Sentra and Organizational Memory | Ashwin Gopinath Interview

transcript

Full verbatim transcript of the attached Lands End interview with Ashwin Gopinath, from the official X video English auto-generated captions for posterity.

Hey, I would have left MIT even if I didn't it just got boring. You started as a physicist How do you end up building an organizational memory company? So I'm a trained electrical engineer applied physicist. My PhD was on trying to understand how molecules interact with light I ended up in Google X trying to solve like systems biology problems Then ended up at MIT The way in which people work together is what actually is the most exciting part for me but also it's the hardest part That's sort of like morphed into what became Centra. For people who've never heard of Centra what is it? At Centra we're building what can be thought of as like organizational memory. It could be like company brain How can we get a team to act as a single unit?

How do you make sure that if one part learns about something ephemerally everybody else learns about it? You don't relearn experiences. You can move fast What is For Centra what problem does it solve? So it makes it easy for us to connect to other systems. We team has a different set of tools. We are live in a fairly large organization with like about two and a half thousand employees What used to take like probably a week of them figuring out when something changed it collapses to 10 minutes I want to ask you one more question though. What's like the highest leverage challenge you're facing right now? Agents our team has got like really great people who will just go do stuff without necessarily asking me for permission Whatever needs to happen to get it done Agents from everything else

Ashwin, hi. Hi

I'm here with, you're the CEO and co-founder of Centra and where are we right now? So, this is near Land's End in, San Francisco I like this place because you know, whenever I have, like, folks visiting, you know, I bring them to you know, places where you can see Golden Gate and, like, this has, like a really gorgeous vantage point You can see, see the bridge, without too much crowd So, so I like to kind of find places where you have like, unique, perspective on things Yeah because when I asked you what's a special place in SF this, this place means a lot to you then yeah? Yeah, yeah, and, and, and and this is a place that I bring my daughter to quite often so, so, yeah That's amazing I'm really excited to be talking with you here. It's very beautiful Let's open it up with for people who've never heard of Centra what is it?

So, at Centra, we're building, what can be thought of as like, organizational memory another way to think about it is it could be like, company brain So, at, at a very higher very High level, what we are asking is how can we get a team to act as a single unit? How do you make sure that if one part learns about something ephemerally, everybody else learns about it? So you don't relearn experiences You can move faster. You can move as a single unit. Because one of this is the problem that I have always faced whether it is an academia or any team the team is only as good as how they can operate. So that's what we are trying to do Wow And what's the thing most people get wrong about what you're building?

So, so people think that when we when we say that we're building organization memory or memory in sense they mistake it for like a glorified search box

That's not true. I mean memory is a lot more than just searching and finding the right thing or like being able to kind of pull things out Memory is much closer to a model and, you know, and, and we are we are trying to solve that for a whole organization. And there's like technical reason why it's easier to solve it for organizations than for individuals okay And before going on this podcast I had a quick call with you just to kind of get to know you a little bit And you shared so many great things which I really want to dive deeper into But before we get into that can you just walk me through your path?

You started as a physicist How do you end up building an organizational memory company? Yeah, so, so I'm a trained electrical engineer, applied physicist did, like, my PhD the whole work was on trying to understand how molecules interact with light That, that, that took me down for a long time and I was, like, a researcher and, like, a junior faculty, like, research faculty at Caltech trying to solve this question and then, you know, a strange set of events happened and I ended up in Google Google X, trying to solve, like, systems biology problems trying to understand how biology can be viewed as a generative system did a bunch of companies around that then ended up at MIT where I started, restarted my lab as a faculty member Did some of the earlier works on agentic systems simply because that was exciting for me at that time And, through it all, you know one point that kept coming up was the way in which people work together is what actually is the most you know, exciting part for me but also it's the hardest part How do you get everybody to kind of be pointing to the North Star working on it, you know have minimum misunderstanding and all of that?

That sort of like morphed into what became central And what made agentic systems so exciting for you at the time? At that time, I mean, I I was interested in it because always it was very clear that that, that AI is going to be Something interesting like, like, right from the time that I was an undergrad but it never it never felt like it was there I mean, and, and about four or five years ago when I had sold my previous company and I was in a transitionary phase you know, I started playing with this and it became very clear that everything is going to change And, but it didn't feel like the the, the, just the LLMs were good enough You had to wrap it around something else and, and that sort of like, led me down the path to you know, doing the work on reflection That led me down the path to realizing that none of these agentic systems are going to be anything without memory and that, that linked to, you know, the other aspect of what I like working on is like, building teams and, you know getting people to work together So it's all, like, m-moffed together Where we chatted before you said diverse experiences are like diverse training data for a model They make you more capable?

