#25: AI helps you with your goals

Hey, this one's going to be a fun one. You know, I. With AI, we're starting to see a whole raft of agent companies, you know, that are going to use AIs to do various things. But you got to start with a goal. And SOCRA is on with me today and that's going to be a lot of fun to talk about. This little company that caught my eye. Who are you?

Yeah, you want to start?

Yeah. I'm Eduardo. I'm originally from Brazil and I've been in the US for about eight years now. I went to UCF and got a degree in mechanical engineering and since then I worked in semiconductors, energy and 3D printing. And now I'm doing soccer and I enjoy traveling and exploring the world, especially if there's nature involved. Working out and building soccer with Mike.

That's good.

I'm not as she actually grew up in a third world part of North Brazil, which is kind of crazy. But I grew up in Colorado and let's see, what have I done? I, I used to work at semiconductors as well. I worked at Texas Instruments and then Apple after that as a test and validation engineer. But my passion during the whole time is building really cool apps and products. So while I was at those companies, I was building cool web apps and things like that. And then a couple years ago we decided to take the leap and build socra and it's been a pretty crazy ride since then.

Did I nail this? That this is a new way to talk to a computer and split, split up your go. Your service is you put your goals in there, right? You say, hey, I want to lose weight or I want to get a raise or I want to complete this project. And then the AI starts giving you ideas of how to do that. Well, the AI is splitting that bigger task, bigger goal up into little piece little things. Like I just said, hey, I want to increase my reputation and said, well, go over to LinkedIn and update your LinkedIn, right?

Yeah.

Tell us about the goals of the company. What are you trying to do with this? And do I have this right, that this could be the user interface for running a whole bunch of different agents that are going to do different things for you?

Yeah, I think that's a great question. If I look at a 10,000 foot view of what the industry will look like in 10, 20 years, I think there's going to be a few really big trends. One of those trends is how people are actually communicating with computers. Right now we're talking to computers and AIs through something like ChatGPT in the future, it might be getting closer and closer to us just thinking something and then the computer being able to act on it. So I think that's one really big trend in terms of how people are going to communicate with AIs and computers. Another big trend that's going to happen in the next 20 years is like, how are people actually going to communicate with each other? And I think that system is going to evolve. And then the other thing is like, ultimately you want the computer to do something for you, so how are you going to orchestrate tasks and having the agent do something very complicated for you in a nice way so that you can just sit back and watch the thing do the thing. Do you have anything to add?

Yeah, I was going to say that right now, since it's the very beginning of agents being useful, because up to now, agents are not able to complete large tasks by themselves. So it's very important that you can break down those tasks and that the agent knows what it can handle and what it can't. Because what it can't handle, then it has to assign to a human or assign to a different agent, like, let's say specialized agent. If this agent is like LLM based, cannot create images. Right. It would assign to one that has that capability. Or there's tons of people right now building agents for all sorts of things. Right. So that's part of the division for SOCRA as well, is to allow people to create those agents in the platform and have them serve multiple purposes.

Yeah.

And ultimately you start with a goal, but that goal could be literally anything that you want to achieve. And in the future, I envision a world where the AI is going to be able to, it's going to become more and more and more capable. It's going to be able to handle more and more and more of your everyday workload. The things that you're handing off to the AI are going to be more and more and more. And so I think building a system that can help orchestrate these things, this human AI collaboration, is really what SOCRA is all about.

Yeah.

If.

What do you, you're a small company, you're not a big open AI with a, you know, billions and billions of dollars of capital you're trying to make. Well, you did tell me where this company is. And, and, and what are you learning about AI? Because you're, you're using AI underneath your own product. Right. You're not building the model that you're using. You're, you're Using the models that other people, other companies are using to make something valuable to the, to, to your customers. Right?

