Unlocking the Power of Personal AI: Redefining the Nature of Identity

AI is big. The news is full of stories about ChatGPT and Large Language models. Is AI going to replace you? Maybe.

Hey everybody, this is Chris Brandt here with another FUTR podcast.

Not everybody is worried about AI replacing them. In fact some folks are looking forward to it. Imagine if instead of dealing with all the mundane tasks in your life, you could hand it all of to an AI version of yourself, that looks and sounds like you and would respond the way you would.

Today we are talking with Suman Kanuganti, CEO of Personal.ai that aims to allow you to build an AI that is based on you. So we are going to chat with Suman about how it works and where it is all going

Welcome Suman

https://personal.ai

Reach out to Suman directly at https://s.personal.ai

Click Here to Subscribe: 

FUTR.tv focuses on startups, innovation, culture and the business of emerging tech with weekly podcasts featuring Chris Brandt and Sandesh Patel talking with Industry leaders and deep thinkers.

Occasionally we share links to products we use. As an Amazon Associate we earn from qualifying purchases on Amazon.

Transcript

Chris Brandt: 0:00

Hi Suman, this is Chris Brandt. Really excited to do this podcast, just trying to figure out how to get it started. I'm thinking about doing an intro where I text your personal AI. What do you think?

Suman Kanuganti: 0:32

Awesome!

Chris Brandt: 0:38

Hey everybody, this is Chris Brandt here with another FUTR podcast. Not everybody is worried about AI replacing them. In fact, some folks are looking forward to it. Imagine if instead of dealing with all the mundane tasks in your life, you could hand it all off to an AI version of yourself that looks and sounds just like you and would respond the way that you would. Today, we're talking with Suman Kanuganti, CEO of Personal.AI. Personal AI aims to allow you to build an AI that is based on you. So we are going to chat with Suman about how it works. And where it's all going. Welcome, Suman. Thank you for having me, Chris. I'm excited to have this conversation. We talked earlier and, uh, you know, the stuff you're doing is really interesting. And I think in the world of, you know, AI chaos that we currently live in, this is a really, really interesting topic, right? Um, You know, you had a really interesting personal, you know, talking about personal AI, you had an interesting personal story about why you started personal AI. I was wondering if maybe you could, we could start there and you could explain what your thinking was.

Suman Kanuganti: 1:45

You know, the genesis of the personal AI was more about, you know, my desire to have an intellectual as well as emotional conversation with a lost mentor, as well as my previous co founder and my previous partner, my previous company, his name is Larry Bock. Uh, he passed away, uh, because of pancreatic cancer. Right? Um, and he taught me a lot. He taught me a lot in a short period of time. And, uh, we had an amazing 18 months of building the company together. And I had this mantra of like, what would Larry do once he passed away? And always like thinking through this, you know, strategic and the thought process of, uh, him was really helpful. And over a period of time, I always thought like, you know what, I wish I could talk to Larry's AI. Can there be a situation, can there be a model that we can algorithmically create that is grounded in one individual person that is grounded in their facts, in their opinions, in their point of view? Yeah. Stylistic and their voice. I think was all part of the Genesis story and, you know, that's the reason why Personal AI exists. And this was back in 2020, so this was before. AI was a thing, uh, but our company started off with solving this problem of being able to have access to people, both intellectually as well as emotionally.