Was there a moment where that played out? Yeah, I mean I wouldn't say there is one moment that it played out It plays out in like every day in the sense that things that you learn in biology the way in which you view the world the view like the noisiness in biology and like how you treat biological systems and like tease out the noise from it You know there are a lot of things that you can learn from that that you can pull and apply it to LLMs and like these systems now because they are much more empirically driven now You don't have, it's not science I, I I wouldn't say that LLMs are a science yet It's much more closer to biology than a science but it gives you a certain discipline in how to actually look at these systems But because you also or like, because I have had training in physics there are these questions of trying to grapple it and trying to understand what is the root cause how do you get like, what why does it behave the way in which it does?

So that, that, that that friction allows you to do a lot more What made you leave MIT to build Centra? I mean, I would have left MIT even even if I didn't build Centra It just, it just got boring I mean, it's like boring in the sense that in the sense that I I don't I mean, MIT is great My department was great It was a wonderful time great students, but there is just so much going there's so much fun outside And I see my friends and colleagues basically having fun outside And it's like, I want to play there Let me go there and play You told me once and I want you to restate this for me memory is not a knowledge base What's the difference?

Yeah, so that's, that's, that's a great point So memory is knowledge base and that's a mistake a lot of people make Whenever you hear the term a memory company they think that it is a glorified knowledge base and you are trying to search for something among a whole bunch of things that are out there The best way to understand the difference between knowledge base and memory is if you have perspective Like if you have three people looking at the same piece of data and I think I told you the and like part of the reason why we ended up here is because you know the example that I like to give everyone is if you have three people looking at the Golden Gate Bridge which you have like over there you have like three different perspectives Let's take an example right? Like let's say a designer and an engineer and a biologist a marine biologist were to look at the Golden Gate from the exact same point They're going to walk away with three different memories One would remember the color the designer might remember the color the engineer might remember the Cable and the beauty of the engineering in there And the biologists might completely forget about the bridge and might remember some sea lion or the ecosystem around here So it's the same knowledge but they walked away with different memories So there's something there. And that's very profound right? Like memory is part of the knowledge that is going to be useful for whoever is having that memory or the system that has memory. So the designer because they have a utility function they care about the colors and the beauty of design they remember that The engineer because they have are trained on like thinking in an analytical fashion they would remember that part because that'd be useful that they think it'll be useful in the future. That's a big difference And there it is more it's much more closer to a model than a database or a knowledge base And you said, if you remember everything you essentially remember nothing. Let's unpack that Yeah, it's, it's, it's sort of like this right? You, you don't remember everything Like, no one remembers everything I mean, I, I think, I think you, whether you know it or not you implicitly make certain

you know, you, you, your attention goes towards certain, part of, piece of data, whether you do it consciously or unconsciously

okay? And that is because everything that you have seen till that point sort of attracts your attention over there You see that in the model systems as well If you give it a bunch of data and you look they're, they're not, they're not, the attention is not uniform okay? And there is a relationship that are there That, that, that, that's what I mean You, you need to have like a priority Like, what is important for you? Yeah, and about that so you know how like when you're watching a movie or something you sometimes if you watch it the second time you notice like things that are completely different different And that's that's really crazy how like maybe at that moment in time when you're rewatching it you're thinking about something else or maybe you have a different perspective and you see things differently. Precisely And, and, and, and, and that's that's part of the reason why memory can't be solved for individuals Because like for an individual you don't know the motivations of that individual You don't know why they are doing what they are doing They could be doing certain things just because they had like a great cup of coffee that day morning Okay But but none of these AI systems will be able to capture that So you don't know the utility function like why of that system most of the time if it's for an individual And that's partly the reason why we are doing it for organizations because companies and teams you can define why they are doing what they are doing So like most companies or all companies at least in the capitalistic systems would be, you are you care about maximizing profit You have a product that you want to sell and you have people who are building those products. So anything that is not touching those things in some sense, the company doesn't care And, and, and you don't the company doesn't need to remember So in a meeting if nothing gets decided you don't need to remember that like the details of the meeting That makes a lot of sense Okay. And that's why organizational memory is so much easier than individual At least that's the thesis that we are going with. Okay Every other memory company decides upfront how to look at the data The schema determines everything. What's wrong with that?

