Behind The Bots

Max Jacob, the founder of TAP, joins "Behind the Bots" to discuss his innovative AI companion powered by Artificial Intelligence. TAP is a hardware device that integrates with operator, a software system designed to create a personalized AI experience. Unlike other AI assistants, TAP focuses on memory-based interactions, learning about the user over time to provide more relevant and helpful responses. By leveraging large language models like ChatGPT, TAP aims to become a user's best friend, understanding their needs and offering support in various aspects of their life. Max shares his vision for the future of AI hardware solutions and how Artificial Intelligence will shape the way we interact with technology. With its unique approach to AI companionship and the potential to enhance human memory retention, TAP is poised to make a significant impact in the rapidly evolving world of Artificial Intelligence. Discover how TAP is revolutionizing the AI landscape by tuning into this insightful episode of "Behind the Bots".

#ai #artificialintelligence #aitools #singularity 

TAP

https://heytap.tech/
https://www.producthunt.com/products/tap-11
https://twitter.com/heytap_tech

HOLE SYSTEMS

https://hole.systems/roadmap

FRY-AI.COM

https://www.fry-ai.com/subscribe
https://www.youtube.com/@TheFryAI
https://twitter.com/lazukars
https://twitter.com/

Creators & Guests

Host
Ryan Lazuka
The lighthearted Artificial intelligence Journalist. Building the easiest to read AI Email Newsletter Daily Twitter Threads about AI

What is Behind The Bots?

Join us as we delve into the fascinating world of Artificial Intelligence (AI) by interviewing the brightest minds and exploring cutting-edge projects. From innovative ideas to groundbreaking individuals, we're here to uncover the latest developments and thought-provoking discussions in the AI space.

Max Jacob:

So TAP is actually our hardware side of the brand. The brand itself is Whole Systems. What we're creating is basically a personalized AI companion, which will the the point of it is really the memory. So unlike, for example, Rabbit, where they have more of an action based approach, we're going for more of a memory based approach, which means that almost like a human, when you're talking to it, it will personalize itself over time to you based off your interactions. And TAP is our hardware side, basically.

Max Jacob:

So, on the software side, we're building out operator, which is the, what is going to integrate into TAP and basically allow it to have this memory.

Ryan Lazuka:

Are they both the they both need each other to work, or is operator like, can you use it right now on the web?

Max Jacob:

So so, actually, it's funny. So TAP is going to launch first before operator.

Ryan Lazuka:

Okay.

Max Jacob:

And you you will be able to use operator by itself. We're also planning on making an Apple Watch app for it. So but Tap is almost like an extension of it, so you could use it in a much more seamless way without having to constantly pull out your phone. There's a listening mode where, for example, let's say I would set it right now in listening mode for this podcast, and it would just take information and, in intake different types of context that it finds interesting.

Ryan Lazuka:

It doesn't do that automatically or do you need do you need to prompt it to say, hey, can you pull out like the 5 most important pieces from this conversation?

Max Jacob:

What it's doing is basically based off of your inputs. It's always extracting details

Ryan Lazuka:

Okay.

Max Jacob:

That it finds important. And over time, based off its personalization to you, these the details that it retrieves will get much and more much more relevant to you. So it will learn what what is important for you, if if in a way.

Ryan Lazuka:

So it's sorta like you don't you don't have to prompt it really because over time it's gonna learn what you want to prompt before you even have to prompt it if you could, if that makes sense.

Max Jacob:

Exactly. So, like, out of the box, it's quite like, like, a baby that you need to have it kind of, like, learn about you. And as it grows older, it starts actually being very helpful to you with connecting certain conversations you're having, being able to brief you on certain things, and eventually doing actions for you.

Ryan Lazuka:

Alright. Cool. And what what, like, got you into this? Do you have a background in, hardware or, like embedded devices or anything like that? Or what what's your background?