Yeah, that's exactly right. I think one analogy that I really like to give is kind of when the Internet first started, the kind of the base infrastructure layer was laying down all the cables and getting all of the servers capable of communicating and implementing this thing that we call HTTP now, which is how the computers are communicating with each other across a global network. Right. So in the beginning of computers and the Internet, people were basically building the core infrastructure layer. We see the models, the AI models, the generative AI models, whether it's image, text, multimodal, whatever, as that core infrastructure layer. And some of the really cool applications are going to be built on top of that core infrastructure layer. So you know, it's, it's really great that the, that ChatGPT can generate text for me and it makes sense and that I might be able to get something from it, but what could I do with that text in a more complicated scenario that, that something like ChatGPT can't really cover? So that's, that's kind of like what it's all about with, with this trend that I'm seeing with, with how, or this, I guess this allegory between the Internet and how AI is now. It's just like right now we have this core infrastructure layer that is super capable and, and I think we're going to see a lot of really big companies that can build on top of that and maybe they might even be like two person companies like us who could scale into very big solutions and markets.

I was just going to add that another one of the main applications that we are building SOCRA for is to help us build Socrates. We, our dream world, Socrates, handling 80% of our work for us while we sleep. And we want that to be a reality, not just for us, but to everybody else because we think that's the promise of AI. The promise of AI is to create a world of abundance, not in the sense of necessarily more mature wealth, but a world that's not a zero sum game. A world where everybody can succeed. Where like currently, right now, most businesses fail, most people fail to achieve their goals.

Yeah.

So that's, that's the whole reason why we are rebuilding Socrates.

What, what are you learning about goals by doing this? You know, the, the human task of, you know, oh, I, I, I want to run a marathon in a couple years or I want to, you know, change my job, or I write something like that, or I want to change my Career or, or I want to you my kids, you know, like teach them soccer or something like that. What are you learning about school, about how people do goals and, and your own use of the, your own tool? What, what are you learning about goals?

Yeah, yeah, it's, it's kind of crazy the number of things that we've learned about how humans complete goals, how people set goals, achieve goals, work towards them, etc. Etc. Through this platform that we've been building for over a year, year and a half now. One of the things that really, that really sticks out to me is that the neural circuit in your head for goal achievement is effectively the same. Whether it's a big goal, a small goal, an intermediate goal, a technical goal, whatever the goal is the circuit in your brain that's actually wired for helping you make progress towards that goal and the way that you feel and your determination and, and what you get out of it is basically the same. So that's one of the big things that we've been working on is like how can we optimize this, our platform, so that people are kind of working with that neural circuit in the most efficient way so that people are making the most progress towards their goals and all of that. There's so many other things that we've learned about goals. Maybe you've got a couple, but I think that's the one that sticks out to me.

Yeah, I would say main learning is that most popular goal is to lose weight. I was actually very, very surprised that third was learn Python. I was not expecting that so many people were interested and makes sense to use AI for that sort of applications. It's learning. Programming is one of the best applications for AI at the moment. So, and another thing is like we, we had people use the platform for many things that we didn't expect expect. So that's even like for one time we, we met with an investor. It was like our first time ever meeting with a potential investor. And it seems like one of the things that they didn't like is that we didn't have a niche of a goal, but we started to see people use it for different things we didn't expect. And that's when we decided we are going to keep this agnostic and let people use to bring any idea they have to life or any goals they want to achieve. One in particular that comes to my mind in the very beginning of SOCRA was this woman, she sent us a very nice message that she was using Socrates to prepare for surgery. She was going to have a Surgery. And she was not very comfortable with doing the whole planning or talking about it with her family, and she didn't want to be very dependent on that. So she wanted to prepare up front and have everything set up. And she said that SOCRA was very good to help her with that. So you will have some cases where. Where it's a goal like that that might take you just a week of working and planning. But another very common goal in the platform is to start a business. That one. It's a goal that has multiple goals within itself. That's why the thing that we focus most on soccer is hierarchy, structure, and hierarchy, because work is just hierarchical by nature. And that's how you take big challenges, by breaking them down into smaller steps.

