Joe Benarroch:
I'm really excited that you guys are here, but we were talking just a little bit. I wanted to get a little bit more of your origin stories because I think the origin is actually kind of good. So I'll start enjoying my oatmeal. But let's start with your origin story. Was there a moment in time where you said to yourself, "Okay, Vector Database, this idea isn't just something that needs to exist, it's something that it's going to change industries"?
Edo Liberty:
Yeah, 100%. That was, of course, when I left AWS to build Pinecone, at that time I was building AI infrastructure as a part of AWS, you saw these two massive trends in technology converge. So as a society, we spend decades and millions of human hours and billions of dollars in building machine learning. From GPUs, to models to... I'm an academic, I still am an adjunct professor-
Joe Benarroch:
Right. In Tel Aviv. Right?
Edo Liberty:
The amount of... Tel Aviv... I actually also taught at Princeton last year. The amount of effort that we put into getting these models that we have today and getting them to be smart is stupendous. You actually have exactly the analogous story with information and knowledge. Search, knowledge basis, the internet itself, open, like Wikipedia, organizing knowledge as a human endeavor is a massive investment. Being able to search stuff, being able to retrieve relevant information, all the advance in databases and search engines and Google itself and all that stuff. And late 2018, '19, you started seeing those worlds converge. Suddenly it's not search engines and models, it's search engines and models. AI is going to make search 1000 times smarter and search is going to make AI 1000 times more usable, and they become one thing. And that became this colossal pincer movement of technology, of take two of the biggest investments ever made by humanity in technology and have this supernova in the middle. And if we could be the thing that connects those two worlds, that's the place to be.
Joe Benarroch:
Because most data is unstructured. And I want to give you a little bit of time to eat, though, and I want to talk about why you think just most companies can actually structure it well or find any meaningful value in it. And I'm curious about that because I love when you say that there's this difference between knowledge and intelligence and on the face of it, those just seem like the exact two words, the exact two meanings-
Edo Liberty:
No completely-
Joe Benarroch:
... But they're not.
Edo Liberty:
... Completely different. I don't know about you, but I don't think I'm any smarter than I was when I was 18. Just in terms of-
Joe Benarroch:
[inaudible 00:03:46] and CPU.
Edo Liberty:
CPU. Right?
Joe Benarroch:
Yeah, exactly. Rock and CPU.
Edo Liberty:
I'm probably less.
Joe Benarroch:
I'm definitely less. Definitely less. Definitely less. Hundred percent.
Edo Liberty:
I would think I'm a much more capable CEO and executive today than I was at 18. I knew jack ****, the things that change is my knowledge and my experience, the fact that I now have-
Joe Benarroch:
I see what you're saying.
Edo Liberty:
I have these experiences. I actually remember actually drawing relevant information. The same thing is true in [inaudible 00:04:16]. Like you're a legal firm, the senior partner is not smarter than the whippersnapper on day one, right?
Joe Benarroch:
Yeah, the associate.
Stefan Weitz:
The whippersnapper.
Edo Liberty:
No, it's true.
Joe Benarroch:
I agree with you. I agree with you.
Edo Liberty:
The young kids that come in.
Stefan Weitz:
I used to be that guy.
Edo Liberty:
Smart as right? They don't have the wealth of information. When I speak with large companies like enterprises and so on, the most amount of volume in their business is not in their data lake.
Joe Benarroch:
Right. [inaudible 00:04:47].
Edo Liberty:
It's in their contracts. It's in their meeting transcriptions, it's in their-
Stefan Weitz:
It's in their Slack. It's in their [inaudible 00:04:53].
Edo Liberty:
Slack, in their emails, in their historic context, in their documents, memos and decisions.
Stefan Weitz:
The filing cabinets and [inaudible 00:05:01]. Yeah.
Edo Liberty:
Exactly. And you ask them, what do you do with that? And the answer is almost always nothing. "We store it."
Joe Benarroch:
That's interesting.
Edo Liberty:
Right?
Joe Benarroch:
Right.
Edo Liberty:
Whatever.
Joe Benarroch:
It's just a repository [inaudible 00:05:12].
Edo Liberty:
We have a wiki, we can go and-
Stefan Weitz:
We have a wiki.
Edo Liberty:
No, I'm serious. They do nothing with it. In the best case, they make it accessible, then nobody accesses it. But if you could now take the smartness of AI, so think about how much... We've grown to assume these models are pretty God damned smart. Right?
Joe Benarroch:
Right.
Edo Liberty:
Think about what would happen if you also and you... Not only are they that smart, but they also know everything. They've read all my documents, they've read all my contracts, they listened to all the meetings, they've read all the memos.
Joe Benarroch:
Yeah, they've extracted the right things.
Edo Liberty:
They actually read the entire wiki, they remember everything.
Joe Benarroch:
Yeah. They haven't glossed over it.