Max Jacob:

Well, so I'm an indie maker. The first project that I made was actually digital contact cards. Do you guys know the, like, NFC contact cards that you, like, take on? Yeah. So that was my first project.

Max Jacob:

And then from then, I started working on this other tool, which was called Bara. It was actually launched on Product Hunt. What we made was an, a tool for documentation for AI documentation. So it was really meant for college students to take notes. Compared to Otter and some other competitors in that space, we had something unique where it would actually take quite comprehensible notes over long tie over long periods of time.

Max Jacob:

So it was able to keep track of that, of conversations over long periods of time. And we basically just took that core, and we're remaking it into operator because we found that it's much more useful in that case.

Ryan Lazuka:

Sure. So you how did that tool work? Is did it, like, record classes from a a laptop or and then sort of create notes from there for students, for example?

Max Jacob:

Yep. That's exactly what I did.

Ryan Lazuka:

Wow. Awesome. How'd that go?

Max Jacob:

It went pretty well. We just didn't really do any outreach or marketing. What we did is I gave it to my friends and just family, and they found they just really like the notes coming out of it. When they had it recorded for, like, an hour or 2, let's say, they just have a conversation, and they want, detailed documentation in a specific format. So they were able to get that over 3 hours of conversation.

Max Jacob:

So that was really useful. But we realized that the system that we built out for basically, the context, the context system that we built out to facilitate that, we just decided to remake it into operator.

Ryan Lazuka:

Cool. Because they can sort of instead of doing just notes, it can do pretty much anything that you'd Exactly.

Max Jacob:

You just it'll just keep track of your whole life, basically, your whole memory.

Hunter Kallay:

So is there some platform that you can log into to kind of edit how you're, as far as the device goes at least, that you can edit kind of your conversations and store things on and stuff like that? Or is it how does that all work?

Max Jacob:

So it once it's integrated into operator, there will be what we will call a portal, and it will be a like a web interface where you could log in. It will give you not only will it give you just the data that's stored in there, it will give you much more detailed, connections between your data. So if it sees correlations between certain topics that you're working on so let's say you had a conversation here with a person that it listened to, and then you watch the podcast and it finds some interesting correlations between that that you could, work on or that it sees that you're, like, working towards, it will bring that up and showcase that, like, hey. This is maybe a interest point of yours, for example.

Ryan Lazuka:

It looks like you have a team of people. Is it just 2 people, Max? Or you got more than that?

Max Jacob:

So we have a team of 3 people, and we also have 1, person that's helping us out with the launch currently.

Ryan Lazuka:

Okay. And are you just to be clear to the people listening, are you the you're the you're the founder of this or is there a cofounder? What's your role in the company?

Max Jacob:

Exactly. So it's me as the founder and then my brother who is the cofounder. I'd like to say that I'm more of, like, an idealist, and he's more of, like, he's more practical, he's more like he's more of the CFO, I would

Hunter Kallay:

say. Okay.

Max Jacob:

And then I'll be more of the CTO.

Ryan Lazuka:

Gotcha. Gotcha. Well, I mean, that brings a a good point is, like, if, you know, if you're more like the CTO doing the hardware, is that that like, a lot of people have a lot of experience with software, you know, software developers, coders, programmers, but doing hardware is a completely different niche. Do you guys have a background on that, or you sort of just figured it out as you went?

Max Jacob:

Well, I have a background with talking with manufacturers and so does my brother, based off of, like, just back other stuff that we've done, business business stuff with, like, the family. So we're quite comfortable with that. So and we had quite a lot of connections, that we could reach out and ask, you know, can we make this? How can we make this? Let's you know, basically, at this point, batch 1 is a sample run, to see how the market will basically react, how people will use it, what will be the feedback, what features they will want.

Max Jacob:

And from there, we will continue building out the software.

Ryan Lazuka:

Awesome. And when is when do you guys look for batch 1 to be available? I know you guys have a it's Kickstarter. Right?