Chris Brandt: 3:02

I love that. Like, what would Larry do? You know, that's, that's, and there's so, so many people in our lives that, you know, we would love to have that, that ability to, you know, just go back and ask. Um, so, so tell me a little bit about, I mean, cause this is a very. The AI is, it's different than a large language model. You're, you're using a foundational personal language model, right, is what I understand. And that kind of operates differently than what we're seeing with the chat GPT large language models and things like that. Um, but how do you get it, how do you get it to be personalized? How do you get it to have your voice or the voice of somebody, you know, you want to wanna replicate. Suman Kanuganti: So the most interesting thing about AI that I've learned is, of course, it's all about the data, right? Right. Uh, for this context of the conversation, we can consider the data as memory. You know, the memory of the internet is a DataCorp of the internet, which is technically what a large language model is trained on, such as chat, GPT. So you can think about it as, okay, it has the knowledge of the internet and it has some sort of a collective voice. The idea of like unique voice, unique identity comes with unique memory that individual person possess and the unique facts and preferences and opinions that you possess. To be able to do that, it comes back to the memory of you. So the data that goes into this model is actually not from the internet. It is the data that comes from an individual person. So it's not a pre trained model as much as You come into the system, you start creating your AI, you tell your AI everything about you in the form of these memories, you know, your favorite things, your favorite books, you know, how we, how your knowledge, your opinions, your views, your perspectives, and over a period of time, it will not only know who you are and your facts, But it will start to sound like you, you know, both stylistically, as well as from the voice and the facts and, uh, opinions that you take. So the more memory that you have, you know, the smarter it gets, the more it becomes you. I do use ChatGPT a lot for, you know, just writing things. You know, helping it spark, help me spark ideas and things like that. And it's great because it, it's thing, it'll come up with things I hadn't thought of. But the challenge I do have is that like when it writes something, it writes it in such a different way than I would approach it that, you know, I've got a lot of rewriting and things to do because it's like, it tends to be very like, everything's big hype and it's like, I'm not, I'm not always wanting to be, you know, that gregarious in my writing, you know, so. Uh, you know, like having something that more closely resembles my style would be phenomenal. If

Suman Kanuganti: 5:52

you are using TRGPD, you can probably try various different prompt techniques and be explicit about how your style is, how, how, how the style in the voice you would want it to be for a particular piece of task, right? What we are trying to be James

Chris Brandt: 6:07

Joyce, right? In the style of James Joyce, but I am no James Joyce.

Suman Kanuganti: 6:11

Exactly. Our goal is to be more implicit about it. Meaning... Your model should already know your style, your voice. Your memory, your facts, and when you are even ideating with the external world, it is still personalized to your choice, you know, very simple examples, right? You may have specific food preferences and generally, you know, when, yeah, when like someone meets you and like Chris, you know, where do you want to eat, right? Normally, foundationally grounded within your, you know, food choices and food preferences. Now you can ask Charjeevity, Hey, you know, what are some of, you know, favorite. Uh, food places, uh, specifically around sushi restaurants in, um, you know,

Chris Brandt: 6:56

New York. Sushi in New York sounds really good, so I'm down. But

Suman Kanuganti: 7:01

if your memory already knows that your favorite thing is sushi and there may be some specific things and specific allergies, right? It will be... a response that is tailored to you even though, you know, that information is coming from the internet. But then if you do have like specific restaurants in New York that is our favorite to you, then you can add that to the memory as well, right? So if I like this restaurant, what are the other restaurants I would like? Or simply straightforward answer, hey, what do you recommend for me to go? It's like, hey, go to this, this is one of my favorite restaurants. So I think it's just more about like how we live every day. Yeah.

Chris Brandt: 7:34

We've heard a lot about large language models, and I think everybody's. I'm just generally confused about what that means, but, you know, we're introducing a new term here yet again. Could you explain to people, like, what does that mean, a foundational personal language model?

Suman Kanuganti: 7:49

Foundational personal language model, so it's essentially like a ensemble model that uses like A large language model, by definition, uses, you know, pre trained set of data and uses generative transformer to create this model. Uh, what we do is, uh, our, our model is 120 million parameters instead of like 170 billion parameters of a typical GPT 3 model. Uh, and the reason is our purpose is not to learn on, you know, the internet of data that exists out there as much as To learn on individual sets of data. And this, this is also like grounded within one, uh, one's memory. So the priority for these models is more about, you know, the given intent or the given context that is coming to the model. Is it, uh, can, you know, can, can AI model can be constructed using that data? If not. Then I would go consult into a large language model, get the data, and then personalize it. So it's a unique model that is trained on every person, and it is continuously trained based on the memory that is being pushed. And given how small it is, it is very effective, and it is also cheap, if you will, compared to some of the, you know, super large compute. My analogy always has been like mainframes versus personal computers. Mm. Uh, mainframes had, you know, huge, uh, machines and it's almost like shade between all the people. And then you're at the of, you know, the application that mainframe supports and you go in and you, you know, type in the document and you get out. Right? Right. But, but when the personal computer came, everybody has their own use. Everybody has their own Yeah. Configuration, if you will. Uh, so that's kind of the personal AI model. I think one of the misconceptions that exists out there is the more the data, the better the model is. Uh, that is true when you are thinking about being able to handle like very general broad use cases for everybody to understand biochemistry and marketing at the same time. Right. Um, it's not true. It's, it's basically depends on can the intent, um, you know, that a particular model is getting, can it be addressed by the, you know, the model of the data corpus. And given now we have large language models, we can, you know, create a nice unique experience on, you know, hey, everything is personal until you don't know that information and we can consider a large language model. And, uh, use that information to personalize it further.

