This has been an exciting year for vector search and vector databases. Embedding models got better and more accessible, the use cases for vector search expanded, and new applications powered by vector search generated a lot of buzz. While a lot happened over the past twelve months, we wanted to share a few highlights of our own.

In 2022…

Demand grew, and so did we

We launched Pinecone back in 2021 – introducing the vector database as a way to make it easy for developers to build high-performance vector search applications. The adoption of Pinecone exceeded even our own ambitious expectations.

That’s why at the beginning of 2022, we needed to make some changes to keep up with our quickly growing user base. And we did just that - starting with a $28M Series A funding announcement in March. Since then, we’ve continued to observe the adoption of vector search in various applications and across new use cases. Pinecone is now being used by customers like Workday, Mem, Clubhouse, BambooHR, Expel, and many others for things like semantic search, anomaly detection, recommendation systems, mutli-modal search, fraud detection, and more.

To support all this growth, we also needed to grow the team. At the start of 2022, we were a company of 21 employees. Today, we more than doubled that number with employees (especially in engineering) spanning 25 cities and 6 countries and quickly growing hubs in New York City and Tel Aviv. We celebrated this amazing growth in person during our offsite in Greece!

Research advancements kept us on our toes

This year was also focused a lot on learning - whether that be learning more about new use cases and the capabilities of vector search, or for creating learning opportunities for our users and the community.

We created many new learning resources around the latest technologies, including:

• Learning how to train models in low-resource scenarios with GPL.
• Exploring multi-modal ML with Open AI’s CLIP and in the Embedding Methods for Image Search series.
• Searching through video and audio with Whisper.
• Making Stable Diffusion more accessible with “vector caching”.
• Collaborating with Deepset AI to release the HaystackDocumentStore for Haystack users to build better, faster, and feature-rich semantic search tools.
• Helping users get started with semantic search in the NLP for Semantic Search series.

In 2022, we attended and sponsored several conferences, including Southern Data Science Conference, SIGIR, and the AI Summit. We also hosted many webinars with guest speakers from Pinecone partners and supporters like Yury Malkov (advisor to Pinecone), Nils Reimers, and Julien Simon. Subscribe to our mailing list and join our meetup group for the latest news and upcoming events.

Our users inspired us to keep innovating

Since launching Pinecone, we’ve worked hard to innovate and add features for our users. We had many product launches this year to help our users build bigger, better, and faster. Some highlights include:

• Hybrid vector index for keyword-aware semantic search
• 10x faster searches with proprietary graph-based vector indexes
• Partial and real-time index updates
• Combined vector search with single-stage metadata filtering
• Zero-downtime vertical scaling

Sign-up for free, and experience the ease of building high-performance vector search applications with Pinecone.

What’s next?

Overall, it’s been a year full of growth, learning, coming together, and pushing the boundaries of vector search. We are very proud of the company and culture we’ve built, and can’t wait to continue growing, learning, and improving in 2023. Want to join us and continue to shape the future of AI and ML applications? We’re hiring across all functions!