Edo Liberty:
[inaudible 00:05:58] that smart. They actually know everything about my business. I can actually ask, "Hey, why is this the customer of mine saying this to me and what's actually happening and what do I need to know that I don't know yet?" Or, "Oh, the analysis of this, we are running an analysis, some business, how does this relate to other stuff that we've seen before and how should I react?" These are deep questions. It's not a model thing. It's very specific to your company, your history, your content and so on. So connecting those two worlds, getting those really, really smart models to actually consume, know and have access to everything in your company, that's where you see enterprise really just like, the eyebrows just go like, "Holy ****."
Joe Benarroch:
Sure, totally.
Edo Liberty:
"This is possible?"
Joe Benarroch:
Okay. You've totally changed, you've opened up my aperture on that. I want to get to your origin story.
Edo Liberty:
Sure.
Joe Benarroch:
Because if we're talking all about this, I mean you pretty much originated and built Bing.
Stefan Weitz:
Well, yeah, one of many, but yes.
Joe Benarroch:
One of many. But okay, still.
Edo Liberty:
Yeah.
Joe Benarroch:
We don't have to go all the way back there, but get me to HumanX. Give me the origin. What was that aha moment?
Stefan Weitz:
Yeah, that's a good one. Well, it was actually from those early days, I've been a nerd, not a professor, certainly not at Princeton, but I've been a nerd for a long time, since I was a little kid writing code and always wanted to have that C-3PO and I wanted to have R2-D2... Oh no, he couldn't speak. I wanted to have Kit from Knight Rider. I worked on this for a long time. In the 2010s, we had super simple models that kind of worked. We tried to recreate the entire world in digital. All this crazy stuff, just wasn't ready. The tech wasn't there yet to do a lot of what we wanted to do. So I honestly felt like AI was something I could still aspire to see in science fiction, I could have my Star Trek communicator, but the reality was in my lifetime, probably weren't going to see that natural human computer interface that we all were so excited about.
And then we saw Transformers pop out. I got a little more excited and we saw Chat pop out people now were actually understanding what the **** I did all those years. My mom actually knew what I was doing back in the day, which was great. She was, "That's what you were work..." "Sort of, Mom, sort of, sort of." So I got very excited that we now have seen this opportunity for more people to experience the power of enhanced human computer interaction. So I was super excited and all these companies were getting funding and coming out I'm like, "Wow, this is a great opportunity for humanity." And then I started calling my buddies who were running these large companies and I said, "Hey, so what are you doing with all these new pieces of technology?" And I got three forms of answers.
One, it's like, "Why are you calling me?" It's a bad answer. The second was we got-
Joe Benarroch:
"Who are you?"
Stefan Weitz:
Yeah, exactly. "Do we work together?" "We did. We did. I promise we worked together at some point."
Joe Benarroch:
"Who?"
Stefan Weitz:
Yeah, "Weitz? No."
Joe Benarroch:
[inaudible 00:08:52].
Stefan Weitz:
"Forget about it. Forget about it." The second batch was, "We have a great plan," which was a complete lie. They didn't at the end of the day. The third one was kind of like, "I'm not really sure what to do." The reality was the vast majority of large enterprises really didn't have a clue as to how they're going to deploy this technology. And so as a guy who also runs a venture capital firm and invests in these companies and as a nerd who wants to see AI come to fruition and actually benefit all of humanity, it's a bad place to be. If the buyers don't know what they're doing and the sellers are just creating new and more and better technology every single day and there's no ability to match those needs and opportunities together, we're going to under-deliver on the promise of AI and we're going to have dramatic corrections in the market as companies fail to achieve sufficient revenues to justify the increasingly high valuations. So that was the reason behind HumanX, it was how do we create an opportunity? How do we create a forum, a marketplace, a place to come and be inspired by these innovations, but also then to take real action and move into, "What do I do with this stuff now that I understand potential of what it could actually be?"
Joe Benarroch:
Well it's definitely... I mean you've already signed up 50 speakers for [inaudible 00:10:01].
Stefan Weitz:
Actually 170, but we'll advertise 50, but-
Joe Benarroch:
Good correction. Good correction.
Stefan Weitz:
Yes, there you go.
Joe Benarroch:
All right, so before we... I have a question both. You've talked about legacy behaviors, you've talked about some new behaviors. So I have a quick question. Are the legacy gatekeepers or the new gatekeepers, are the legacy gatekeepers completely toast or are we toasting the new gatekeepers? Are we celebrating the new gatekeepers of information?
Edo Liberty:
Look, I think the beauty of technology is that nobody has the patent and whatever can fence people off. The question is how good is the technology? How well you can do this and what is the right pattern, right? At the end of the day, you need to get better results. You need to actually unlock value. You need to really empower people to do amazing work. What we see is that, yes, the new way sort of combining search, AI, knowledge now with agents taking on the world, making those trustworthy and useful and efficient and sort of an integral part of day-to-day business doing, right?
Joe Benarroch:
Yes.
Edo Liberty:
It's just part of work. It's still hard. And if you rely on legacy technologies, if you build on sort of like bolt-on, sort of like, "Hey, I take my old database and my old search engine and my old Wiki and I Frankenstein-ianly smoosh them together and hope for the best and now I call that AI," that's what you're going to get. Right?