Chris Brandt: 10:14

If you're going to replicate me, I don't need that much data because I don't, I don't, I don't know as much as ChatGPT. Exactly. So, a much smaller model is probably more appropriate. But you mentioned something that's interesting about that because I know that when you look at like some of the, like ChatGPT can only handle so much context before it's kind of got to kind of start over, right? And the thing that's different about the person foundational personal models is that you're constantly integrating that data back in to the model, right? So it's, it's, uh, I want to say in some ways it has a perfect memory of every conversation you have since, you know,

Suman Kanuganti: 10:50

it's almost like, uh, it's almost like, uh, the context is not limited to in that moment task. The context is your long term

Chris Brandt: 10:58

memory. It opens up a lot of more options for different things.

Suman Kanuganti: 11:02

And also, like, more exploratory, like, within your own, like, individual person. Uh, one interesting thing that I observe is, you know, people who come from the chat GPT world, who are looking for personal, you know, personal AI or personalized AIs, uh, they don't have to think too much about a specific prompt and queue, all the different variation of the context. Meaning, you probably will still tell, okay, Hey, you know, based on your food preferences, hey, what is the food that you like? But you, you will not say, hey, here are all the food choices and food that you like in one sentence. And then you send it to chat GPT, you get the response back, and then it is forgotten or it's lost, right? Yeah. So I think that's the beauty. The prompting idea gets even more simpler, more natural. to human.

Chris Brandt: 11:49

The prompts are built in because it's based on, on you. Exactly. Yeah. Yeah. We got down a real deep rattle really quick talking about all the possible uses for this. I love the idea of replacing myself to do, take care of some of the mundane communication tasks that I have in a day. In fact, I would, I would love to have several of them just, you know, sitting in on, on various meetings. Um, but you know, Could you talk about some of the cool use cases that you've seen, you know, that you've been part of?

Suman Kanuganti: 12:22

I will give you maybe three to four different variations, right? One person who is almost like 92 years old has been documenting his entire life. Uh, from 60s, 60s or 70s. It's like 30, 000 blocks or something. It's pretty amazing, you know, some of the habits that people possess, which I wish I had. And now he essentially built his entire memory, you know, that were written, that were in previous journals, multiple different notebooks, et cetera, et cetera, and created his memory and the model is up and running. Wow. Uh, his name is Hemant Parvek and his goal is that he wants to, you know, quote unquote have his presence for the people around him, for his grandkids, for his kids, and... Almost like if somebody do need to talk to him, you know, they can still talk to him. Now, his motivation, yes, could be legacy, but he also has like tons of knowledge, tons of political changes, uh, that happened through the history, um, and the specific knowledge that he possesses and people come to him for, you know, specific reasons and unique needs. So I think that's fascinating, uh, because he forced, uh, during the beta times of personal AI before we even released and, uh, built his entire AI around personal AI, which is pretty fascinating to me. Wow. We are currently focusing on a number of, uh, individual brands, or we call them, you know, small business owners, right? If you think about small business owners, not the enterprises, not the huge corporations, but firms where majority of their currency is their personal currency, right? Their income is based on individual brands, right? And the unique knowledge that they possess, you know, consultants, coaches, authors, uh, influencers, thought leaders. If you think about like large language models, right, it is a composition of like all this different data that is available publicly on the internet with consent or without consent, you just basically index it. Right. But for these people, their whole core mode is their personal data, personal knowledge, personal mode. For example, Jacob, um, he is an immigration lawyer and, uh, uh, he gets like, you know, thousands and thousands of clients on every, uh, every year. And he helps people through, like, various different situations and complex things. Uh, and he cannot fulfill the demand that people come to him. Just him taking all these requests, now he created a personal AI of himself, which is his AI version. That represents, you know, his personal brand and all his knowledge. So now it becomes more like a knowledge AI, right? Like people can come and talk and Now, almost like, uh, okay, uh, doing an interview of Jacob, uh, Jacob doing an interview of them and then finally get to the person where, okay, you know, this is the, this is something I can help you with. Yeah. And then I have parents as well. Like parents, like literally, uh, creating, uh, specific views and opinions and their thoughts. for their kids, you know, to talk to them, like, you know, even myself, I personally do that as well. Like, rather than my daughter talking to Alexa AI, who I don't have any control on, you know, what it says or when it says, and who is controlling behind the scenes, they're already, they're already used to, uh, talking to devices, talking to AI, right? So, why not, you know, have that, you know, connected experience to the people that you love? I would rather have her talk to dad AI instead of You know, and Alexa, yeah, which I don't have.