So again, it's the same problem right? It's like you have already deciding what is the perspective that you want to take on the same piece of data And so a good way to I mean, just like the Golden Gate example the example that I would go with is let's say you're getting an email or a meeting wherein some customer is complaining to the sales rep I'm not happy with this feature I'm going to churn It's costly, whatever it might be Now different people in the same company would look at that piece of data and come up with different ideas The salesperson might look at it and be like oh, what could I have done different?

Could I have offered them a better deal? Could I have reduced the price? All of that The product manager would look at that or the engineer would look at that and say what is the feature that caused that complaint to start to begin with? Now, the CEO might look at it and say oh, this person, no, fine, even if I lose it it doesn't change my top line revenue is not going to change Nothing is going to matter My board is not going to affect So he'll forget about it So different people and different parts of the organization would remember these things differently That's what needs to be captured And if you decide how Look at the data So that's why memory companies end up creating CRMs CRMs are for customer relationship management That's the salesperson's perspective Or they try to create it from a product manager's point of view But once you decide that you can't reuse the data You set the perspective Why can't you reuse it?

Because you forget about the like it's sort of like the like going back to the Golden Gate analogy right? If the design the designer forgets about the the, the, the marine ecosystem or like the you know, the, the fish in the water or like the like the dolphin in the water So there's no way like if I ask them the question about that that they can answer it So once you create the structure you forget a whole lot of things and you can't reuse it That makes sense. Wow Okay let's get into some of the big takes that you have So what do you think is the most underrated tension in what's happening in SF right now?

Something that's not, someone's not saying out loud? I mean, nobody's held responsible when AI is going wrong does something wrong Like in the sense that in the sense that like AI makes mistakes all the time but if a individual makes a mistake if my if somebody who's working for my team makes a mistake at least I can actually hold him accountable Like I can't hold OpenAI or like Anthropic accountable when something goes wrong in a coding tool I mean, the only thing option is I can just basically you know, like churn and go to somebody else But ultimately, who's responsible when things go wrong?

Yeah And is this the AI's peak socialism and peak capitalism? No, no, that is just, no, no I, I mean, it's, I I find it very funny that like the biggest companies in the world right now are using the collective intelligence or the collective knowledge that has been like, that, that, that has been taken by the entire like, human history and trained their tools So in some sense that is owned by everyone Like implicitly, the, like every human till now sort of participated in that in some sense And that Is used to create a tool that is a representation of peak capitalism in some sense. I mean their valuation and everything. There's there's something very funny going on there I mean, I'm not saying whether it is what what, what, what somebody should do but there's something very funny about that Oh, okay Everyone talks about AI taking over AGI, super intelligence Do you think that's a real risk?

No I think, I think, I think more the more things change the more they remain the same I think, I think I think it's not going to happen I think, I think that it's not going to be Terminator around the corner basically taking over or you know, or, you know utopia is going to be somewhere in the middle Ultimately, it comes down to this right? It's like, it's the liability issue It's like who's going to be responsible when something goes wrong? Like, and it can't be insurance Somebody, like whenever something goes wrong at least in most companies or most teams somebody gets held responsible Some person needs to be held responsible for it Until that gets figured out you know, this is not going to change And, and like, like the big company the, the AI, the big LM or the big AI companies are not going to be the ones who will be responsible when things go wrong So somebody is going to drive the AI and they are going to basically Maybe pogs in the machine and so on Okay, yeah, because I heard this take where it's like you know, people are worried about AI becoming more human but what about, yeah, humans becoming like very much I mean, we, I mean, we use AI for everything right? So we're like Yeah yeah, that's, I'm, I'm, I'm way more I'm way more afraid of humans becoming more like AI than AI becoming more like humans. Because I see this right? Like, and and this is part of the reason why I decided to sort of leave academia Like I was in the in the, in the like around the time ChatGPT came out was when I started noticing weirdness in academia in the sense that suddenly the quality of essays that you end up getting in admissions packets, went up. I mean that can't be true. I mean I've seen it over many years and it remains approximately the same I mean, almost the same You see like good ones bad ones, but then around 23 onwards 23 and 24 like that, that, that, that, that pool of applicants there was a market change Like everything became better. It's all almost like it was like Filtered through ChatGPT And, and, and, and the more you use it you see this, right? You, you, you see, you have seen you have instances of like cognitive atrophy Like, an example is when GPS was introduced to the

taxi drivers in London their ability to navigate without GPS went down dramatically So now the question is what happens when you are constantly using ChatGPT or Claude for every single one of your actions? You stop being able to think You think more like that or, or you become you don't have the ability to think anymore And that, that terrifies me Yeah, you become less confident in your decisions What's does someone who's gone fully AI native look like in 10 years? What is this going to how's that? Is it going to be worse?