Joe Benarroch:
Yeah.
Stefan Weitz:
Yeah.
Edo Liberty:
And if you're like, "No, I understand this is the wave of the future. Yes, this is a new technology. Yes, this is a little bit scary. I'm not 100 percent sure what I'm doing, but I have smart engineers and smart scientists and they can help figure this out. I'll take a leap forward. We'll bring in the right technologies, we'll bring in the right infrastructure, the right model, the right database, the right talent to the team and say 'Let's move forward.'" Vanguard started working with us almost two years ago. Vanguard, it took some time, but now their support is like 12 percent more accurate, a lot more performant, a lot more [inaudible 00:12:27].
Joe Benarroch:
Took some time to get them...
Edo Liberty:
For them to learn how to operate their own systems and improve and use the models correctly and use the database correctly.
Joe Benarroch:
And training, and...
Edo Liberty:
It's a new technology. You have to learn how to use it. It doesn't pop out the box ready to go. It's like everything else. Same thing, we have startups like custom GPT know, building GPTs for a million people. Again, they're natively building it from the ground up. Within a few months they were way ahead of the competition of other stuff. So we 100 percent see a new wave of people who are embracing new technology and accelerating amazing outcomes. I think the old guard is trying to catch up and is trying to bolt on whatever we are doing, sort of like, "Oh, we also do that now."
Joe Benarroch:
Sure. Yeah. They're trying to modernize pretty quickly.
Edo Liberty:
Again, it works to some extent in the sense that they already have a big platform, a big channel, and they have a lot of customers and some people already have whatever it is they are selling. And so like, "Oh, okay, fine, we'll try that first." The pattern of success we see is people doubling down and say, "No, we're not going to try to half-step this. We want to really embrace it and say, 'Hey, agents are going to change the world. We have to figure this out proper. We can't just-'"
Joe Benarroch:
Totally, you can't [inaudible 00:14:00].
Edo Liberty:
Exactly.
Joe Benarroch:
What's your-
Stefan Weitz:
I mean, I don't think they're toast, but I think they're in the toaster at this point and they're getting browner. But I think that's actually a good thing. I was actually talking to a big producer yesterday out west and he was talking about how this is going to impact the production of TV, film, everything else.
Joe Benarroch:
It absolutely will.
Stefan Weitz:
And he was projecting that they're going to be able to save 50 percent on pre-production, 25 percent on production itself, and 60 percent in post-production using all these new tools, which of course some people will say, "Oh my gosh, this is the end. Now we're seeing the destruction of the creative industries and for those in the space-"
Joe Benarroch:
Not at all.
Stefan Weitz:
And his point was, "Sure, we will have fewer people on each project, but we can do a film for one third the cost of traditional, which means we're not going to make less films, we're going to make more stories." So actually the employment's going to probably be roughly the same if not more, but spread across more stories. So for me, when I think about gatekeepers and think about incumbents, I think about those who have... If you think of Hollywood, for example, the traditional studio systems, a lot of infrastructure, a lot of costs is baked into that. These new players can come in and actually disrupt how it's all done and really expand the opportunity for more people to make great stories. So I think it's probably both, to your point there.
Edo Liberty:
But it's also just changing the way work is done. If whatever, animators used to draw and then if they wrote software-
Stefan Weitz:
Cell by cell, exactly.
Edo Liberty:
There's nothing that says that, "Yeah, the job of an animator to train a model next," right?
Joe Benarroch:
That's right.
Edo Liberty:
You are a model trainer, you model will... Whatever. So the work itself changes because the tooling changes. That's what I said. You don't fight this by trying to say, "Oh, we'll animate faster." You fight this." Oh no, no.
Joe Benarroch:
Definitely.
Edo Liberty:
"Okay, the world changed now it looks like this. I need to get really good at this."
Joe Benarroch:
But what do you think or have we seen it... What's that tipping point where people aren't afraid of it anymore, businesses aren't afraid of it anymore, you're not talking to companies like Vanguard who are taking a lot of time to deploy it or... What's that tipping point? What's in your mind where just a certain thing has to happen?
Stefan Weitz:
To me, I think it's when people actually are able to touch and feel a breakthrough that otherwise wouldn't be possible without it. And so that's the mass scale. So imagine we finally crack the code on cancer, as an example, thanks to the alpha-filter work that's happening, or identify a protein which causes Parkinson's. And I think when people actually come to see this as a little dystopian, see it as inevitable, I think that will then, like it or not, it will create this level of, "Okay, now we really... It's not a flash in the pan. It's not a 3D printing [inaudible 00:16:52] this consumer thing back in the day. It's not blockchains that's going to revolutionize everything and has done great stuff, but also is necessarily blowing the whole world up." I think this is one of those situations where when more people can see the actual positive and potentially negative impact of this, I think that to me is a tipping point where it'll be obvious that it has to happen.