Chris Brandt: 15:55

Have you done that Alexa integration yet? I have done the Alexa integration. Hey, Alexa, talk, talk to dad.

Suman Kanuganti: 16:01

Yep, um, I also done the, uh, text messaging integration. So, um, I can actually, you know, play the video to you. But, uh, uh, the voice is also cloned after me. We haven't released the voice cloning as of yet. But it actually sounds like me.

Chris Brandt: 16:16

The gentleman who is like preserving all his memories is sort of a legacy project. It's very interesting because when you think about what is a person, you know, and, and are we the sum of our memories, right? And if you can input a lot of memories, you know, into an AI, you know, it does. Come closer to, you know, this sort of permanent existence, potentially, you know, it's, it's, it's really fascinating. And I guess that's sort of streams into the, the metaphysics rattle that we probably shouldn't go deep down because we'll, we'll get lost in that one. But, you know, the other thing that, that really strikes me as interesting too, is Um, you know, you talk about, you know, people who have a lot of knowledge around a subject, you know, and that they could teach people. And you know, we look at, I love like, you know, masterclass. That's an interesting where you have like people coming on and sort of teaching how to do things. But, but there's no reason that couldn't be a personal AI teaching you those things. I mean, so in a way, once you get these things trained, you could. Utilize instances of these to, you know, as a way to monetize them, right? Or, or a celebrity, you know, like people always want to talk to a celebrity. I'm sure, you know, how many people would want to have like a personal AI of Taylor Swift that sounds like Taylor Swift that's based on, you know, her memories and things like that to talk to. Just opens up so many interesting, interesting areas.

Suman Kanuganti: 17:41

Masterclass education is definitely a very, very interesting area, right? I think it will probably take some time for being. people to kind of comprehend, I guess, the use cases as well as the methods that need to go into train. But once I take a class, it's almost like the continuity, right? Like, you know, 20 years ago, the professor that I, you know, that I worked with is still in contact with me. But, you know, I wouldn't, I would normally would wish I can have like easy access, you know, to her rather than, you know, me trying to email her or text her and hoping she will get back to me or even remembers me for that matter, right? So it's, it's about like this, like continuity that exists, you finish the course and then you continue communicating or staying in touch with that professor, with that, you know, person for a long period of time, right? It's like, Oh yeah, I do remember this, but hang on. In current situation scenario, it's like, okay, Yuval Harari in Homo sapiens book said something about this concept. Oh, I will never be able to find that remotely, even with all the search technologies that exist, but rather if. If I do have Yuval Harari's authentic AI, you know, that, that then combined with like, okay, my, my personal understanding of things, I could probably get to that, you know, comment or thought process much faster.

Chris Brandt: 19:01

Yeah, I can take parasocial relationships to a whole new level. The, the, the possibilities here are so interesting and, and, and, you know, it, it brings about a lot of really interesting scenarios too, because I would imagine if you had somebody's in a personal AI that you're conversing with, and over time, those reintegrated back into the personal AI, you're going to have. personality drift, I guess is the best way to sort of describe it. So at some point you're going to have to get a, get a 2. 0 that has all the latest personality features.

Suman Kanuganti: 19:38

If, if indeed, you know, to your original point, if who you are is the sum total of all your experiences in the past or some total of all your memories in the past, it may be also be anchored. The person who you are today, or the behavior that you are expressing today, maybe is influenced heavily with what happened in the past month, or yesterday, or last week, right? Traumatic event. Yeah, traumatic event, yeah. Uh, talk to me about it. So, the memory that we built is time bound memory. So in other words... It favors the recency in sort of this personality and the facts and the opinion. So even the knowledge, right, like what you think about or how you talk about AI, like maybe two years ago is probably likely, you know, it's likely different from what it is right now. Even, you know, the use cases that I was talking about personal AI to an investor or in a podcast, like, you know, two years ago is likely different from what it is right now, right? Um, so it does keep track of that. additional dimension of it. And that's what personally I makes it very personal because it's also time hard.