That's. That's awesome. Have you learned tips for. Well, let's get back to the weight, because I lost 60 pounds, and I did it by one, giving up alcohol, you know, not drinking alcohol, which was a big, big change in my life. But, you know, all when. When I drink coffee now, I don't have milk in it. Those are my only two changes I ever made. And I lost 60 pounds. Right. And the mil contributed to that a lot. Does your tool go through and, you know, if I say, hey, I want to lose weight, does it look at both the big picture, like, are you drinking a lot? You know, versus the small picture? Oh, do you want to do something small, like change? You know, go to no milk in your coffee in the morning? Right. Does it start splitting up? Like, because part of getting somebody to lose weight is making it simple for them. Right. And making it. Make it. Making. Making something that they can change, that they feel empowered to change. Right. And change for a month.

Exactly. The small habits, one of the biggest challenges that we had. And also, I'd say I. I saw another. Another company that was like AI for goals, and. And recently, unfortunately, like, they decided to stop pursuing that because they actually encountered the same problem we encountered ourselves, which is people just losing momentum and losing motivation, even though they. They are able to, like, use AI to find their starting position where they want to get and. And find the habits that they should have or the things that they should do to get to that point. It's usually like, they start strong but then lose momentum and give up as it happens in real life, with your goals in real life. Right?

Yeah.

So, like, for us, like, a big. A big jump in users in the platform was around New Year's resolution, and we got to see that the same thing happened where a lot of People said New Year's goals and they give up, which means that we haven't solved this problem. And that was back in January that we decided to rebuild everything from scratch. It's crazy. It's a crazy amount of work and we are right now like finalizing this new version. But we decided that how can we solve this problem for people? An AI coach is not going to do that because AI coach was the solution that we had. And we would tell people, you can always talk to your coach whenever you need some extra motivation and so on. But that was not helping because I think that people are still not at a point where they can relate to an AI like they can relate to a human.

Yeah.

So we rebuilt this new platform around collaboration. So our features, everything is around collaborations because we think that this will be a solution that will help other people stay motivated because they will have accountability from other humans that are working towards the same goal as they are. So that's one of the main things that we want to, to have for Socra now and focus on. Because, for example, you have two excellent tips of cutting alcohol and drinking your coffee black. Those are things that you could share with the community and others would know. So not only you can have accountability, but you can also share what worked for you. Because that might work for somebody else as well and might be like a small thing that they are willing to do versus some other bigger, very strict diets and so on.

Yeah. Your scenario reminds me of a study that I read recently. But the study is basically that just being aware of the amount of spending that you have, like in terms of your finances and whatever, just being aware of the number that you're spending, you actually spend like 20% less. So just having that goal and being reminded of it is already a big influence in terms of how you're going to make progress towards that goal. And so I think as we're moving towards this new platform, the things that we like we're really stressing is that it's. You're collaborating on goals with other humans and other AIs so that you can be constantly reminder when you want to be reminded about the things that you're trying to achieve. And the little nudges here and there add up over time. Just like the little, the little milk in the coffee helps out a lot too. So yeah, that, that. I think that pretty something. It all sums it up.

It's interesting that you talk about being accountable to other human beings. I talked to a researcher in Israel who studies addiction and he said the Number one indicator of somebody who is going to stay sober is are they accountable to other human beings? And so that's interesting that you're an AI company and you're trying to build a community with AI, the technology to bring human beings together. You know, that's sort of an interesting thing. You know, a little company like yours. You know, I used to work at a programming magazine and a lot of the advertisers would build little tools for Visual Studio and then Microsoft would copy the tool and do away with their business model. And it seems like a lot of little companies like yours always have to live in fear. Oh, is OpenAI going to do this and add, you know, goals and tasks to open AI, to chat GPT and sort of take away a lot of what you're trying to do? How do you think about that and how do you innovate and stay ahead of them? Because they're, they're pretty aggressive. You know, they have big plans for making operating systems and devices and, you know, things on our face soon that you're going to talk to all day long. How do you, how do you continue to be relevant in that world?