Joe Benarroch:
I agree. I know I'm actually flying today to go meet with Patrick soon.
Stefan Weitz:
Yeah.
Joe Benarroch:
And he was actually mentioning the same thing that he thinks healthcare is where the real critical mass-
Stefan Weitz:
Yeah, I agree.
Joe Benarroch:
... And celebration of these big technological... Technology in general, that... You share the same viewpoint.
Stefan Weitz:
It's the biggest segment of the economy to a certain extent.
Joe Benarroch:
Correct.
Stefan Weitz:
I think it's probably the least loved segment of the economy by people.
Joe Benarroch:
Sure.
Stefan Weitz:
And has a ton of opportunity, and AI in particular is a place where you can actually take all the unstructured data, all the chart notes, all the tests from over the years, a 20-year longitudinal history of somebody and actually now make sense of it. You couldn't do that.
Joe Benarroch:
No, totally.
Stefan Weitz:
A doctor could not do that, can't do that today. It's impossible. It's not humanly possible. So that's why I get excited about it.
Edo Liberty:
When people start... A, when they have experiences for the first time that it's like they have this a ha moment. It's like, "Holy ****, this is like, okay, I can do this now? All right."
Joe Benarroch:
Yeah.
Edo Liberty:
"This changes things for me." And sort of the flip side of it, when you take it away and they feel like-
Joe Benarroch:
Oh, they need it.
Edo Liberty:
What? This is-
Stefan Weitz:
No, that's good too.
Edo Liberty:
Like, "Oh, I search for something. It's like no search results." They're like, "What? What are you talking about?"
Joe Benarroch:
Their experience is subpar.
Edo Liberty:
It's like, well, this, not-
Joe Benarroch:
[inaudible 00:18:26] expect it and it's not there anymore.
Edo Liberty:
They know it's there, right?
Joe Benarroch:
Yeah.
Edo Liberty:
Or at Pinecone we have 1000 customers, we have three salespeople.
Stefan Weitz:
Oh my gosh.
Joe Benarroch:
Okay, you have three salespeople?
Edo Liberty:
Yeah. They don't write notes, they don't enter information in Salesforce, they don't... Whatever. The conversation ends, everything flows through. We get a Slack message of who's the buyer, who's this, who's the decision maker, what's happening, what next steps are? Everything goes in, everything is automated.
Joe Benarroch:
I see your exact point.
Edo Liberty:
My job is to just work and get done, right?
Joe Benarroch:
But I see your exact point.
Edo Liberty:
If you take it away from them, "I need to sit here and type for a half an hour. Are you nuts?"
Joe Benarroch:
Yeah. They don't want to be a stenographer. They want to be an inventor.
Edo Liberty:
It's crazy, right?
Joe Benarroch:
Yeah.
Edo Liberty:
So it's like that, it's both like, "Hey, stuff is now possible." Or again, for me also internally, again, we can talk about knowledge and so on, but yeah, I mean it's what's possible, and then when you take it away, you feel like you're in the dark ages. It's like driving without a GPS or whatever, without your Google Maps.
Stefan Weitz:
[inaudible 00:19:28].
Edo Liberty:
You're like, where am I? What's happening?
Joe Benarroch:
That's actually very true.
Edo Liberty:
:Just take it away."
Joe Benarroch:
You can't, right. The whole tech with a heartbeat type mentality, it's all part of our lives. We're wearing it.
Edo Liberty:
Yeah, yeah.
Joe Benarroch:
It's there. I could never imagine getting in the car with my kids without Google Maps.
Stefan Weitz:
No.
Edo Liberty:
If you sort of stop the AI models-
Joe Benarroch:
Totally.
Edo Liberty:
... And I say AI because it's embedded everywhere.
Joe Benarroch:
Sure.
Edo Liberty:
If you stop that in your workplace, in two or three years, your company would grind to a halt.
Joe Benarroch:
Absolutely.
Edo Liberty:
Driving up a fleet without.
Stefan Weitz:
You'll be lapped by the competitors too.
Joe Benarroch:
I wanted to unpack Vector Database because you brought up the entertainment industry and obviously when I was like, "I don't know much about a Vector Database." 7.5 billion of them within Pinecone alone, is that correct?
Edo Liberty:
What, the-
Joe Benarroch:
Indexes.
Edo Liberty:
Yeah, knowledge. Yeah. Indexes. Yeah.
Joe Benarroch:
But I was fascinated by the idea of searching by style, searching by vibe, helping with understanding DNA sequencing, understanding financial fraud, unpacked, a Vector Database for the individual, the business partner who might not really understand A, what that is and how that can be applied or how they can actually build one for their company?
Edo Liberty:
I'll first say for the business leaders on... Listening to us, your engineers already know. Don't worry about it. They know, they even know [inaudible 00:21:02].
Joe Benarroch:
Spoken like a true founder.
Edo Liberty:
No, I'm telling you. They know Pinecone. They know, they've used it before. Then if they don't, they're going to go on our website and they're going to figure out the docs in five minutes. If not, they're going to ask a coding agent to tell them to go implement this search on Pinecone and within five minutes they have it running. So your engineers know.