It's going to be interesting right? It's going to be like the way in which I think about it is I don't think it's going to be chat It's not going to be not going to be chat interfaces The way in which we think about it is so Centra has a interface which is more like there are two ways to interact with Centra One way to interact with Centra is like an API layer wherein it connects to all your systems sort of understands it and connected to your agents or your apps so it can basically deliver what needs to be delivered The other interface is an app interface And the way in which we think about it is Since Centra knows what everybody should be working on everybody's commitments, what they have, who they have committed to what are the priorities of the company all of that, it can basically you can think of it sort of like TikTokification of work in some sense wherein like it tells you you need to be working on this let me help you do this Here is a way in which I think about it brings it to you wherein it doesn't, like wherein you're not going to AI AI comes to you and be like hey, you should be doing this Okay, let me help you do this Here is a way in which I think you should do. Let's talk about it and let's let's go, let me go and finish it Okay, and then you walk away I, the way in which I think about it is like I want humans to or like, if, if we are doing everything right I want people to just basically do things like this like be wherever, like talk to each other like hang out and have like human experiences And all the grunt work ephemerally AI is present and it just basically comes to you and say hey, let me do this let me do that and then just basically does it Yeah, yeah, that's the, that's the that's a good version Most people would say the AI race is basically over. OpenAI is the winner Anthropic is second. Everyone else is fighting for scraps. Is that right?

No? I

think the interface is like I said, the interface itself is very very, it is more intuitive than anything else but it is still not anywhere close to what I would consider as human. The way when you stand back and you look at these systems you are not interacting with it the way in which you and I are interacting with it. You don't have sessions You don't have like a tab And I'm talking to you I'm not like, I'm not picking up you know, from scratch It's not a new like you have your experiences you have learned. I have done a whole bunch And then we talk to each other and like three months later when we are talking to each other we'll pick it up from you know, where we left off But also there are a lot of other experiences that you bring to the table I bring to the table AI doesn't feel like that at all It's nowhere close to it So, and, and and the amount of energy that is being wasted in these systems it's like far from optimized

there's, there's, there's so much more I'm not saying that these guys are going to disappear but I don't think it's I, I, I, I I think there's like so much more to be done Wow okay. So there's going to be someone else that's going to going to be a part of the yeah. I mean, when, when, when the, through the whole dot-com early time, when, I mean one could have made the case that Netscape would have taken it all okay? It's not, I mean, things change Yeah If the winner is determined by memory architecture is Centra building it or just enabling it?

I want to say we are building it I think I think partly because of the fact that I don't think like, like, like I said they're exactly what is memory

I have a certain perspective on it that that that that that that is you know, that that's partly understandable

but there are like other aspects of memory that even I don't know okay? I'll be like I don't know how to bring in like continuous learning in the sense that when you remember something how do you make sure that you learn from it? And how do you how do you do it without having to kind of retrain the whole system? Like the architecture needs to be different It's not going to be a rag It's not going to be some kind of graph thing It's going to be something else altogether that is going to look more and more like a a model more and more something that constantly change and evolves How do you actually put utility function into it?

Not sure Okay. I have like some ideas and all of these other things So it's it's far from solved but it is going to be a very very big part of it Like if I were a betting person I would basically say that in the next two to three years more and more just like if you if you roll back Like there were rumblings of agents like two years back but it was still not a common word I think two, three years from now you're gonna have like more and more talk about memory. Like what does it mean? Like, like how do you deliver memory?

Is it Sentra? Is it something else? I'm sure all of these guys are gonna come up with like their own interpretation of memory. Which one will win out? How it interfaces with these systems? What it enables? All of these questions are gonna come up. Yeah. Yeah memory is like very still very, like very new term Yeah. Yeah, I think we, I think, I think memory if you were to compare memory to agents we are, you know, probably like three, like barely, barely at the beginning of where the term agent starts get like, like, like thrown around That's exciting. Wow, you're really, yeah building at the beginning of all of it Everyone is racing toward full automation Why are you building in the opposite direction?

