any large language model that beats it overall. I'm using Gemini a lot of the times for images 'cause their NanoBanana, a strange name for a model, but NanoBanana model is so f- at the moment unsurpassed.
So sometimes when I really wanna create amazing graphics or infographics or images, I'm going back to Gemini. And also because I use Google Workspace, it's already infiltrated in all my [00:55:00] spreadsheets and emails. So I use it then 'cause it's already there. Why would I not, why would I use anything else?
But if it weren't for that, I don't think I'd be using Gemini all that much apart from the NanoBanana model. It's because it's already just there, so why not use it? ChatGPT, I'm getting to the stage now where I did I, have I even been on it in the last two weeks? Maybe once a week or that. So I'm not really impressed with ChatGPT all that much now.
So I would say if you really wanna focus on one large language model, Claude. That's the one to really get involved, really get to know. They're just astounding me with all the different things that they're bringing out on it. It's just incredible.
Philip VanDusen: Mitali, this has been such an amazing conversation, and I'm sad that we're gonna have to tie it up.
But thank you so much for coming to the Brand Design Masters podcast and sharing your knowledge, your deep knowledge of AI. And also f- that last nugget that you gave us about the a- agentic browser through Claude about how to utilize that in a really functional way on LinkedIn to [00:56:00] make a business connection that is sorting through all of the dreck and giving you the highlights of a to-do list very automatically.
I think that's the sort of thing that we need more of from the AI community- Yeah ... is something that's really functional,
Mitali Deypurkaystha: that's what it is. What worries me is when people use AI for, there was a, an article a few weeks ago in the UK where the agentic browser had basically, someone had given it access to their internet banking.
Don't do that. Please don't do that.
Philip VanDusen: Oh, God.
Mitali Deypurkaystha: And then someone had hacked into the browser- Then taking out all this money from this bus- It's so well, don't do that. Don't give it access to no. Don't, we're not there yet. We haven't even put the security around this. So I'm using it mainly for businesses.
Go into my LinkedIn and connect with people, send a connection request, send a, put a comment on someone's post. And I'm watching, I'm usually drinking a cup of tea, and I'm watching it do things, and I'm s- giving it like, "Do not do it straight away. Come back and show me who that person is. Then don't send them a message straight [00:57:00] away.
What is the message you wanna send? Okay, that's good. Go ahead." Pretty much like directing an assistant. If you do it that way, you can't really go wrong.
Philip VanDusen: Good. Keep us away from the Terminator bots at this point- Yes ... Mitali. Okay.
Mitali Deypurkaystha: I
Philip VanDusen: appreciate that advice.
Mitali Deypurkaystha: I'll do my best. All right.
I'll stick my superhero cape on and I'll do my best.
Philip VanDusen: I encourage everyone to go connect with Mitali on LinkedIn. Her company, Impact Icon is amazing and doing incredible work for medium to enterprise level clients, and she's specializing in companies in the northeast of y- the UK.
Mitali Deypurkaystha: That's the go- that was the goal
Philip VanDusen: Or globally But now it's- Or globally
Mitali Deypurkaystha: I've got interest
Philip VanDusen: now- It's globally
Mitali Deypurkaystha: from a couple of businesses in the UK, in the US, and also Australia, which I'm not sure how I'm gonna do that with the time difference. But yeah, I will, 'cause because of the power of Zoom and things like that look at us. There's a whole-
Philip VanDusen: Yeah ...
Mitali Deypurkaystha: Atlantic Ocean between us- Yeah ... and we're here, we're having a chat like old friends.
So yeah, just go to impacticon.ai. It's pretty simple, so impacticon.ai. You can find my [00:58:00] details. You'll find my calendar, so you can even book in to have a quick chat with me.
Philip VanDusen: Thanks again for joining us, Mitali. I'm gonna have you on s- again sometime. There's this is just scratching the surface of this topic.
Thanks for coming and talk to us. We'll see you next time.
Mitali Deypurkaystha: Thank you.