Joe Benarroch:
That's good.
Edo Liberty:
That's number one. But number two, I think it's important to sort of understand what the heck is a vector database? Why do you even need a different infrastructure? What the **** is happening here? And I think it's best understood by analogy. Okay?
Joe Benarroch:
Yeah.
Edo Liberty:
If you think about, if you sort of walk down the street and see your buddy from high school and you're like, "Hey." You literally know who this is. You might take a second to remember their name or whatever, but you know the person. Something happened in your brain. First of all, they don't look the same anymore.
Stefan Weitz:
They got a little fat [inaudible 00:21:57]. Lost some hair. You're like-
Edo Liberty:
Hopefully not. [inaudible 00:22:00] fit and super jacked.
Stefan Weitz:
Let's go with yours, let's go with yours.
Joe Benarroch:
Listen. I'm using Hims. I get it.
Edo Liberty:
No, but something happened in your brain.
Joe Benarroch:
That's right.
Edo Liberty:
That go and went and fetched this person's face and matched to the tens of thousands of other faces you've seen that day and the tens of thousands of people that you potentially know. And just to make things even more sort of abstract is like what your brain receives is not what your eye consumes. Your retina is sort of like a CCD kind of very messy one. But what your brain, gets your temporal lobes is the output of the visual cortex, which is a jumble.
It's an array of activations of neurons that has nothing to do with the original image. It's just a jumble. That's what you remember. That output of the visual cortex is what you remember. There's no face-
Stefan Weitz:
Correlation to it, yeah.
Edo Liberty:
.... in your brain anywhere. But somehow this thing sort of finds its way, "Hey, you know what? This is my buddy, Mark." That layer of activation of neurons, that kind of memory, that ability to remember by analogy, by meaning, by style, by connotation and so on, that's a brain thing. That's a memory thing. The ability to match those by similarity, by context and so on, that's a vector database. The array of numbers of the activation of neurons. That is the vector, right? This is how neural nets remember stuff. We've built them to mimic the way compute works in the brain and we've built vector databases to mimic the way memory works in the brain, and you have to bring them together.
Stefan Weitz:
That's a great, great analogy.
Joe Benarroch:
That is fascinating.
Stefan Weitz:
And sorry for Mark, who I'm sure is jacked.
Joe Benarroch:
I'm sure Mark is jacked. Mark is my husband.
Stefan Weitz:
There you go.
Joe Benarroch:
So I was like, "Oh, that was good." Where did that... So it's basically the study of the brain is analogous to the technology of Pinecone.
Edo Liberty:
I mean it's analogous. It's sort of in the same way that attention models are not quite literally how the organs in the brain that are processing language works. I mean it's analogous. It's sort of like planes and birds and all the usual analogies for these things. But yeah, I mean it's inspired by the functionality is the same. It needs to provide the same function. Of course, the underpinning implementation is different, but it's definitely inspired by it and most importantly needs to provide the same function.
Joe Benarroch:
You must be coming across so many of these types of companies, A, on a daily basis but B, wanting to be a part of human X simply to deconstruct what Edo has just deconstructed for us. Tell us some companies that are coming into the fold. Tell us about HumanX next year.
Stefan Weitz:
Oh my gosh. Well, the companies are coming into the fold are ones that most people have probably already heard of. Like Brett Taylor's coming from Sierra, which is really exciting.
Joe Benarroch:
Yeah, I saw that, that's great.
Stefan Weitz:
And very happy to have them. Ali Ghodsi from Databricks is actually coming as well, which is really exciting. Matt Garman from AWS is coming. These are big players. And then we have hundreds of smaller startups coming in because that's the cool thing about this industry. You do have the big players with a lot of money and a lot of capabilities building incredible innovations every day. And now the garage guys and gals, they're back because the tools, the tooling, they have to create incredible one person, billion dollar companies, we'll see if that actually happens, are there like never before. When I was a kid building software way, way, way back in the day, I remember just the cost of standing up your own colo or your own rack server or going to rack space and getting a rack you could put your boxes in and having some person maintain those things physically. It was absurd, right?
And now literally you can go to your garage and have an idea and build something that could be revolutionary to even a niche customer segment. That's the cool thing. That's what we're seeing at HumanX is a lot of companies who never before would've been viable because the money would've taken to build the solution for a somewhat niche-y customer, the equation would've been wrong. Well now, because the cost to build these systems has dropped so dramatically as a result of a lot of innovations that we're seeing, you can build a product which may only have a $200 million TAM, and that's pretty good. Is it going to be a publicly traded company? Probably not. Is it going to make you a billionaire? Probably not. But is it a business that you can run as a wonderful lifestyle business for the rest of your life and clip coupons until you're in the grave? Probably so, and that's [inaudible 00:27:01].