I wouldn't say that we are building for full automation. I think I think, I think this is this is this is core to why we are doing what we are doing. I think I want to work on things that are like, I want to use be, I want to take AI where

human interactions and human relationships are there I want to take I want to make an AI enable fundamentally human interactions and fundamentally human you know, dynamics So teams and groups of entities working together is about as old as you know anything else We were chimpanzees in a group of chimpanzees and we came down and became group of people and now you have group of people running companies It's been around okay? And there are tensions that happen around them I, I, I I want to basically work on that area I don't want to I, I somehow it doesn't philosophically align with me to kind of remove humans I want to, I, I like humans Oh, I love that. Okay so those are, yeah, so you're like the values you have like you want to. Yeah I, I, I want to make it make it easier for people to kind of work together because people are not just going to be you know, sitting in a in a beach drinking margaritas That's not going to happen And do you think that genuinely it's better like it's a better product?

Or is that like a philosophical position or just you know, this could be? No, because like I said right? Like we are far from

any AI system basically autonomously doing anything Because we as much as we want to say that we can control it and give it certain value systems and it doesn't break it's not It, it breaks all the time It gets stuck all the time Unless you have somebody defining the you know, guardrails and saying what decisions needs to be made these systems are not as foolproof as they are made out to be. nice. Use Composio. Yeah

What does Composio do for Centra? What problem does it solve? So Composio is it makes it easy for us to connect other systems into

into Centra and, you know every company or every team has a different set of tools and different set of

see different way of working with each other with these tools. So the the, the, the, the range of integrations and range of tools that Composio is compatible with that makes it easy for us to serve our customers. Nice Composio connects agents to tools like you mentioned Centra makes sure the right memory responds to whoever is using those tools For a founder building an AI native product today what does the full infrastructure stack need to look like?

I don't have an answer to that question All I can say is like Composio and things like Composio has a role to play And in order to kind of integrate the data that is coming in so that you the agents don't need to kind of rebuild the context every single time to know where to look at to see how it is connected to everything else something like Centra needs to be there I mean, I'm not saying that Centra is the only one but that, that that category of organizational memory or like AI memory that is still forming So it's sort of like what does, like AWS and S3 look like?

Like, like storage, like what storage did to the

cloud infrastructure like what the kind of memory that we are building we hope is going to do to all agentic and AI infrastructure And Composio plays a part in the sense that Composio has the pipes that basically brings the data into it You said there's no formula to building a startup Most people giving founder advice are just doing post hoc rationalization They only remember what worked Do you believe your own advice? Yes, it's it's the core of the what memory is right? In the sense that you forget a whole bunch of things okay? So why, like in the sense that if something worked out you're going to basically come up with rationalization as to what you what you, this is the, at the core of memory Like you don't remember the true facts of what happened you remember what worked or what is important and like, and, and memory change, like, your your interpretation of it changes So, when somebody says that you know, there is a formula to success there isn't There isn't It's sort of like you know, if it worked, it worked, okay?

There are few, there are obviously bad bad decisions, but then, you know sometimes those work as well So, that is not to say that there you shouldn't have, like, best practices and stuff like that but you know Like I I I mean this is not to kind of like throw shades at anyone but I see like 20 year olds who have zero life experience give founder advice

And I I wonder it's sort of like really? Like like wh wh like and and and then there are like people who are like 70 year old who are giving founder advice Again, the their lessons there are like things that they can learn from the lessons but there are a lot of things that they just forget So Right? Like I wish like I wish I remembered a lot of the things I did. Like there's you know and there's probably so much like good advice that like people could share. But also I heard like so advice is pretty much just people sharing about their lived experiences right? So it's like So I I wish everyone says this worked for me Okay. And in my in my particular worldview because of everything that happened this sort of worked out It doesn't necessarily mean that if you do the same thing you will have the same outcome Maybe, may not be. But the point is this is, this is my story Yeah, right. So before you like say tell your advice like so maybe instead of like saying like my advice is like say kind of like yeah, like this is this is what worked for me Like always having those like disclaimers because yeah it's, it's, there's no formula like you said You can't, it's not like math Yeah

Why should anyone listen to like The advice then like how useful is advice just? No it's just take it with a grain of salt It's not, it's not, it's not intended to be I would say, I would say this right? It's sort of it's like one perspective Like you can't, it, again it goes back to LLM training or any of these things Diversity of training, data If you have multiple points of view if you if you train a model or if you train something only on a very tight set of data like XP, you know, information it memorizes it and it doesn't generalize Okay And so the reason why you want to have like diversity of data training data is so that the model sort of learns the meta lessons underneath it and generalizes it So it can do like zero shot learning in a new experience a new condition The same idea. If you listen to the same advice okay, you're like a hundred times you're not going to be able to generalize So everything is like take it with a grain of salt Yeah What's a decision you made building Centra that you can't fully explain in retrospect?