Yeah, that's a great question. I, I have a quote from one of my, my favorite entrepreneurs, Naval Ravikant. I'm sure you've heard of Naval, but one of his next. One of his, yeah, one of his main keys is to escape competition through authenticity. So the idea is that if you can build a platform and a community that is authentic, where it's homegrown, people are there because of other people on the platform there, or people are there because the platform has something unique, some sort of feel, some, something that is not really quantifiable or replicable. Just like how you could replicate any number of features on a web app could be replicable. The idea would be if you can make it authentic and people collaborating together on their goals in a space that they're comfortable with and all of that, that's actually harder to replicate than you might think, than, you know, you wouldn't just be able to throw money at that and build a whole bunch of features and then, you know, take the network. So that's one of the main things that we're, we're looking for when we're trying to make sure that, you know, big companies aren't taking any of, any of our potential users or people that might use our platform. And let's see, what do you have any.

I think the main, for us is just making it the best that we can. We are deeply passionate about making it as best as we possibly can. And I think that gives us an advantage on just having the drive to always try to find better solutions, even if that requires starting all over and building again. We always want to continue improving the platform. So it could be the case that, that, let's say that they caught up to something we are doing now. We are already thinking about what we want to do next to make it better. And I think that passion is another one of those things that you cannot replicate.

Another obvious one is to just stay away from adding features that you know that they're going to add in. You know, like there's a lot of features that, you know, that ChatGPT as a message, a future messaging app with AI is going to be adding in to help make messaging for between humans and AIs a lot easier. But from that same token, an app like ChatGPT has a designated purpose for it, which is to chat with AIs. The farther they deviate from that, the more likely they might lose customers and lose a reason for existing. So they're also tiptoeing a kind of an uncomfortable line where they have to make sure that the thing that they're building is helping people chat and, and do that the best way with AIs.

Yeah, you caught my eye in social media on X dot com, you have a unique style of talking with people and trying to get people to see the product. Tell me a little bit, because that's one thing that small companies have to do. They have to find customers and you were doing a great job of that. That's why you stuck out to me, stuck in my head because I keep seeing you on social media doing things with, with customers and people. That is unique. Tell me a little bit about what you're trying to do there. And does AI help you with your social media?

Thank you. I think that the main thing is I try to speak my mind personally. We don't have much marketing or marketing team or anything like that. Our actual focus is building the products the best that it can be and we believe that that will help us grow. By default, if our users love what we have, they will want to bring others along and collaborate with that and so on. We think that that is something that's more within our control. Also. It's something that we love doing the most and what we are best, best at. I'd say not, not as much the marketing side of it, but just building socra. So that's what I, I try to do. Just speak my mind on what we are doing and how I'm feeling about what we are doing. And that has helped us attract people to, to try the platform.

Oh yeah, there are other companies, I won't name them, but they're, the presentations are very inauthentic and they're very well scripted. And you can tell that they've been through many, many, many meetings and many rounds of script changes and that their presentations and slides are so, so well made that they're just optimizing for potentially getting people to click on a link. We don't want to do that. We want to be authentic. We want to make sure that whatever we're saying, we really feel it and we really mean it. So those, those types of things I think really help people on social media. That's one of the reasons why I think there's so many like small creators that end up getting big followings because they speak their mind and they, they, they, they feel authentic.

When I put a goal in and it starts splitting it up into separate things to try. Tell me a little bit, get nerdy with me what's going on under the hood and what are you learning about AI by building this? Because it's AI is deeply embedded, used in your service, right?