Max Jacob:

It's not a Kickstarter. It's just 400 units we're launching early April. Okay.

Ryan Lazuka:

My bad. Yeah.

Max Jacob:

Yeah. And we're launching on Product Hunt, and it's just 1st come, 1st serve for the 400 units. We're launching ready to ship. So as soon as you buy it, we ship it.

Ryan Lazuka:

Sweet. And those, you still have some available for the first batch if you if anyone wants to buy them right now.

Max Jacob:

Yeah. Yeah. The 400 units will go live early April. Like, they're all available.

Ryan Lazuka:

Awesome. I feel like anybody like, any hardware device for AI, like, instantly sells out. I mean, I hope you guys, you know, do as well, but it's all about marketing as well. So, I think you'll do very well. We'll see how it goes, but it's it's exciting to release a hardware product in the AI space for sure.

Max Jacob:

Yeah. Definitely. This is the 1st week that we started kind of actually publicly talking about it and, communicating with people and potential customers. So, and we've seen a very good response so far. Right?

Max Jacob:

People are very interested in supporting us and just buying the product itself.

Ryan Lazuka:

And it's pretty reasonably priced. What's the price again?

Max Jacob:

$59.

Ryan Lazuka:

Yeah. That's super cheap compared to, like, the rabbit, which is I forget what that is. Was that over $400 or something? But, in in terms of, like, you know, you alluded to earlier, the you have you're different than the rabbit, but can you maybe walk through some specific examples of, like, how your product works, TAP works, and what it does in relation to Rabbit and how how they're different? Like, maybe with specific examples, if you could give some.

Max Jacob:

For sure. Are you speaking to the current form factor of TAP or the one where it's integrated with operator?

Hunter Kallay:

For less

Ryan Lazuka:

when when it's integrated with operator because that's probably your long longer term vision. So I'd love to hear that.

Max Jacob:

Great. Okay. That's definitely much more interesting. So, basically, with Rabbit I mean, I don't have a Rabbit personally, so I I can only go off of what I've seen. But it seems to be much more action based where you could teach it, with their LAN model.

Max Jacob:

You could teach the rabbit to do certain things for you, like ordering Ubers and DoorDash, stuff like that, and other automations, you know, work based and stuff like that, which are definitely very useful. But we believe that creating a memory first approach will later on allow us to make the actions it takes much more impactful. What I mean is, basically, it needs to learn who you are before it could do stuff for you. You wanted to understand what you want as a person. Right?

Max Jacob:

It wants you you need it to be completely personal to you before it's you trust it to start taking actions on your behalf.

Ryan Lazuka:

Gotcha. So, like, for example, right now on just to make it a little bit more clear for the for the end user here or the people that might use your product one day is like, if you're using chat GPT, you know, their context is rather small right now. You might be able to upload a couple of articles, and it only knows about those articles when you ask a question about, you know, that article. If you wanna summarize the articles or do whatever you want with them, create highlights for them. But with your tool, the context is gonna be kind of unlimited from what I'm hearing you say.

Ryan Lazuka:

Like, it's gonna learn, you know, 5 if you have the same product, which I'm sure you'll have many iterations of it. But if you have the same project product now from Tap and you still had it 5 years later in that perfect scenario, like, it's gonna learn over that 5 year time, whereas, know, these chatbots right nowadays just know a very small context window. Is that sort of summarize it in in layman's terms?

Max Jacob:

Exactly. But the main thing that I would wanna from a privacy standpoint that I would wanna say is it's not really have it's not collecting a log of exactly what you're saying.

Ryan Lazuka:

Okay. What

Max Jacob:

it's doing is it's out of the things that you do say let's say you're discussing, a party you're going to on a specific date.

Ryan Lazuka:

Mhmm.

Max Jacob:

It's not going to collect the whole conversation. It's just going to be like, okay. That's an important thing. This date, you're going to this party, let's keep that. Let's, you know, save that.