You don't know exactly why you made it but it was right

I would say I, I, I would say choosing the team to work with

I think

part of the reason why I'm saying that is so it's basically me and my, my, my, my, my students basically ended up starting it And there's a huge age gap difference between us and perspective, like we have generational gap difference between us and, and my my whole team is like a bunch of like young 20-year-olds like, and I'm like early 40s okay? There's a language barrier there is like a whole bunch of other things I don't know exactly why I did it I could have gone the other route of basically having like more experienced folks in the team but it's more fun It's more fun to basically build it with, like in, in some sense, I, I, I I guess one way to think about it is my previous startups are like my XP other like I always had a leg in academia And there's a certain strange set of energy that you end up getting with like people who want to like go ahead and do stuff and, probably that's why I can't, I can't, I can't explain it I can't, I, that's what it is And in some sense I can't explain it because it could have been done in a whole different way Yeah So, would you agree so you said you described yourself as a dog the dog from Up?

Yeah, you just chase whatever interests you Is that a strategy? It's just oh no, it's not strategy Like I said, it is, it is So in my, in my if I were to kind of look at my career arc I'm trained electrical engineer. Then I did my PhD in a whole bunch of like physics stuff I shifted to like trying to put molecules using semiconductor stuff Then I ended up at doing going to Google to go and work on biology of all things And then went back to academia It seems very chaotic. It worked for me The reason why I did it is because I got to a point wherein that experience gave me everything that I could get from it. I felt like I I had learned what I needed to learn from it Moved on to whatever was the next thing And that is, there was zero strategy in all it And, and so just because it worked for me it doesn't mean it'll work for everybody else I did it because I wanted it, it could have gone a terrible Another thing too, you mentioned, a little bit about you know you seeing 20-year-olds giving startup advice and they'll have a lot of experience. Following onto that a lot of younger people are dropping out of school to just you know, build startups. A lot of people are you know getting into YC and they're kind of taking a break from school What are your thoughts on that? Because I feel like you're you know it's a very formative part of your life I feel like, I think it's I would say I think it's really important that you finish school If you're very smart obviously, like, you know, drop out, but So I, I would, let, let me put it this way right? Like this is the advice that I would give my daughter okay?

Like once, like, and she's, she's still like, she's still very, very young, okay? like do stuff, build things, okay?

Yeah school is great for like the social aspect of things. You end up learning a lot a lot about like how to interact with humans

which you wouldn't learn by just basically building things But I think formal education

I think is where traditional media was when YouTube was starting out I think it's like one or two platforms away from basically disappearing in the way it is I don't know what they are but I, I mean that's another reason why I decided to sort of like like, like be, be like, okay, I, I, I, I mean like higher education is great but I don't want to I don't want to it felt like it was a dying a dying institution It's not dead. I mean it's like traditional media has not died You still have news and you still have CNN of the world but it doesn't have anywhere close to the importance as it used to in a pre YouTube or a pre TikTok pre Instagram world, like pre social media world So I think in the 10 years time Academia would still exist but I think these newer tools are going to be like radically changing what it looks like and feels like Yeah, because like everything is evolving but like academia is not I wouldn't say it's not I think it's, I wouldn't say it is not It is, it is, but it's just slow Slower, okay. It's just slow Yeah yeah If you get, get a right with Centra what changes?

What does an organization with real memory look like versus one without? And do you think every organization should have memory? Yes, I think every organization should have because there is it's not as if organizations don't have memory right now You, people call it as the culture of the company Like, like Google and, you know, Amazon and all all that have like certain ways in which they operate Like, like Google has the whole I mean, the whole Googliness idea okay? And then you have like Amazon has the whole memo culture okay?

And those are all parts of collectively a group of people coming together and saying we're going to work in this way okay? But what Centra loves you to do is like turn that notch to 10 and to basically say you know how do you get a group of people to sort of Walk towards the North Star with minimal friction between each other okay? So that, so so that you act as a single entity what I call as like collective intelligence okay? And you see things like this right? Like I, I, I I love to kind of compare what Centra is doing to social media in the sense that social media somehow gets groups of people to think like the whole concept of groupthink wherein, they all have like similar ideas and they they sort of feed onto each other and like think in the same way Now, the question is what does that look like in a purposeful manner inside an organization?