Sure, yeah. It's a pretty complicated agent. Depending on the state that it's in. It has a finite state machine of a whole bunch of states that it can potentially be in to help you either make progress on a goal or remind you of something or break something down into smaller steps. But at its core, when you are trying to create a goal, the system needs to know what you're trying to do. So it's trying to figure out what exactly you're trying to do and where you're coming from, what is your background. From there it can help you figure out whether you know plan A is the best way or maybe plan B is the best way or plan C is the best way. Once you figure out a plan that you is optimal for you, similar to how when you're using Google Maps or Apple Maps, when you select a route for how you want to do it, like I want to go bicycling or I want to go on my car. Once you, once you've decided the route that you want to take, our system gets to work breaking down and planning that route for you and your goal so that you can basically have the steps that you need for it. And then you can complete the steps or plan the steps recursively inside of the platform. Because each, each step is itself. And you're talking about journeys in this feature or in this, in this example, because each step is itself a small goal, you can again break down that goal into smaller goals, plan it, execute it, et cetera. So under the hood, it's a pretty complicated finite state machine with a pretty complicated tree structure to help you basically plan recursively down a tree and then complete things back up the tree to make progress towards your goal.

You bring up something that I learned about AI. If you ask it, Write me a book. It'll try, but it won't be a very satisfying book. Right. If you tell it to write you a scene in a book or a chapter or a table of contents, in other words, a smaller task, it does pretty well with that. Right. And it seems like you figured that out intuitively. What, what other tips have you learned for. From working with AIs? You know, because I don't build companies with AIs, I just, you know, tool around with chat, GPT and stuff.

Yeah, I would say, like, one of the main things that I have learned is that the entire infrastructure and tooling for this industry is brand new. There are libraries being created every single day that supplant previous libraries that are capable of doing things. There are like, at its core, just being able to generate text that makes sense and decide on things is an extremely powerful thing to have in an LLM. So really, I think one of the next things that you'll see in the next five years or so is people being able to create compositions of tooling that use these LLMs to decide things. So I would say at its core, like finding the right tools or building them ourselves, we typically build everything in house. We don't use, pretty much, we really don't use many other libraries for this. We build a lot of the tooling and agents in house because we feel if we have full control over those, those different parts, then when we put them together, it comes up into something that is, I guess, more. More valuable, more. More meaningful and more effective.

I'd say some of the tips too, that we've learned from that. One thing you notice is Socrates. He's helping you there plan your journey and he's asking you questions.

Yeah.

And that was one of the things that we figured out early on that if the AI system can ask the right questions to get the right context it needs for that specific thing you're. You're trying to achieve or trying to do, then nobody needs to be a prompt engineer because the AI is prompting you.

Yeah.

So that that was another thing that we tried to do as well.

Hey, yeah, it's something like I'd like to plan a vacation to Spain next year. Right. You know, it should come back and say, well, where do you want to stay? Or, you know, what's your budget? Or stuff like that. Right. You know, it needs a lot of data. And this is one thing, you know, people talk about privacy and we should talk about privacy because, you know, when people are putting their goals in, that's that's a very private information. You know, there's some of that those goals might be very personal. You don't want your, your wife or your kids to see them even maybe. Right? Yes. Yeah.

When it comes down to this is a tool that we want to use ourselves and we take privacy very seriously, especially me. Like I use, you know, all the, you know, private, private Messaging, private browsers, etc. Etc. I'm a very techy guy. That's privacy is very important to me. So we, we make sure that, you know, all of the data that we have about users that could identify users is very, very secure, et cetera, and that we don't use it for anything else other than the helping the user make progress towards their goals. It's not used for training any models or anything like that. Once you, you have your data, you opt in to either sharing it with others or keeping it private. So.

Yeah. And how do you store the memory? Because you know, this is service and if you put in a goal, it's there tomorrow. Right. That's chat. GPT doesn't really remember things like that. Right. It, it's starting to, because they just are starting to add memory. But how, how are you thinking about that and how is that going to. Because soon we're going to talk to our AI all day long about different things. Having the memory of things like the questions you're going to answer really helps the system out and helps the AI help you.

