Vector database vendor Pinecone eyes future under new CEO

Pinecone, a vector database specialist, is entering a new phase.

Vectors are numerical representations of data that make it discoverable through similarity and keyword searches. Because vectors enable users to find relevant data amid potentially billions of other data points, including unstructured data such as text and images, vector databases have become a key part of AI development pipelines in recent years.

In addition to New York City-based Pinecone, niche vector database vendors such as Milvus and Chroma. Broader data platform vendors, including AWS, Google Cloud, Oracle and Databricks, also provide vector database capabilities.

On Sept. 8, Pinecone named Ash Ashutosh, a longtime technology industry veteran and entrepreneur who most recently served as Google's global director of solution sales, as its new CEO. Edo Liberty, who founded Pinecone in 2019 and had been its CEO, will now concentrate solely on developing the vendor's technology as its chief scientist.

Ashutosh's appointment represents an inflection point for Pinecone. Like many niche data management vendors that establish the value and validity of their technology, Pinecone has different potential paths.

Some data management startups select acquisition as their end goal, seeking a larger company to buy their technology and integrate with a larger ecosystem. Others look to grow by expanding their platforms beyond their initial specialty, becoming broader-based data platforms that serve multiple layers of the data stack. And still others choose to remain specialists, trying to stand out from larger competitors by touting more sophisticated capabilities than those provided by vendors with broad data platform offerings.

Ashutosh founded Serano Systems in 1997, AppIQ in 2001 and Actifio in 2009. Eventually, he led all three through acquisitions. However, despite a history of finding buyers for data management vendors, an acquisition is not the goal for Pinecone, according to Ashutosh.

In a recent interview, Ashutosh discussed his vision for Pinecone, including why he believes vector databases are so critical in modern data and AI stacks. In addition, he spoke about what attracted him to Pinecone, as well as the vendor's strengths and weaknesses.

Editor's note: This Q&A has been edited for clarity and conciseness.

What attracted you to Pinecone to become its CEO?

Ash Ashutosh: It's very simple. There's an AI market that is not just exploding from a mind-share perspective but is moving from what's possible to what is practical. This means there is a perfect opportunity for someone to build a real business around infrastructure that's at the core of AI. Pinecone has founder Edo Liberty, who is a brilliant scientist and invented this [vector database] space and continues to innovate in the next phase. Meanwhile, there needs to be a second part that needs to be kicked in, which is to drive the business, go help many of the organizations that are trying to change the way they do business and change their business workflows.

When Edo first [went public with Pinecone], that was in the first wave of creating the vector database category. Now, it's about scaling it to meet enterprise needs, building the critical infrastructure for AI that is accurate and reliable, and being a true partner in helping businesses. That's what I've been doing for more than 30 years.

What are your goals for Pinecone as you take over as CEO?

Ashutosh: It's very simple. There are large language models (LLMs) that are very intelligent machines, and what these intelligent machines need are memory and context so they can become knowledgeable machines. That is what we provide. We want to be the company that provides the component that makes very smart machines very knowledgeable machines. AI guesses, and we provide the knowledge that makes AI accurate, safe and powerful. If you look not just at the vision part but at the execution part of that journey, we started with a vector database, we simplified [application development] with Pinecone Assistant, and we plan to keep moving up the stack to add things like multimodal context and integrations to become the foundation of agentic workflows.

At the end of the day, our vision is to make AI knowledgeable.

Who do view as Pinecone's main competitors and how do you plan to guide Pinecone so it can differentiate from those competitors?

Ashutosh: There are a lot of vendors in this market because there are so many enterprises trying to figure out their AI journey. Each vendor supports different parts of enterprises' journey.

We are a pure play vector database designed from the ground up to become an enterprise platform for AI, and we focus on building our business around large enterprises that are building production applications. At the same time, we've been developer-first and AI-first, and we've been educating the entire market. We will continue to do that. There are two dimensions of our business. One is to continue to innovate, continue to educate and allow enterprises that are experimenting with this new technology to get familiar. At the same time, we will continue building the business around things that we are starting to see customers build their businesses around.