Chris Brandt: 20:41

The kind of metaphysical, you know, like kind of conversations that sort of can come into this too, because, you know, as this AI evolves and becomes more intelligent, you know, the value of it grows and, and the way it integrates within your life, you know, could, um, Lead to just different ways of thinking about how AI is governed and how AI is used and, and, and the, the, the value of AI. I mean, it just, it just really, I didn't feel, I didn't think we were going to be. Here yet, you know, I guess and I didn't think I would have to think about these things as much but now it's really opening up Some really philosophical kind of conversations, especially when you are basing it on an individual

Suman Kanuganti: 21:33

So the way the way I think about it is, you know, most like people, you know We are very much favorable to getting things done faster and instantly gratification So we end up like using, you know, the tools and the services to get that log return, to get that task done. It's more like a service, right? We use most of the AI like a service. Like you go to Amazon, you, you know, use Amazon's AI to recommend you a product and you buy the thing and you get out. Uh, the thing about personal AI is it's, it's the mindset of creating an asset. It's, it's almost like investing in yourself. Right. The more memory that you build, the more valuable your asset becomes. And we are giving that ownership of the asset to you. So it's not just about the physical assets anymore. I tend to feel for the first time, I have a mechanism to create a digital asset. That is mine, that is not, you know, spliced and diced and rented to multiple different platforms, you know, having conversations in one platform, posting in one other platform, blogs in one other platform, audio in one other platform, you know what I mean? Yeah. The more memory it grows, the more it becomes valuable to you and you would eventually carry this asset into your trust and you know, you share with family and friends and it lives forever. It's like a living asset of you.

Chris Brandt: 22:50

I think to some people, this might be kind of a scary idea. Um, you know, that they're putting all their information into this virtual entity and, and, you know, I mean, can you speak to, you know, like what are the, you know, kind of privacy and security implications?

Suman Kanuganti: 23:07

You know, I always say like privacy is subtly different from security. I guess we normally tend to combine those two things. So from a security standpoint, yes, you know, it is secure. as secure as it is, you know, your Google services or Apple services. But when it comes down to privacy, from my perspective, privacy comes up with, comes with control. Like what kind of control that you have on your data, on what data is going in, and where is it going to. Traditionally, I mean, Internet 1. 0 or 2. 0, if you will, we are the platforms to get a service out of it. That means you don't have control over your data and how it is used or how it is used for other ads. For me, the control is more about, do you own the data? Do you own the model? Do you have controls to be able to do that? Uh, and do you have a choice, you know, when to use the platform, how to use the platform and how not to use the platform. And that's, that's the route we have chosen. We, you know, if you go to our terms of service, it is pretty clear that, you know, you get to keep the data. The data will not be used for. Uh, any other purposes other than training your model, uh, eventually in the future, you know, when we are able to fulfill our roadmap, we can download the model and run it on a device, right? So you don't have to keep it on the cloud per se, you are living in a device that is at your home. Uh, so I think, I mean, those requires like some development, uh, but eventually given. You know, how small and how efficient these models are. That is the path that we have chosen to do.

Chris Brandt: 24:37

It is such a fast evolving area of technology. And it has, it's one of those things that, um, I don't think I've seen anything this impactful. In technology, maybe in my entire career, but certainly not in a very long time, it has so many broad reaching implications, and it's it's really rapidly getting integrated into everything. And I and I do think there's there's parts of it that are scary in that regard, because I don't think a lot of good thought is always going into where it's being applied and how it's being applied. But, but I do love the idea of like being able to own and control your own data, especially when it's something as personalized as this. There are going to be people who are going to be utilizing personalized AIs that are, you know, going to, you know, we talk about. Uh, that, that, that concept of post humanism, you know, like what, what does it look like when, you know, we, we kind of approach the singularity and, you know, we're integrated with machines and things like that. But I, I, my contention has always been that we have already passed that point because the second we started sticking an iPhone in our pocket with Google on it, I don't remember anything anymore. I don't have, I don't have no idea what anybody's address is. I have nobody idea what anybody's. phone number is, I don't know how to, you know, get anywhere. I don't know, you know, like I just pull that down as I need it. And it's an extension of my brain. And I think that's why when you don't have your phone with you, sometimes it's really nerve wracking, right? Cause it's, it's like, you just got a lobotomy. Yeah, this is that, this is this on steroids.

