Powering Production ML: Why Wing Invested in Pinecone

Peter Wagner
No items found.
Few people on the planet understand the operationalization of ML as well as Pinecone’s founder and CEO Edo Liberty.

Everything we do at Wing aims to help realize the concept of the Modern Enterprise: an agile workplace, built on data and powered by AI. Essential elements of this concept are Cloud Data Platforms (such as Snowflake) and AI-powered applications (such as Gong). Their intersection – the Cloud Data Platform that is purpose-built for AI-powered applications – is the next critical step. Pinecone was formed to provide this capability, and Wing is excited to lead the company’s $10M Seed financing.

The world abounds with databases and it is reasonable to ask why it needs another. The answer lies in the distinctive requirements of AI-powered applications, which use model-generated vectors to represent data such as documents, videos, and user behaviors. Accurately filtering and ranking large sets of such high-dimensional vectors in real time requires a specialized data infrastructure. Existing data platforms can be pressed into service at limited scale, with modest performance. But they were designed for tables and documents, not vectors, and struggle to deliver low latency and high accuracy at scale. New workloads and their core data types have always been the catalysts for the creation of new data platforms. ML and its vectors are next in line.

Pinecone’s new ML / vector database supports dynamic transformation and indexing of billions of high-dimensional vectors. It answers queries like nearest neighbor and max-dot-product search accurately and in milliseconds. These and other must-have capabilities are required to take AI out of the lab and make it available for workhorse operational applications. ML and data infrastructure engineers will use Pinecone to support large scale production deployments of real-time ML-based applications such as personalization, semantic text search, image retrieval, data fusion, deduplication, recommendation, and anomaly detection. Looking ahead, it is hard to imagine many interesting applications that aren’t grounded in AI in some fundamental way. AI will be a pervasive property of modern software, as ubiquitous and important as oxygen.

Wing is fortunate to have had the opportunity to work with the Snowflake team since their Seed financing. One of the key things we learned to appreciate was the power of its business model. Snowflake made it shockingly easy to adopt cloud-based analytics. The result was an unleashing of pent-up demand, the scale of which we are probably still underestimating. Pinecone also offers a fully managed data platform service, providing ML and data teams faster time to market and effortless scaling without needing to think about infrastructure. ML-based applications like the ones Pinecone targets are more nascent than the relational analytics the early Snowflake focused on, so one should not expect the same avalanche of demand in the short term; but there are clear analogies between the value propositions.

Few people on the planet understand the operationalization of ML as well as Pinecone’s founder / CEO, Edo Liberty. Edo is himself an elite AI scientist, having led advanced research groups at Yahoo and Amazon. But unlike most AI researchers, he has also been responsible for standing up ML in production, including services like AWS SageMaker. The Pinecone team has used its cutting-edge experience to build the powerful back-end they would have loved to have had access to in their prior work. This is the kind of deeply authentic origin story we love to back at Wing. We look forward to working closely with Edo and the entire Pinecone team to enable the next wave of ML applications as AI takes center stage in the Modern Enterprise.

Wing Logo
Thanks for signing up!
Form error, try again.