Like if you have a group of 10 people who are all who have like minimum friction between each other who are all constantly like learning from each other they would be able to do a lot more They would be able to outcompete a team of like a thousand maybe Now what does that look like when every team has that? Yeah! That's what Centra is trying to sort of like bring about. That's amazing You said the AI era is still a transitional phase The new norm was coming but we're not there yet Yeah Yeah? Yeah, I, I I don't know what the new norm is going to be but I, all I know is that it is going to like, see, here, here's the thing, right?

This is what I meant by saying like, 20-year-olds have not actually seen transitions

And I'm not saying I've seen a lot of transitions but I've seen things like when like, where, first, like social media was supposed to be the greatest thing and then it changed to oh my god, social media is the worst Yes! Okay, same thing as, like internet is going to change everything Internet changed some things but many things remained the same People just, see, it's going to be the same thing and this has happened multiple times If you go back in history it's, like, happened again and again and again So, I don't know what it's going to be looking like but it's not going to be like we're, we're not going to go into a you know It's not going to be a dystopian future It's not going to be a utopian future It's going to be somewhere somewhere in between. I mean people are still going to like like hang out with each other People are still going to have like human problems. It's not as if it's all going to disappear Yeah, it's not like the WALL-E movie Yeah, exactly, WALL-E. Yeah, that's just patently absurd Yeah. Oh yeah, and about the social media Yeah, it's crazy. I see a lot of people more people like trying to disconnect as much as possible detoxing and I'm quitting social media And if you go back and if you go back and read articles from like 15 years back social media was like this is the new norm Now everything is going to change Like people are going to like this is going to change everything What happened to metalers?

Yeah, it was supposed to be like yeah, it was supposed to be the greatest thing And, and I so I'm not saying that it is going to go the same way like AI is going and and we're not saying that these tools are not going to actually you know, resurrect and it's going to it's going to evolve but more things change more they remain the same Yeah Yeah, that's a really good phrase of yours Yeah So what will the winners have though that that the losers won't? Perspective I want to say perspective I mean, it, it goes, see, this is the thing that that, that, that, that that makes me so happy about the work that I'm doing because memory is actually you can sit down and really think about memory a whole lot. I mean like

it, like the different perspective of on the problem is the most important thing Like, like to somebody who knows the value of a golden nugget okay? They will see it not as a rock but if somebody didn't know about gold they would look at it as a rock So now perspective is based upon memory and what you have learned right? And, and, and that is, that is your experiences or or you could have you could, and there there are like other small things that you can say. So whichever team has perspective and interesting perspective that is valuable

that's going to win. I I have no idea what that is but, you know, we have a certain thesis on what what it should be and we are like chasing that and we'll see what happens Interesting perspective is very would you say, very subjective It's just like the way you live life and the people you're surrounded by the, yeah. Yeah, yeah, definitely, definitely This is, this is, it goes again and again it goes back right? It goes back to what is the utility function of an individual That determines as to what your perspective is The perspective determines as to what you're looking at and what you you, you memorize in some sense And that like it's very hard to define what that utility function is. If I knew it I wouldn't, like, I wouldn't share it with everyone It's like, I would want to use it Yeah What's the idea you keep coming back to that you haven't had the right conversation to fully work out yet?

The thing that keeps coming like, the thing that, that that keeps coming back to me is I don't know what the future is going to look like okay? I know, like, in the, in the sense that that, like, all these tools have like, Really wonderful, positive sides

and in the long run that will sort of like pan itself out and it'll it'll sort of like come to the fruition But in the, in the I would say in the next 10 15 years, as people sort of work things these things out I don't know what that's going to look like And, and, and, and again like my perspective on some of these things are because I have a like I have a little child in my life So it's sort of like I don't know what her like, education is going to look like how AI is going to look like for her Like, I like like something that makes me very uncomfortable is she talks to my grand, my, my, my parents over the phone okay, very regularly And the other day and she like has babbling conversations with my my, my mom and dad The other day, I was talking to ChatGPT okay, it was on speaker and she started reacting in the same way okay? And that freaked me out because it's sort of like how do you explain to Children who have no idea what AI is

but they have interacted with people on the phone that this like how do you introduce AI to somebody who has experienced AI but never don't know this difference between an AI and a human?

That, that, that, that, that, that that freaks me out. I don't know I'm very, that's a very uncomfortable thought Because like we lived through you know, we lived when AI wasn't a thing and but like all the the future generation, they're they're not going to know what it was like without. So how yeah, it's going to be How do you explain that this thing ChatGPT doesn't exist? Right And like kids are so naive I'm like, you know, and they're, it's so yeah. That freaks me out Yeah, definitely Me too What can you tell me about your product which you haven't talked about publicly yet?