So we're seeing a lot of those at HumanX as well. People who are coming in, and it's exciting for the larger companies who come and for the enterprise executives who come because they too see these innovations and they'd go ask their engineers, "I saw this thing at HumanX, why can't we do that?" And I'm sure all the engineers love that conversation when the CEO gets back and says, "Hey, I saw this cool thing. Why can't we do that?" Sorry, engineers. But we're seeing a lot of that bouncing around as a result of those connections that are happening there. So I'm excited just for the diversity of people we have at the event, both, you have all the enterprise customers there, all the enterprise companies there, and the enterprise innovators who are doing great work. And you have this amazing middle and lower layer of companies that I don't think there's really a place today for them to go to be seen by these enterprise buyers and have inspiration for them.
Joe Benarroch:
You've essentially triangulated the model of a development conference at Davos, an Inksgroko. I mean, you've kind of put that into the heart of what is in our future within the AI space, which is pretty fantastic.
Stefan Weitz:
That was the idea. It's like how do you... I told the whole team from day one, how do you move from inspiration to action? You could go to a lot of events that you get the person on stage who was just giving you the "Want to be cool. And we all have flying cars, and we don't have to go to school anymore, and we all have infinite amounts of money," and it's great. Everyone's like, "Yes, rock on." But then there's that translation layer between, "Yes, I want to go there," and how do I actually pull it off? Thank you.
Joe Benarroch:
Good, 30 Rock.
Stefan Weitz:
30 Rock. Yeah. I'm glad someone got that.
Joe Benarroch:
Good, 30 Rock.
Stefan Weitz:
Most of my team is too young, so all my dad references just go over their heads. It is the saddest day of my life.
Joe Benarroch:
Liz Lemon.
Stefan Weitz:
Yeah.
Joe Benarroch:
[inaudible 00:28:46] forever.
Stefan Weitz:
So how do you actually get them to that level? And that's what I was missing. It's like I can get excited by a lot of stuff I do every day, but then how do I actually do it? And you're like, "Oh, now what do I do?"
Joe Benarroch:
All right, so actually I'm glad that you brought up teams because I do want to talk about culture. I was fascinated by the fact that you said you have three people-
Stefan Weitz:
Yeah, me too.
Joe Benarroch:
... For a thousand customers.
Stefan Weitz:
That's crazy.
Joe Benarroch:
There's so much-
Edo Liberty:
Actually we have 10,000. We have 1000 that we are actively managing and talking with.
Joe Benarroch:
[inaudible 00:29:12]. 10,000. Okay. That's called a massive model for scale and efficiency.
Stefan Weitz:
Scale.
Joe Benarroch:
Your teams are obviously inundated with the warp speed of the world. How do you personally stay focused? How are you keeping them focused, to your point? And then I want to ask a little bit about some cultural shifts. You were joking around about how a lot of your jokes go over their head, but are they changing the way that you lead? Let's start in the place of... Let's just start in the place of how are you staying focused with just the inundation of ideas and things coming at you?
Stefan Weitz:
Well, I can tell you one thing. That is not my superpower, staying focused. And I think I hire people who drive that focus, because if it's up to me, I would do 50 things every single day at once, which is part of my problem. So I can tell you tool-wise though, I'm using a bunch of AI tools that help me at least corral my ideas into something that is both actionable but also trackable. So that's a lot of... So it's just ways to keep me from kind of spitting off into 1000 different directions. How do I narrow down my ideas into something which is concrete versus something more amorphous? That's me. And then as far as what the teams have taught me, I've been managing for a long time, but this was not in this era. Obviously no one has in this era been managing for a long time.
And it was interesting, one of my team members said I tend to go to the how, not the what or the why. That's just as an engineer, I just think about how do I go solve a problem? My wife loves that, obviously.
Joe Benarroch:
[inaudible 00:30:49].
Stefan Weitz:
[inaudible 00:30:48]. Super big fan of that. They're like, "I just want to talk." "Why? You have a problem. Here's the answer. What is your problem?"
Joe Benarroch:
[inaudible 00:30:56].
Stefan Weitz:
But that's something that the team is there too. It's like, "Hey, can you just... Rather than just give us a solution, can you actually tell us the why they want the context and the how?"
Joe Benarroch:
Context.
Stefan Weitz:
And I'm like, "Okay." But it does actually force... It's actually beneficial to me even as I think about working with AI tools, rather than telling, "I want you to go just do this," you say, "Here's what I'm trying to accomplish." And then let the tool actually figure out the best way to... So to me, it was an interesting analog of how the team would tell me, tell us the context and then how I use that to actually work better with AI tools, because they actually want the context. They can get to the how real fast. I don't need to tell them-
Joe Benarroch:
Sure.
Stefan Weitz:
I need to tell them what I want to do and then [inaudible 00:31:33].
Joe Benarroch:
They might be training the models better to give more context, and also that's fascinating.
Stefan Weitz:
Yeah. [inaudible 00:31:38].
Joe Benarroch:
They might be asking the better questions to these tools.