With many broader data management platforms now providing vector database capabilities, can a specialized vendor such as Pinecone continue its relatively narrow focus, or do you have to broaden your capabilities to continue to grow?

Ashutosh: I've thought about this quite a bit. I've been in the storage and data business for almost 35 years, longer than most people, and I've seen an evolution. There are times to experiment using whatever you have as a matter of convenience, when you don't need serious best-of-breed capabilities. But we absolutely believe that AI is a different workload. AI data patterns, how people think about data, how people secure data, the basic structure of underlying data, is so different. If you think about a vendor adding vector database capabilities onto an existing platform, it's like extending data in a format built for human consumption to a vector database built for AI consumption. As more data becomes agentic, you need a purpose-built AI foundation, a market that is focused on serving AI. We are exclusively focused on that market, focusing on innovations to make AI knowledgeable.

Will there be add-ons? We'll see. But we are absolutely focused on purpose-built storage and memory for AI workloads.

Within being a platform for AI workloads, is there an opportunity to expand beyond being a specialized vector database?

Ashutosh: There will be. As more workloads become agentic, there will be a new kind of data that goes with it. We think about vectors as a language for AI. That language will continue to be enhanced, and we will continue to enhance around it. But the idea is to continue to focus on being a foundation where LLMs and agents can communicate and think and change the way business is done.

If you think about the early days of HTML, they were simple pages. But then there were additions as the entire market around it became much bigger. The language itself has changed. We expect the same thing for vectors as a basic construct.

Bringing an outside perspective to your new role as CEO, what do you see as Pinecone's strengths?

Ashutosh: The team has lived this problem. It didn't think that there is an interesting thing that someone else built and then try to do it faster or cheaper. Instead, the team lived the problem and was effectively customer zero. When they didn't find [a purpose-built vector database], they went out and built it. Anybody who builds for themselves fundamentally has a different perspective on what it means to be a user. Every aspect of what this company has done is around facilitating developers -- having an API-first approach, making it very easy to use and spending a ton of investment educating others around this ecosystem. I think the brand is amazing. But the best part is the team.

Most vendors have engineering organizations, but we decided to create a research organization. I'm coming from an organization that respects that enormously. I had an opportunity to do many things in the next phase of my life, and it is very hard to find a unique company like this.

Conversely, what do you view as something that needs to improve?

Ashutosh: We need to get better awareness of why [purpose-built is better than] just using a platform's add-on -- something that is in a customer's cloud platform -- and why they shouldn't take an open source tool and just use that. We need to drive awareness that we now are going to focus everything on helping businesses build enterprise AI applications for [production workloads].

Secondly, internally, we are evolving from an organization that was ideating this entire space to now building a business around this entire space. That's an internal change that we have to make to align with how the market is changing and what our customers are expecting.

Before you came to Pinecone you guided three companies through being acquired, so is that the long-term plan for Pinecone?

Ashutosh: It's definitely not on the table right now.

Each of the companies that I started … were trying to solve problems that I lived through. I had a bunch of partners and customers who worked with us to help us figure out what the problems meant to them and how they impacted their business. That's what you can control. I was also part of a company that we took public, though I wasn't the founder. What happens, the outcome, is impossible to know. You can start out with a very clear view of what problem to solve, and control that. That's all you can focus on.

As far as fueling product development and other expansion, is Pinecone relying on venture capital?

Ashutosh: Yes, but we are also engaging with several strategic partners. Most of the vendors in the AI market are starting to see Pinecone as a leader that they want to partner with. There will be strategic partnerships, and we will fuel and finance Pinecone in different ways.

What are you looking forward to as you take over as CEO?

Ashutosh: I spent the last few weeks looking at Pinecone and looking at this space. I've never seen so many vendors in a space come up as fast, which is absolute proof that there is more than hype around AI. As more people enter this market and explore what is possible, there are plenty of people that are quietly building businesses to solve problems. I want us to be thinking about what people are doing in real businesses, and how they are changing business. It'll be nice to start observing those rather than a lot of the hype, things that are sensational. It's nice to look at what people are really doing. It's an amazing time.

Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.