Max Jacob:

And later on, maybe you say something, and I realize that you're saying that you're gonna go to something else the same day, and it's gonna tell you, oh, wait. You have a party that day. You know? So it's able to act as a reminder, as a calendar, and all of that at once, basically.

Ryan Lazuka:

That's really cool because most of the time, you would think it would need all the data all the time, but it's gonna learn what data it needs via your past conversations with it. So, like, that's a huge help for privacy. Awesome. In terms of, like, using it, is the device always it's always on listening, or how does that work?

Max Jacob:

So we will have different inputs. Most of the time, people will use, like, the conversation input where you just tap you tap press and you talk to it. You say something. You let go. It answers.

Max Jacob:

But we'll also have a mode which is, like, listening mode, which you could we're not sure exactly how you're gonna switch into it on the actual tap device yet. It'll probably be, like, a double click or something like that, where you'll basically let it, like, listen to a whole conversation. So it could be 20 minutes, and you're just letting it listen along. So there's different inputs, and, obviously, there's a privacy light to let people know that's it's active when it's working.

Ryan Lazuka:

That's cool. That that's one of the questions I was gonna ask you. And what about like, if say what's the battery life like? Do you know yet since it hasn't really been tested in the wild?

Max Jacob:

Well, I have the sample. It's sitting behind me, actually.

Ryan Lazuka:

Oh, sweet.

Max Jacob:

And my whole team has their samples. Operating time is 17 hours. Awesome. And that's if

Ryan Lazuka:

it's listening the whole time? Or is that

Max Jacob:

Exactly. So that's when it's actually active when you're talking to it and when it's talking back. When it's on standby, it's around a month.

Ryan Lazuka:

Holy crap. That's awesome. It's just nice that you can have a device that you can plug in overnight, and it will last all day. And that's what your product will do for the most part.

Max Jacob:

Yeah. That's something we say that it's gonna outlast you during the day. So

Hunter Kallay:

So how big is it, and how well does it listen? Can I hear a conversation that I'm having with somebody far away, or do the person have to be, like, right into the

Ryan Lazuka:

Yeah? We'd love to see it.

Max Jacob:

One second. Let me grab it. Yeah.

Ryan Lazuka:

No worries.

Max Jacob:

Here's the tap.

Ryan Lazuka:

Wow. It's so tiny.

Hunter Kallay:

So it's actually a lot smaller than I than I thought it was even off your on your website. So

Max Jacob:

Yeah. It's it's really small. So let me do you guys wanna see it work? And actually

Ryan Lazuka:

love that. That'd be great. Yeah.

Max Jacob:

Hey. What's the weather in New York like right now?

Tap:

The current weather in New York City. New York is 34 degrees Fahrenheit and sunny. The temperature is expected to reach 49 degrees Fahrenheit during the day.

Ryan Lazuka:

Sweet. It's really a very, like, crystal clear.

Hunter Kallay:

Yeah. Clear.

Ryan Lazuka:

I was about to say that.

Max Jacob:

Yeah. I was worried that you guys wouldn't hear that. But if you have headphones connected to your phone, it'll play through the headphones.

Ryan Lazuka:

Oh, sweet. So how does it work? Like, we I've got this. I don't know if you heard this. There's this I've got a a 2 year old daughter, and we have this toy called Grem.

Ryan Lazuka:

It's by, what's the girl that dated Elon Musk? His name, I forget. But, anyways, it's a AI toy, and you can talk to it, and it responds back. He uses chat gbt, and it connects to your Wi Fi, but it's always on. So, like, if I'm asked if I can't have it on and just have a conversation with my wife or have a conversation with my kids because it's always listening to all 3 of us talk.

Ryan Lazuka:

And then it will it will, like, answer questions even though I'm I'm asking the question to my wife. I guess your the solution that you guys have to that is you can have it you would just turn it off at that point, not have it the listening. You wanna have it on the listening, the feature that for you for your tool at least.

