Pinecone

“We have software engineers who are completely AI-pilled and using Cloud Code, Codex, our own internal agent we call Pinecone that we used to run much of the company on, and they estimate they are between three and 20 times more productive in terms of features …”

“Just one that we did recently called Pinecone, raised over a hundred million. There was a point at which they pitched 40 VCs, all passed, they had one meeting left, and it was kind of four months of runway, make or break, and, you know, they ended up Making th…”

“Even now to this day, we're five years in. Pinecone is a raving success and so on. I have, I would have done much to this day. I mean, put my stock aside, I will have done a lot better had I stayed at AWS, right?”

“The term vector database didn't exist. It just didn't exist. We'd, nobody talked about it this way.”

“We literally asked them like, Hey, you internally in your team, you talk about using this pine cone thing. What do you think it is? What do you call this internally? Seven out of 10 would say, we just call it a vector database. It's like, that's where it's, we…”

“I have at that point spoken with Every VC in the Bay Area have gotten a brutal rejection from each and every one of them. Like I swear to God, like there's not a third rate VC in the Bay Area that I have not pitched to with like all my, you know, with, you kno…”

“We got to the point where we would saturate whole, like, regions in, in, in AWS and GCP. We just couldn't meet the demand.”

“We finished that year with roughly two million in revenue and roughly... We finished with roughly two million in revenue with a few hundred customers and an insane growth.”

“VCs would send me their pitch deck on Pinecone about why it's a great investment. Their entire research, they would send me like, 50 page documents on all the market research and exactly how we're the leader and how we're gonna win and how we're, what we're go…”

“We closed the round between when we signed the term sheet and when we, when, and closing, we five X'd the number of five paid customers. And this is the year go, you know, go, this is the year after. Yes. This is a beginning of 23.”

“If we need to do this at what, if we need to do this at the, you know, hundred X scale of where we are today, we're going to have to build something very, very, very different. Separate, separates writes and reads, that multi, that has multi-tenancy sort of bu…”

“Well, we had to rebuild the entire system, which is a huge bet for a company, for a database to go rebuild the, like, what it is mid-flight and keep serving all these workloads was a massive strategic bet that was sunk in probably more than a year of the compa…”

“So now we're, we could run, we can offer the service to be 10 X cheaper. So the cost, so people who used to pay us, you know, three, 4000 dollars a month started paying 200 bucks a month.”

“I'm telling you in five years, the thought that maybe the conversational agent that you are talking to inside your company hasn't Listen to all the sales calls that ever, ever happened. The company knows everything about your product and can talk about it inte…”

“I think the one thing that founders Tend to misjudge is how much control they have over anything in both directions. Sometimes you feel like you're constrained by your team or by your customer or by this or by that, and you feel super, you know, stressed and b…”

“Compared to one of the leading vector-only databases, Pinecone, a PG vector scale is 28 times faster for a high recall scenario.”

“The monthly costs of running such a PG vector scale deployment on AWS is 75% less expensive than PyIncome.”

“We invested seven million dollars at 35 post, right? Like, oh, 35 post for a company with, like, no revenue and no customers and blah, blah, blah. That's what we did.”

“In 2019, the earthquake had already happened. Deep learning models and so on have already been grappled with. Large language models and transformer models like BERT and others started being used by the more mainstream engineering cohorts.”

“And you could see that if you augment all of them with RAG on, even on the internet, which is data that they were trained on, you can reduce hallucinations significantly up to 50% sometimes.”

“Notion Q&A now runs on, on Pinecone, and they serve essentially question answering with AI to tens of thousands and probably hundreds of thousands of their own customers.”

“Gong does the same thing with sales calls. Again, serves all of their use cases for all of their customers, and so on.”

“We've tested it with tens and tens of billions with live customers and live traffic.”

“Those other products don't work. They don't work either because they don't scale in terms of the efficiency scale, cost, the trade-offs that they can offer, because they're not designed to do this. They're designed to do something else. They kind of thought ab…”

“Our research actually shows that when you do this well, we, you very rarely need keywords alongside embeddings, but getting embeddings to perform perfectly is, is actually, it could be quite intricate.”

“And in fact, if you look at statistics, even companies that are open source, 99% of the contributions are actually from the company itself. Not 99, but high nineties.”

“And in fact, we already see, even though new players in the vector database space that, that, that basically started to try to take us down, all took the open source angle. We already see them, even young as they might be, they are already struggling, struggli…”

“Serverless is the fourth complete, almost complete rewrite of the entire database at Pinecon.”

“There's plenty of evidence that increasing the context size doesn't actually improve results unless, you know, you do this very carefully, right? So just what's called constant stuffing is not helping. You just pay more and don't actually get much for it.”

“Already today, we have users who use not even very large models, you know, maybe a few billion parameters, and the vector database next to the model contains trillions of parameters. And they get, you know, much better performance that way.”

“We didn't, by the way, foresee any of this ChatGPT thing happening. We knew it was, it would keep growing, but that, I think, completely took everybody by surprise, including us.”

“One of my favorite applications is somebody built a face detection for cows”

“In fact, I was looking at experiments today from one of our teams, and we literally measure reduction in hallucination as one of the core metrics that, that, that we try to drive.”

“I know for a fact they're looking at it. I know for a fact they will have something.”

“And actually for prototyping, it's great to use that. Now get that into production at scale. And that's when you're going to start running into hiccups in terms of scalability cost wise, scalability performance wise on the ingest side, and then on the query ru…”