Stefan Weitz:
Yeah, I think so. So thanks to my millennial team for forcing me to go to the what and the why, not the why [inaudible 00:31:46].
Joe Benarroch:
To get to the feels.
Stefan Weitz:
To the feels. Hasn't helped at home.
Joe Benarroch:
How many people at Pinecone?
Edo Liberty:
We about 100 now.
Joe Benarroch:
But you're scaling, I mean you've scaled. 10,000 customers?
Edo Liberty:
We automate everything.
Joe Benarroch:
So how are you keeping focus? Is there a cultural credo? How are you staying focused on the value that you know you ultimately-
Edo Liberty:
Yeah.
Joe Benarroch:
... Versus just scaling to create more capabilities?
Edo Liberty:
I suffer from the same, of course-
Stefan Weitz:
Every founder does.
Edo Liberty:
... The same kind of situation. And we're both technical folks. Again, it's very natural to think about solutions and just find cool ways to do things. I do think that has actually a lot of value. I tell my team a lot in product and engineering and so on, that sometimes the product is the fact that something is possible. It's not that, "Oh, it would be nice to have a flying car." It starts with somebody's like, "Holy ****, if I do this, it can levitate."
Stefan Weitz:
That's so right.
Edo Liberty:
That's the real inception of the flying car. Not, "Hey, it'd be great to have a flying car." The vibe makes no... That doesn't help anyone. It starts with somebody figuring it out.
Stefan Weitz:
Sure.
Edo Liberty:
And so oftentimes it's like, "Hey, engineers, scientists. Tell me what's possible. What can we do today that we couldn't do yesterday?" And then, "Okay, can we make a product out of it? Can we improve something? Can we streamline something?" Maybe it's just a... Whatever. It's just a gizmo that nobody cares about.
Stefan Weitz:
It's just cool. Yeah.
Edo Liberty:
Great. Onto the next thing. And so I think we need to give... I mean, again, as a scientist, as a technologist, I love that pattern. I was like, something is now possible that wasn't possible before. And then, okay, now what value does that unlock? Rather than, "Okay, what do customers ask me to do? And then, okay, can I do that?" Because oftentimes they ask for something that a million other people asked before, there are 1000 other ways to think about how to deal it. And odds are that your guess of how to solve it is not better than anybody else's guess on how to solve it. And so you sort of go into this sort of old-fashioned way of building businesses. So that's one thing. But the second thing is sort of like you stay focused by staying small to me.
Stefan Weitz:
Yeah.
Edo Liberty:
You almost create scarcity and you're like, if you can't do everything, you'll naturally do the most important things.
Joe Benarroch:
Sure. Well, you're proving that staying small and scale are definitely not at odds.
Edo Liberty:
I mean, it doesn't help.
Joe Benarroch:
Sure.
Edo Liberty:
I mean it's a lot of tension, but it creates this... The environment of scarcity creates prioritization everywhere. We don't have a problem of salespeople chasing data opportunities-
Stefan Weitz:
[inaudible 00:35:03] leads, yeah. Exactly.
Edo Liberty:
... Because whatever, because they don't have time.
Stefan Weitz:
No.
Edo Liberty:
Engineers are not stretching their heads on what to build next. I mean, there's a mountain of execution.
Stefan Weitz:
Backlog [inaudible 00:35:15]. Yeah.
Joe Benarroch:
So I imagine... Well, I would think. I mean, I've been at companies, Facebook, X, who are founder led. I imagine a lot of people look to you for the inspiration, the innovation, the ideas. What is a couple of things that your team has taught you that has given you a way to look at things differently?
Edo Liberty:
I mean, I learn from them all day long. I mean, I tell my team all the time, "I can't do any of your jobs."
Joe Benarroch:
That's totally right.
Stefan Weitz:
That's good.
Edo Liberty:
"I'm not a finance guy. I'm not accounting. I'm not a product manager. I'm not a VP of engineering. I'm not a marketer. I don't know how to do any of those jobs. And so I, A, rely on you to teach me and to tell me how we should best do that." But I have to rely on them too. And I'm learning all the time. And I love that. That's a part of maybe... A part of the affliction of being a founder. Is that.
Joe Benarroch:
The affliction.
Edo Liberty:
No, I'm serious.
Joe Benarroch:
I hear you.
Edo Liberty:
You have to be somebody who's very happy to learn all day long and be sort of average.
Stefan Weitz:
The [inaudible 00:36:21].
Edo Liberty:
Below average on 20 different professions.
Joe Benarroch:
Yeah, absolutely. Absolutely.
Stefan Weitz:
Yeah, yeah.
Edo Liberty:
But enough to actually understand what's happening, be a backstop a little bit, be a little bit of a sounding board, be able to sort of play ball in these 20 different professions. And again-
Stefan Weitz:
Okay.
Edo Liberty:
So I learned, again, maybe you're asking for a specific thing.
Joe Benarroch:
No.
Edo Liberty:
I learned of six plus years of running Pinecone. I've learned an infinite amount on anything from accounting to marketing, to sales, to running engineering, let alone, of course, the science and our product building. So yeah, there's something good.