Max Jacob:

For sure. And besides that for operator, we're also working on a voice ID option. So in the app, you'll be able to basically create an ID for your voice. And in loud environments, like at a party or something, you'll be able to talk to it, and it will actually pick up specifically on your voice and know that that's the question that you're asking. And whatever else is going on is other people.

Ryan Lazuka:

That's awesome. That's a huge win because it's so annoying to pat things on. So that's sweet that you guys, thought of that. That's awesome.

Hunter Kallay:

I wanna ask your opinion on, you know, we see a lot of the AI hardware solutions gaining a lot of popularity. Like Ryan said, TAPS, we see it's your product, and then like we've alluded to before, like the Rabbit R1. We've seen the Humane AI Pin. We there's been rumors of AI glasses. What is kind of your vision for AI more generally as it makes its way into more hardware solution?

Max Jacob:

I think that something Sam Sam Oltman said in a Greylock interview is really our vision for this. Basically, what he was the question he was asked was what startups will win in the AI space? Right? And he said that that that he believes that there'll be a couple of foundational models. So you see that in terms of GPT from OpenAI, Gemini from Google, and there's a couple of more.

Max Jacob:

But besides those, that's not where the win for startups in AI is. It's really about building the middle layer. So not creating some wrapper app around gbt, for Instagram hashtags or something, but really creating something that is, you know, the way you interact with your computer, something that you talk to medical advice about, or, in our case, like a friend. Right? So something that is basically your best friend and understands you to that level where you can you don't have to really overexplain yourself to it because it understands you based off previous context.

Max Jacob:

And just like people understand each other based off of I'm sure you guys are able to talk to each other much more fluently than I'm than I'm able to talk to you guys just because you guys know each other for so long.

Ryan Lazuka:

Do you think the middleware that is gonna help start ups survive? You know, because like you like you said, there's only gonna be a few major players. It's gonna be a hardware interface like like you have that sort of, you know, could we like, the NVIDIA just released, with, Hippocratic dot AI, an AI nurse. Right? And it's I think it's uses iPads with avatars on the screen for nurses.

Ryan Lazuka:

But is that how you sort of envision it from what you're saying?

Max Jacob:

I believe that it could either be a software approach or hardware or both in our case. There's a lot to do to build around large language models and other AI models that will give you this, you know, middle layer approach if that makes sense.

Ryan Lazuka:

No. It it totally does. Yeah. It's like like because Hunter and I, we we're trying to go launch a few products in the AI space, and it's it's hard because you think, oh, I've got a great idea. And then you a couple weeks later, OpenAI released something that was sort of, like, put your startup out of business.

Ryan Lazuka:

So it's a it's a very it's a very different world to think about because sometimes you have to think about the future, but it's a risk to take if you're gonna build your whole interface around, like, OpenAI's, API or something like that.

Max Jacob:

Yeah. For sure. And I and I think that's something that, is a win for TAP, that the other competitors are kind of stemming away from

Hunter Kallay:

Mhmm.

Max Jacob:

Which is basically being powered by, these large APIs. I don't think that, humanes and I'm not trying to down them. I think their products are great, but the most correct approach to an AI wearable, I think leveraging what OpenAI and or Google has built out, and just kind of connecting your product to that, is the best way currently.

Ryan Lazuka:

One thing I that stood out from what you said is you said your tap is gonna be like your almost like your best friend. Is that is that sort of like kinda I o in a way because it's gonna know everything about you, but that's makes a lot of sense from what you described this product is.

Max Jacob:

So does that scare you, or does that excite you?

Ryan Lazuka:

I would say it, at first, I would the only reason why it's I I don't think it would it doesn't scare when you first mentioned that, it doesn't scare me at all. I think it's, like, cool. I guess the scary thing about it, and this topic keeps on coming up in our podcast over and over again, is the fact that why would anybody you know, if they ask it's gonna get so good, why would you wanna speak to anyone other than your AI best friend? You know, because it's gonna know you better than anybody else. It's not gonna fight with you, and it's just gonna be feel more comfortable.