Yeah. So we have a few approaches that are already implemented and a few approaches in the new version that will differentiate us quite a lot from memory. I don't want to say too much, but in general, your brain uses a few different ways of storing and retrieving memories. Your brain even augments memories when you retrieve them. So based on the recency of the memory, under what category it is, what types of references, so on and so on, memories can be categorized and filtered and sorted and ranked based on their relevance in a given situation, under a given context, and so on. So one of the main things that we want to do is, is make the memory feel really authentic and make it feel like you are actually talking to an agent that knows you or a friend that knows you, that has experienced the same thing that you have. So this is one of the things that we have in plan, is a really comprehensive, what we're calling a response pipeline or socrate actions combined together. But the idea is a very complex multi step process that can grab the right memories, format them in the right way, discard memories that aren't needed for a given situation, and then generate a response that makes sense for you in a given situation and format it the way that you would want to be talked to. So there's a lot going on there. I think if you look at like the OpenAI example with ChatGPT, they're basically grabbing single sentences here and there and they're storing them in like a global memory for you, which is not particularly useful for a given task or a given subset of tasks. So one of the main things about our memory is going to be categorical memory. So like, is this memory even relevant? We're talking about fitness, but why am I providing a memory about like relationships in, in the context window? You know, there's not really a good need for that. So there's a lot of different techniques that we're going to be using and that we are using in this upcoming version.

I was also going to say that ChatGPT memory is not necessarily memory memory because it's summarizing something and adding that so it can remember later on, which means that it does not have the details our system. Instead it traverses your entire tree of journeys and everything that you have worked on. And it rates each of those memory, ranks it according to weight and the criteria that he mentioned. Another one is how often that memory is recalled. So it knows that memory is important because it's using it many times. The ones that are not recalled very frequently are going to be more forgotten. How it happens with our own memory and in relation to privacy, one of the biggest challenges that started with us is like, how are we going to solve privacy if people are collaborating? Because we don't want to be collaborating on a goal. And now the AI is using my memories and letting you know. So the way we solved that was by implementing a collaboration system that has granular access control. So you're able to give granular access to specific, let's say very specific folders, each folder and content that you have. You're able to set different access to different people and different types of access and the AI. If we are collaborating and the AI is working with you and I together, it's only going to use the memories that we have in common. So for example, Mike and I have a gardening Socra together and we have a journey to build Socra together. Whenever I'm working, even in a new journey with the AI, the AI would be able to recall those memories that we have in common from gardening, from building Socra, but nothing else. So basically the memories on the things that you work individually because you're able to work goals and things just yourself and AI, the AI will keep just and use that and be able to access that just in those circumstances. And then when you are working and collaborating with others, it's only going to use the memories you have in common.

Yeah, and I think this is a really important point is let's say you're communicating with your co worker on Slack and you and this co worker have a good rapport and you're talking about something you're trying to do, you're trying to achieve something, right? But you realize you need a third person or I guess a third person for now, but it will be an AI soon. A third person to help you out. Well, that person doesn't have that rapport between you two. So you basically have to describe that problem to them as if they haven't heard it before. Which I think is an important privacy thing. Like when you and I speak about something, the only memories that we have are the ones that are shared together. So that's a really important point with, with our new system where we're going to be having AIs and humans collaborating together is that you really, really, really only want to have memories that are, that are shared across all people in the conversation or in the goal, etc.

So it's an intersection, which means that if you, so we are having group, you can invite as many people as you want to work with you on a journey. That's how we're building the system. Now let's say that you are working with 100 people together on a group. That means that AI pretty much we only have the memories of that specific thing because what are the chances that 100 people also intersect memories in something else? Right? So it's basically like the, the bigger, the larger the groups you get, the less memory it's going to have. If Mike and I are working just the two of us on things, the AI will have a lot of memories that it's going to be able to use soon.