Suman Kanuganti: 26:13

You know, one thing I tend to think is. This idea of technology in our lives is always going to be a foundation. And then the experiences that we create or we gather over a period of time will be built on top of the technology. In other words, how much of our technology will suck some of the human traits that We used to have, you know, in the past life and take that with generations and generations, if we will create and we will experience things on top of the technology, like we were only talking about my five year old, like being able to talk to an AI, right? Uh, so that means, you know, how experiences and how creations will be built on top of. those foundational technology that exists as she growing up, you know, when I was growing up, I mean, our technology was what, like a bicycle, a scooter, and maybe a car, you know, when I was like 22 years old, I grew up in India. Uh, so that's the level of technology. It's like, and if, uh, if, if my bicycle broke down, it's like, Oh, I'm terrified because I cannot go to school. I cannot get things right. Yeah, you have to figure out like how to walk. So I think we can like make a statement or we can, you know, think like many different things that happened in the history. Uh, and we are, you know, the cycles of technology is happening more and more faster, which I think is okay because I think humans are also like trying to keep up with the adapting of the technology. So as long as we have this open mind that we do not fear technology, we embrace it and we build on top of the technology. Yes, the shifts will happen. The economy will change. The jobs will change. Uh, but eventually, uh, we will always be, humanity will always be built on top of the foundational technology that exists for generations.

Chris Brandt: 28:00

In order to participate in, in The future you're going to have to utilize some of these tools, because I mean, I think about it in terms of like, you know, right now, technology like for security products, we're seeing a I integrated into basically every security product, right? And part of that is because. One, you know, humans can't react fast enough to, to do, to deal with things, but they can't, humans can't also parse the amount of data that's necessary to just understand what's happening in any given moment. And I think that's analogous to some of the things that are happening in our lives. And there's so much information, so much data coming at us that our brain can't. Parse it all, you know, and at some point we're going to have to have something that's doing a lot of that lifting for us so that we can, you know, stay connected to everything that's happening because I, I believe that the pace of technological changes is, you know, like I think our, our ability to adapt to change is sort of linear, whereas the pace of technological changes, especially when you introduce artificial intelligence, that's sort of. Yeah. a multiplier factor on that is sort of this Malthusian curve that just, you know, skyrockets to the sky while we're just kind of, you know, putting along this very linear path that we're on. And unless we do something to, you know, get us caught up, you know, we're going to fall, fall behind, right?

Suman Kanuganti: 29:25

Well, what I think will happen is when the technology picks up, it picks up rapidly and the adoption for humans is going to be linear. But then the technology cycle will also start plethoric, and then humans will catch up. And then there's the next wave. It's almost like a connecting S curve that is happening, whereas human adoption is going probably like a linear. When there is a gap, that's where the changes are happening. The shifts are happening. The laws are changing. The jobs are changing. You know, we are having these conversations in the podcast. Um, same thing happened with the internet and then happened with mobile phones and then happened with like communications, right? Like now my mom and dad like, you know, uses internet and social media like everybody else as well. But for initially they were like, Oh, you know, I don't know what is going on. Like, what is Google? I remember my dad asking me and now he asked me like, what is AI? Right. But eventually I think, you know, that linear curve and then once AI start to plateau and gets integrated, we will catch up. And then there'll be the next curve, which is robotics. Synthetic humans, sir.

Chris Brandt: 30:25

What's next for personal AI? What is, what's next on your radar, your roadmap here?