Ryan Lazuka:

And, it's gonna feel more comfortable to talk to you with than than talk to a human counterpart. Even if it even if you love your human friend, your human best friend, you know, I can see a future where you don't wanna talk to anybody human with and I guess that's the scary part, but no one knows how this is gonna play out.

Max Jacob:

I guess I I see that, but I think humans like the imperfections in a way, a little bit. So I don't think there this is the end of humans talking to each other.

Ryan Lazuka:

Can I have, like, be be in the matrix with neural links hooked up to our heads and everyone's reading our minds?

Max Jacob:

At that point, I I think when we're sending thoughts to to each other and not speaking, that's the end of conversation.

Hunter Kallay:

Yeah. It's it's interesting, that that you said that because, Neuralink just came out or just in they've been working on this for a while, but their first patient was just able to play chess. I don't know if you saw that. And then, they're able to play chess and then also post on x, and that and they did that all without using they all they did was use their thoughts to do it, from a brain implant by Neuralink. And I'm like and Ryan and I were talking about this a little bit earlier.

Hunter Kallay:

I'm just like, woah. This is this is starting to get freaky. You know? Like, I'm all for the the chat bots and stuff. And now now we're using, like, this mind control where we're literally hooking stuff up to our brains.

Hunter Kallay:

And it makes me wonder, like, it's kinda I mean, it's obviously super cool, but it makes you wonder about, like, the implications. Like, is there going to be, you know, 10, 20 years from now, it's just a given that, like, when a baby is born, you just put, like you know, like, some people get, like, circumcised or whatever. You just put the little chip in their arm, and it's like, oh, do you want us to chip your kid? Oh, yeah. Go ahead.

Hunter Kallay:

They put the little, like, chip in their arm and you know, or or their brain or whatever, and they have this little chatbot that follows them throughout their lives. I don't know. It's just an interesting little thought. I mean, nobody knows where this is gonna go.

Max Jacob:

It seems like, for me, it seems like there'll be kind of this division, almost like a new Amish community, that that emerges. The, like, non AI, non body invasive kind of, like, tech, or, like, community, basically. So it'll be interesting to see. Actually, I need to think a little more about that, but that's an interesting thought.

Ryan Lazuka:

Yeah. A 100%. I could see that. Like, people, especially, you know, if it gets politically divided, people are just gonna take a side whether it's gonna be divided down the middle. Right?

Ryan Lazuka:

Your anti AI or your 4 AI, and then, you know, communities will form because of that, maybe. I don't know. We'll see. But that's I can I can see your point a 100%? I just saw something super interesting, and it said, you know, a AI is gonna be really cool when it can prompt humans.

Ryan Lazuka:

So right now, we prompt chat gpt or Gemini or or whatever chatbot we're using, But it'll be cool when the AI knows us well enough where it can prompt us. So it might ask like, hey, Ryan. I noticed, you know, you did this today. Can I help you with this? Like, is that something you think Tap could eventually do as well?

Ryan Lazuka:

Like, it will ask you out of the blue if you need help with something?

Max Jacob:

That's definitely something we are already thinking about. We're about 2 month 2 months ago or so so, we're talking with my developer about, basically, such actions which are forthcoming actions. The problem that we would have currently that we realized that would be quite a big hurdle is to know when we can say something. Where where a human, they have ears, and they're always kind of listening. When can I say something?

Max Jacob:

When can I not? In this case, we would always have to listen. When can we say something, when is you know, there's a conversation going on, or something else.

Ryan Lazuka:

Yeah. That's a good point. So maybe you could do it in you could do it in your listening mode. Right? Like, if you wanted to?