I have Humanoid robots. I visited a humanoid robot company last week and they spilled some coffee on the table and then said, hey robot, clean up the, clean up the table. That was a goal, right? Keep the table clean. But the human AI today isn't self directed. It, it didn't know to clean up the table before somebody told it to clean up the table. Right. Somebody had to give it a goal and keep the table clean. Right. How are you, let's say it's, you know, three, four or five years from now and, and are you going to be running my robot? Am I going to robot some goals like keep the house clean, you know, that'll be a goal for my robot, you know.

Yeah. This is one of, one of the features that we have for our new release that we're calling SOCRA Actions. But SOCRA Actions are basically a really complex action based system that are, and each action is always triggered based off of specific events. So in your scenario there, the event that happened was somebody spilled coffee and the action that the AI determined that it needed to do was to go grab a napkin or whatever and then wipe it up. So with this kind of like action and event based framework, you can accomplish quite a bit because when you think about goals, most things are happening after something, some sort of event happens or you do something and then you should figure out what you should do next. This is also how humans work. Or maybe you forget to do something for three days and then you should be reminded of it. It's another event that could be happening. But the idea is events can trigger actions and workflows that you need to go do or that the AI can do for you. So I think in five years this type of, we might consider this type of technology like every everyday normal, normal thing, but it's just definitely not right now. And it's the capabilities are going to be really, really cool.

I used to run every day to Las Gas Ferrari to stare at the cars and dream about owning one of those nice Italian cars. Right. I've seen a lot of entrepreneurs have a picture on their wall and remind them what they're doing it for. Their dream maybe that you want to buy a house, maybe you want to buy a car, maybe you want to start a family. These are bigger dreams that yeah, aren't quite so goal specific. They're, you know, buying a car is sort of a dream. Do you separate those? And does the AI treat you differently when you give, you know, sort of a big dream to it versus I got to take out the garbage today.

It's It's. I think it's. That goes back to the thing that Mike was saying. Like, it's the same circuitry, your brain, or it's. It's the same process that we're following. Even if it's something very simple as, like, taking out the garbage, then you. You have in your head a very basic, simple workflow of I need to go grab the garbage, take it out, walk, and. Right. The same thing. If I want to buy a car, I want to buy that specific car. I want to. I need to save for it, or I need to find a different way of making money so I can afford. Afford it, and I have to go.

We need to get a college education so you can get a good job, so you can get, you know, make some money so you can afford a nice sports car.

Right?

That's what. That's what humans need. Big, big, hairy, audacious goals. You know, Elon. Elon Musk wants to go to Mars. That's his big Harry go. I have smaller, big hairy.

Yeah. I think, you know, ultimately, when it comes down to it, a goal is something that you don't have yet, but you want to have, whether that is some physical thing or something about yourself, whether it's like learning something or like changing your body in some way, like getting more fit or, you know, going to college or buying a house or whatever. But all of these things require kind of sustained change over long periods of time. So, you know, none of these goals happen in one day or one week, except for taking out the garbage, because that can. That can be very fast. Right. It doesn't require a lot of neural rerouting in order for you to have accomplished that goal. So that's one of the big things that we're aiming towards with Socrates, like, how can we help build the agent and the AI in a way that's just giving people just the right little nudge to keep them helping map their brain into where they want it to be over time, over that. That time frame so that they can accomplish their goals.

Yeah. Do you. Does the AI go and get inspiration for you? You know, like, if you did this goal, here's what you would look like, or, you know, I don't know, and I don't know where I'm going with that.

You know, we actually had people ask for that. We had people ask. It's one of the. Like we. Last year, basically, with this current version of SOCRA that you're looking now, we tried a lot of things. So even everything that it's there now it's a small fraction of all we have built last year. And trying to see what works, what actually people are using and what is actually helping them. It was a lot of try and error. But one of the things that people asked was being able to see like, oh, I want to get in shape, Can I see a picture of myself in shape so I can feel inspired to actually pursue it.