Joe Benarroch:
Yeah, when the founder is specific.
Stefan Weitz:
[inaudible 00:37:07].
Edo Liberty:
Yeah.
Joe Benarroch:
Were you the inspiration behind the Pinecone Pioneers?
Edo Liberty:
Pioneers are really, there's a lot of people who really love building on Pinecone, and they become really, really proficient in vector search and building these amazing applications. We have now 100,000 developers join the platform every quarter.
Joe Benarroch:
Geez, a thousand?
Stefan Weitz:
Hundred thousand.
Edo Liberty:
A hundred thousand.
Joe Benarroch:
A hundred thousand.
Edo Liberty:
New developers on board to Pinecone every quarter.
Joe Benarroch:
[inaudible 00:37:39].
Edo Liberty:
And many of them are just learning. Many of them are just poking around. Many of them are just, "Oh, what is a vector database?" But there is sort of the funnel, at the core of that community, there are few thousands of people who truly get it, who built amazing things, who created amazing volume, figured a lot of things out, and they're excited about it. They want to share it. They want to share their knowledge, they want brag about it in a good way. I mean, they deserve to brag it, but they build amazing things, right?
Joe Benarroch:
Absolutely. [inaudible 00:38:13].
Edo Liberty:
It's great. It's great. And those are the Pioneers. People who love our product, know what to do with it, are excited to share with others, are excited to unlock others, and we say, "Let's celebrate that. Let's give them a platform. Let's drive them forward. Let's find ways to make them more engaged, more happy, give them more love, give them more resources, and set them loose on the world. Have them teach everybody."
Joe Benarroch:
We need to get the Pioneer program on the bell.
Edo Liberty:
All right. That's [inaudible 00:38:44].
Joe Benarroch:
That's a really, no, that's a really-
Edo Liberty:
I would love that.
Joe Benarroch:
I mean, yeah, because you have some really good founders.
Edo Liberty:
Yeah. We'll fly them in.
Joe Benarroch:
I mean [inaudible 00:38:50], and I think that's great.
Edo Liberty:
Yeah.
Joe Benarroch:
No, because you're ultimately just, you're giving them a stage.
Edo Liberty:
100 percent. I'm for it. Let's do it. I'm serious.
Joe Benarroch:
That's great. If you were to toast one big idea, one big moment, either from the year or that you've seen, what would it be? An idea, a person, a moment.
Stefan Weitz:
I am so excited that we are finally seeing some of the breakthroughs in fusion energy generation. That's where I get excited. That and some really amazing breakthroughs from Microsoft on quantum computing. So those are my two big things. If I had my Bingo card for the next five years of excitement, if I had to write that up, it would be quantum plus fusion, because at that point you have free energy forever. Free energy forever basically, and basically unlimited compute. So I'm excited for those two things.
Edo Liberty:
It's hard to top that. He went all out.
Stefan Weitz:
I'm just thinking, what's [inaudible 00:39:47]. That's what I excited about.
Edo Liberty:
[inaudible 00:39:49].
Joe Benarroch:
That's good though.
Stefan Weitz:
I read a lot of physics paper.
Edo Liberty:
[inaudible 00:39:50] that was published [inaudible 00:39:50].
Stefan Weitz:
[inaudible 00:39:50] founder.
Edo Liberty:
Once you go cold fusion, okay, fine. This is, you [inaudible 00:39:54] cleaned the table. All right, free energy for everybody forever.
Stefan Weitz:
Yeah.
Joe Benarroch:
I'll take that.
Stefan Weitz:
Not bad.
Joe Benarroch:
You guys. This has been great. We're getting ready for the opening bell.
Edo Liberty:
Yeah.
Joe Benarroch:
Speaking of, it's Lithia [inaudible 00:40:07]. You talked about flying cars earlier. I didn't know if that was a throw to Lithia.
Stefan Weitz:
Yeah.
Joe Benarroch:
But they're getting ready to ring the bell, so cheers, gentlemen.
Stefan Weitz:
Cheers, yeah.
Joe Benarroch:
Thank you.
Edo Liberty:
[inaudible 00:40:17] a lot.
Joe Benarroch:
Breakfast on the balcony. Really appreciate it.
Edo Liberty:
[inaudible 00:40:20]. That was quite fun.
Speaker 5:
That's our conversation for this week. Remember to rate, review and subscribe wherever you listen and follow us on X at Icehouse podcast from the New York Stock Exchange. We'll talk to you again next week. Inside the Icehouse. Information contained in this podcast was obtained in part from publicly available sources and not independently verified. Neither Ice nor its affiliates make any representations or warranties expressed or implied as to the accuracy or completeness of the information, and do not sponsor, approve or endorse any of the content herein, all of which is presented solely for informational and educational purposes. Nothing herein constitutes an offer to sell, a solicitation of an offer to buy any security or a recommendation of any security or trading practice. Some portions of the preceding conversation may have been edited for the purpose of length or clarity.