Max Jacob:

Definitely. So in our in in the current listening mode, in our demo, what happens is, inner thoughts pop up on the screen. So it's an it's a mobile application. So when it's in listening mode, if your screen when it when you're in the app, you will see, inner thoughts of the AI pop up, what it's thinking about. So, oh, that was interesting.

Max Jacob:

Let me save this part. You know? Things like that. In that case, maybe a nonverbal kind of, cue to ask something.

Ryan Lazuka:

Got it. No. That's that's awesome. I think I think it'd be a big help to have it's like having someone you know, you it looks like you have a clip on the device. You know, it's clipped onto you all day, and it's like it just like sort of I don't know.

Ryan Lazuka:

It's probably a bad analogy, but, you know, we're we're broken as human beings. We're not perfect. So it's like having a having a cast on your arm or something, you know, except for your it's actually that's probably that's like a negative way of looking at it, but it's gonna help you throughout the day. It's like having someone that can aid you at all times.

Max Jacob:

Right. Exactly. It's it's, you know how before talk shows, usually, there's a person that briefs, the talk show host on what's what's about to happen, like a like a summary, for example.

Ryan Lazuka:

Yeah.

Max Jacob:

Yeah. That's basically what this would provide for you, 247 plus a million other kind of benefits, but that that'll probably be one of the main ones is just being able to, for me, jumping into this podcast, I would be able to ask who's who's Ryan. Right? Can you give me some information on this podcast? Can you brief me on what what's kind of going to happen?

Max Jacob:

And it would just in before you know, for me, this would take around 10 minutes to research, at least. For it, it would take 10 seconds and, you know, you get the response.

Hunter Kallay:

It also makes me wonder about retaining information because think about how many conversations we have and the information that gets lost, throughout those conversations. Like, you'll probably have, like, 20, 25 conversations a day, and you remember and pieces of those conversations. When you're in the conversations, you're like, oh, this is maybe an interesting conversation. Afterward, you may you barely remember anything. Like, if you think back to throughout, like, yesterday and you're like, think about your conversations yesterday.

Hunter Kallay:

What did you, you know, what do you remember from them? You're probably like, nothing. Like, one one thing one joke somebody told me. But it's like if you could, at the end of the day, review your conversations and be like, what were some of the highlights of the conversations I had today? I wonder if that reinforcement would, like, do wonders for, like, how well people, remember the conversations they have, the fruitfulness of their conversations, and then also, like, I think it there's a way it could enhance relationships in this way and that, hey.

Hunter Kallay:

I remembered what we talked about the other day, and now I'm you know, I've thought about it more, and this is what I think or something like that. So I I think it could really help us retain information a lot better, and really tap into our lives in ways that we haven't before.

Max Jacob:

For sure. That's that's definitely a very good way of explaining it. Internally, with the team, we refer to this as the as a beach analogy. Basically, your memories are sand grains on a on the beach. And every day, it's like grabbing a handful, and your the sand just falls through your hand.

Max Jacob:

Right? And just is mixed in with all the millions of other sand grains that you have. This will really allow you to kind of pick out that those current those really important sand grains and always be able to recall them.

Ryan Lazuka:

Yeah. For sure. You know, our memory can only go so far. So, you know, it's like having a boost to your memory for you pretty much. Is there anything, like, that you thought of in terms of, you know could you have it send you a daily email to say, here's the recap of your day or something like that?

Ryan Lazuka:

I mean, I know that's kinda more generic, but is that something like that possible?

Max Jacob:

So that's definitely a couple of features like that, we're definitely thinking of for the portal, in terms of just giving you notifications and updating you during the day, basically, on what's going on. And email is an interesting way of doing it, but I'd say notifications are gonna be better, just to your phone.

Ryan Lazuka:

Cool. Yeah. Because it's gonna it feels like it's gonna know you so well. I could send you a notification. It will say, like, hey.