So there's also another point to that, which is, and they've actually done studies on this, but seeing a picture of yourself having accomplished the goal isn't actually the best picture that you want to be shown. The one that's more effective is showing you a picture of what happens if you don't complete that goal. And so obviously that's a very like hairy line to tiptoe, because if you don't get that right, you're showing somebody them not completing a goal and it's sending them down a wrong spiral. So like, if you implement something like that, you have to be very, very careful. So all of these features that could potentially, you know, really help that may be the most effective. You also have to consider from like a user experience point of view or from like a human perspective, is this, is this like ethical to create in an AI system?

You know, that's why we decided not to build it, because we thought the other ones, it's proved to be more, to help more. But at the same time, we don't want to be giving something that would be harmful to somebody or that would make them feel down. And, and so we, yeah, we discarded that idea.

Yeah. But yeah, we definitely had these, these types of ideas. And like, there are ideas that you have to consider because these are some of the most efficient and effective ways of accomplishing goals. But you also have the human ethics part that you really have to consider because ultimately we're trying to build this to make the world a better place.

Yeah. Not destroy people. Anything else I should know about what you're doing and how you see AI evolving? You know, maybe talk about what, what you're seeing happening in the AI space for businesses like yours. You know, what are you hoping to see, you know, in the future?

I think in the next five or 10 years, we're really going to see this core infrastructure layer of generative AI being laid down. You're going to see the, the GPT model going to look a lot like the Gemini model and it's going to look a lot like Claude. So these models, you're going to see them kind of converge on Higher capabilities and being closer and closer to the same thing, which, which by definition is becoming a commodity, it's commoditization of that product. So I think what you're going to see from like generative text models is kind of a general commoditization over to this, this neural net that can really generate any kind of text that you want if you prompt it correctly. I think with generative images it's going to be something similar. Generative video, it's going to be something similar. There's going to be all sorts of variants and fine tuning in specific use cases to make them more effective, more efficient at a particular thing. But I think in general you're going to see this core infrastructure layer being developed and commoditized for other people to use. That's the equivalent of, in my Internet analogy, people building out the fiber optic cables or all of the cables and the server racks and building the HTTP protocol and you know, how are people going to connect with people with DNS and so on. And I think in five, 10 years, the companies that will build on top of that to produce these amazing agentic systems. I think Devin is a really good example of this. I think the GitHub copilot workspaces is going to be another example of this. But I think companies that can build on top of these things, that can kind of do a lower level bit of thinking, are really going to have a big advantage for a lot of things that they're going to want to do. One of our main goals with SOCRA is to replace as much of our daily work as we possibly can, make ourselves that much more efficient so we can focus on more creative things. And so that's one of the biggest trends that I think you'll see is that AI is not going to wholesale replace all jobs all at once, but you will see people becoming more creative, more efficient at the jobs that they already have and getting more done, which is going to help the economy and companies overall.

Does Socrate ever say, hey, I actually can do the marketing plan you want to get done, I'll go do it for you?

Yeah. In the new version, that is one of the things. It's going to have a big comprehensive set of things that it's capable of doing and then for any given context, any given thing that you're looking at, it's going to go figure out if there is something that it can do, whether that's something that a capability that we have implemented or in the future a capability that hopefully maybe a developer has implemented through an open source ecosystem that we plan on having in the future. But ultimately the AI choosing what to do is going to be the future.

Yeah, the AI choosing what to do is going to be the future. That's cool. Thanks so much. This is fun and it's a cool little tool. I think this, you know, starting with a goal is how we're going to control our robots, right? You know, hey, the goal is to keep the house clean, you know, now you have a goal. Now the robot, the robot has to figure out how to do that part of it. But humans need their own goals for their own needs, right? It's real cool. Where do we learn more about this? Where do we learn more about you?

Yeah, just socar. Com. The link is in our background.

Yeah, I see it. Thank you so much. This was awesome.

Thank you.