Pretty much, like internally, at least I don't use any Traditional

I would say, systems of record anymore Like in the sense that at least in Centra we don't have any other tools other than the communication tools and Centra Because Centra basically can do everything else that you're asking for. Like in the sense that like since it's connected to all the meetings and all the messages and all the emails of everyone I can basically get up in the morning and say you know, who am I talking to? Who else have talked to these customers? What is happening? And it just gives me a report Like I don't need to have go into a CRM to kind of like know what is a pipeline. All of that is done Like I can ask it like in our conversation like what should I know about this thing?

And like what has happened? Like that has changed and it just basically tells me what I need to know It changes the way in which you operate We don't have anything other than the communication Like internally we have only the communication stuff and Centra don't have anything else because everything else can be taken care by Centra itself And are you launching anything new or? So the only thing that we are like, we have and so one of the things that we have been very very cautious about is

center sort of like when you have the whole whole organization work as a single entity it introduces weird problems. Not not weird problems not from an access control and all of that. People think that you know, if everyone sees, like has access to all my memories it causes problems. That's not the kind of thing The thing is what happens when the time compresses? Between the time when when sales complaint happens to the CTO

Knowing about it and figuring out you know, that, that that something needs to be done Like we have a we, we are, we are live in a in a, in a, in a in a fairly large organization with like about two to about two and a half thousand employees wherein we look at KPIs that they are getting from their retail stores And not only are we any change in the KPIs we automatically go back and say you know this particular person and this conversation and this particular thing is probably the reason why this changed What used to take like probably a week of them figuring out when something changed collapses to 10 minutes. Now that that's weird because it changes the way in which you operate Now I don't, and like, so, so, so so we've been very careful to kind of work with design partners and like slowly get it like push it out rather than say that oh, we have an enterprise product like everybody should use it Because we also notice that you know 90% of these enterprise pilots don't go anywhere So we want to like that we want to really understand what went wrong what goes wrong when you actually put these things in there and work with them to kind of Get it right Why don't the 90% of the pilots don't go anywhere?

Because it's hard to kind of quantify how AI is useful yeah. Like how does it how, like, can like the problem is whoever controls the budget needs to know did it actually give us a value? Did I, like, I'm, I'm using, hold this AI, everybody's using AI but did my profits go up by you know, by, by, quantitatively by the same amount? And that that doesn't turn out to be the case And whoever is actually writing the check will be like I don't want to I don't see the value Oh, yeah and this is where you're talking so much It's like it's easier sometimes to hire a person because you kind of know how much I mean, you know you know how much you're going to pay them and everything. Exactly. So so the thing is if I'm hiring somebody I know how much I'm going to spend on that person. And from there from my experience I can kind of guess what that value is going to be Like, as opposed to an AI that is constantly being metered if I give it a task I don't know how many tokens they are going to use I don't know what are the downstream problems that are going to be Eventually, these things will get figured out But right now, those are not figured out So when you're trying to get this through an enterprise sales there is always one or two people in there who are like and rightly so, like tell me what is the value that this brings?

Am I spending a dollar on this? Am I getting $5 or $10 like either on my top line or you know, is it saving money? So my, you know, bottom line changes Like those are the questions that needs to be asked And unfortunately, that is the reality So if you say that you know, I give center, I mean I give AI or like any tool to everybody does that really change our revenue? Do we get more revenue? Do we have more profits? If you can't show that you can't make the sale I want to ask you one more question though What's like the The highest leverage challenger facing right now?

Agency Agency? What do you mean? Yeah, I want to, I want to have like people like, I think I think our team has got like really great people who will just go do stuff without necessarily asking me for permission or do just do things. Okay and like whatever tool needs to be done like whatever, like whatever you need to learn whatever whatever needs to happen to get it done I think that is like that, that, that's the most important piece of like not, not education, not your qualifications not all of that I mean, agency trumps everything else Okay

and can you, can you build agency? Like, or is it character? Because I feel like it's character It's like doing things Yeah Like not waiting for permission and just doing things Like some folks, I, I don't know there are like some people who just have it and the others don't I, I don't know what what, what makes the difference but But I think I mean, being in SF surrounding yourself with people that are better than you I mean, that's gonna definitely It helps. It helps Yeah. It helps. But those pockets exist everywhere You're from Waterloo. That's true That's true, yeah Thank you Ashlyn. This is such an incredible conversation Appreciate it