Suman Kanuganti: 30:30

Right now it's, uh, it's all about going to market, like getting personal AI in the hands of as many people as possible. Educating the market that this is a mechanism to take control of AI into your hands. Uh, you know, put it to your benefit. Um, current use cases expand into people like replicating themselves to represent in everyday communications. You know, we do have these modes of like co pilot as well as auto pilot. In our definition, Copilot is, you know, human in the loop, meaning AI which just will draft messages for you if you send me a message, and then I can review it and, you know, send it to Chris, even though, you know, it sounds like me. And for some situations where it's more like, you know, team engagement or audience engagement or community who would benefit from You know, my knowledge and interacting with me that could be autopilot where AI automatically responds to a person. Uh, but all these conversations are then like going into this reinforcement learning, you know, continuously. So if I do, you know, if you do ask me a question that is not in my memory, I can basically address it and then add it to the memory as well immediately. So there is this like the continuity nature to it. So it's not only investing time in. Just creating your initial model, but like an ongoing basis, you're reinforcing a factor, an opinion, or a thought and continuously, you know, building this asset for yourself for every day. So from You know, everyday communication, text messages to emails, to people talking to you, having conversations. Those are all, uh, the current possibilities. We have native applications, such as iPhone app. Android app is coming out. Uh, there is desktop and mobile applications as well. Uh, there is API. So think about this API as an API just to you. So technically you have your own, your, your own server and for data output and data input. Um, so yeah, so people, um. are discovering it. It's a, it's a, it's a unique, uh, different, uh, product than most exists out there because it's not a pre trained model. Uh, but we have integrated like large language models as well into personal AI. So people will come in at least, uh, uh, we'll have like the base. public knowledge, like, you know, public information, and now they can start building their personal AI. That's kind of where we are, and, you know, our goal is to make this an asset for every person and make it as cost effective and as cheap as possible. So it's not just for the richest, not just for, uh, you know, uh, Taylor Swift, like you mentioned, right? I think, I think large language models will take care of Taylor Swift. I think our goal is to get the 80 percent of people, you know, where their knowledge, you know, may not be widely popular. But they are still important to humanity overall and to people around them.

Chris Brandt: 33:17

If people want to get started with, you do have a free version of this that people can get started with, right? Right. There's different tiers.

Suman Kanuganti: 33:24

The system, uh, functions with the memory. There is a, uh, free version. You know, you can think about free version as like a consumer experience where you can add pretty much your hobbies, your food preferences, your personal preferences, your, you know, book preferences, anything that you would think. Um, uh, and there is a limit to the memory, uh, and that's all free. So that's where you get the experience of how this model like starts turning it to sound like you for some of the personal preferences. And then as you increase the memory, then it goes into a paid subscription. So 15 and 40 and even we have. Uh, a, you know, custom solutions for, you know, small business and major brands where they have lots and lots of data. Um, but yeah, but they get unlimited memory, you know, at 40 per month and you can create like multiple different personas. Uh, so think about it as like, okay, do you do it yourself model, which is essentially everything is self service and pretty easy. You basically put a message and add a message as a memory or your conversations as a memory and, uh, the model will start learning so they can try it out. Uh, there are a whole bunch of other AIs. In the Discover tab that people, we are starting to like give people to publish their own AIs to discover as well. But people can experience, understand, add their memory, and then figure out like, okay, what is their purpose? Uh, and we have a training course for helping people. You know, get to their eventual goal as well.

Chris Brandt: 34:48

If you just go to personal. ai, um, you can get access to that and get started, right? Yes, of course, yeah. Okay, so everybody go and make your own personal AI. I love that.

Suman Kanuganti: 34:59

And you can obviously come and talk to me as well. s. personal. ai is my personal brand domain. Uh, so if you come there, it's like me plus my AI, depending on if I put you in copilot or autopilot, meaning you can talk to my AI directly. Uh, you just talk to me and then, you know, I use my AI to assist me in that moment to, you know, quick respond to you, faster responses to you.

Chris Brandt: 35:21

So if you're, you're really eager to talk to Suman, get on SSTEP Personal AI and have a chat with him. And it may or may not be him, but who, who knows? This is such cool stuff. And I would love to talk to you for a week about all the, you know, philosophical. You know, thing, directions this could go. Um, it's very cool technology. It's a cool thing. I'm, I'm really excited to see where this goes and where you go with this. So, um, you know, keep up the good work, keep me posted. And thanks so much

Suman Kanuganti: 35:52

for being on. Thanks a lot, Chris. It was wonderful talking to you as always. Uh, had wonderful conversations so far. Thank you.

Chris Brandt: 35:58

Thanks for watching. I'd love to hear from you in the comments. And if you could give us a like, think about subscribing and I'll see you in the next one.