Ryan Lazuka:

Like, it's not gonna say this to you, but it might be thinking this. The AI might be thinking this. It might say, hey. I noticed you usually forget stuff like this, so here's a little reminder or something like that, you know. It it'll give you a reminder for that.

Ryan Lazuka:

It'll pop up on your on your phone. The, use cases are endless. What are are there some specific use cases other than saying, like, hey, summarize can like, specific use cases that you think this will be really useful for?

Max Jacob:

Well, a lot of times, you want, like, a friend, that you can discuss a certain idea or topic about, and then you can correlate that to something else. A lot of times, you need something to recall a conversation. Let's say I just had a team call, and afterwards, I want a quick keynote of that basic like, of what I'm what I should follow-up with them or what I should you know, anything to keep track of memory, basically.

Ryan Lazuka:

Got it. So how does it work in terms of, like, say if you're having a conversation, you know, with a meeting and you need the keynote you need the most important parts highlighted of that meeting. It it sounds like it doesn't send everything to the cloud. It only picks out the parts that it thinks you want. But what if you ask it a question about something that it didn't record?

Ryan Lazuka:

Or does it record everything locally and then delete it after a while? Like, how does that work?

Max Jacob:

Right. So locally, it it does have a log locally, basically, of, that's actually so the system that we have built out for it is called LCR, large context recall. So there's 3 memory, layers. The first memory layer is you're human, so that's basically you, and its persona for itself that it, changes over time based off of you talking to it. That's the core layer.

Max Jacob:

Then it has the recall layer, which is what you just, started talking about, which is almost like a log of everything that you've talked about. But that's stored locally, and it it could search that via time stamps. So it could search that based of, date of and and time. An archival layer that is basically a vector database, a formatted vector database for it to organize its memory. That's that's the insight engine.

Max Jacob:

That's what we call it.

Ryan Lazuka:

Awesome. So it's like the best of both worlds. Your privacy is on the device, and it it might remember some key facts, but it that that that it puts in the cloud in, like, a vector database maybe. But for the most part, it's not gonna take all your conversations and put them in the cloud.

Max Jacob:

Exactly. The the actual conversations that it's fully recording, that's locally stored, and only the key details that it finds interesting are stored in the cloud.

Ryan Lazuka:

And now that we're talking on technology, can you go into a little details of how it works in the background? I mean, share as much as you want. You don't wanna give out give away your secret sauce to the project, but, do you use, like, OpenAI's API or use some other large language model to do to do the AI magic?

Max Jacob:

Well, that doesn't give away anything at all, honestly. So

Ryan Lazuka:

Yes.

Max Jacob:

The way that the LCR is built out is it's a framework that you could plug in any LLM. So you could plug in Gemini. You could plug in GPT. You could plug in Claude, into it and or you could even run a llama in in your on your NAS or something and plug it into it. And it'll it'll work actually at that point fully locally for you.

Ryan Lazuka:

So you would just is the interface gonna work like you open up the app and you type in your API key for whatever large language model you wanna use, and then it would sort of work from there?

Max Jacob:

Exactly. We we think that we're going to be basically facilitating the memory storage and organization of it. So, basically, running the actual operator, the LCR of the operator, and you plug in the LLM that you would wanna use.

Ryan Lazuka:

Awesome. Very cool. Is there, anything else you wanna tell us about the project before we start wrapping things up here, Max?

Max Jacob:

Other than you guys should go follow it on Product Hunt, Search up TAP. You'll find it. Yep. That's it.

Hunter Kallay:

Yeah. So it's you can find it on Product Hunt if you search TAP. Also, you can just type in your browser, hey tap dot tech. You can check it all out, how it works. You can see the model, everything like that.

Hunter Kallay:

And then, also, you can check out our Ryan and I's weekday newsletter, fry hyphen AI dot com, where we keep up on the latest in AI news. And then be sure to check out fryAI on YouTube, and then also be sure to subscribe to behind the bots so you can see cool interviews like this one.