Webinar: How to Build a Streaming Data Mesh | Register Now

Confluent Integrates with Pinecone Serverless to Make Real-Time, Cost-Effective GenAI a Reality

Get Started with Confluent Cloud

Apache Kafka® reinvented for the cloud — start free, launch fast

Start your free trial with Pinecone

Written By

Confluent is excited to closely partner with Pinecone as they unveil Pinecone serverless—a completely first-of-its-kind vector database architecture, enabling more developers to build real-time, highly performant GenAI products with up to 50x cost-savings. When Confluent Cloud—the cloud-native data streaming platform proven to lower the total cost of ownership for Apache Kafka® by up to 60% (Forrester TEI study)—is paired with Pinecone, engineering teams can focus their time on building intelligent, contextualized AI applications at any scale while avoiding the costly distractions of infrastructure management.

Now available for integration with Confluent, Pinecone serverless lets companies add practically unlimited domain-specific knowledge to their GenAI applications, with a hands-free and truly serverless developer experience. This provides a significantly easier and faster path to reliable, effective, and impactful GenAI applications for companies of any size and GenAI maturity.

Key innovations in the breakthrough architecture of Pinecone serverless include: 

  • Separation of reads, writes, and storage significantly reduces costs for all types and sizes of workloads.

  • Industry-first architecture with vector clustering on top of blob storage provides low-latency, always fresh vector search over practically unlimited data sizes at a low cost.

  • Industry-first indexing and retrieval algorithms built from scratch to enable fast and memory-efficient vector search from blob storage without sacrificing retrieval quality.

  • Multi-tenant compute layer provides a powerful and efficient retrieval for thousands of users, on demand. This enables a serverless experience in which developers don’t need to provision, manage, or even think about infrastructure, as well as usage-based billing that lets companies pay only for what they use.

Successfully deploying GenAI requires a constant supply of trusted, real-time data

Many organizations struggling to get started with AI encounter the same challenge: data. The 2023 Global Trends in AI report by S&P surveyed over 1,500 AI decision-makers and revealed that access to clean and trustworthy data is one of the biggest barriers to AI innovation. Connecting AI/ML models to enterprise data in real time has been one of the most challenging problems data-dependent teams have been trying to solve.

These models require fuel by way of high-quality, reliable, and fresh data from across the entire business in order to deliver the results that AI promises. Use cases including chatbots, virtual assistants, and other large language model (LLM) applications are expected to drive effective customer engagement on behalf of the business—but they can’t do this if they don’t know anything about the business or only have access to stale data.

To succeed in this new era of GenAI, a data stack built atop an enterprise-wide supply of real-time data streams will be required.

Confluent data streams are the foundation to the GenAI data stack

Realizing the full potential of GenAI is often hindered by outdated data integration methods. Batch-based, point-to-point pipelines throughout the enterprise result in a “bird’s nest,” monolithic system that has nothing to offer downstream applications but stale and inconsistent data. Not to mention, scaling applications built on these legacy pipelines is an operational nightmare and data governance controls are limited, at best.

This is a roadblock to AI innovation, impeding developer agility, hampering data reuse, and slowing the overall pace of advancement. The crux of the matter is clear: your AI strategy is intricately tied to your data strategy. Until the foundational challenges of real-time data infrastructure are addressed, your AI capabilities will be constrained, perpetually waiting for fresh, contextualized data that can bring a meaningful interaction to life.

Built by the original creators of Apache Kafka® and recently named a leader in The Forrester Wave™: Streaming Data Platforms, Q4 2023, Confluent provides a cloud-native and complete data streaming platform available everywhere it’s needed to fuel AI applications with an enterprise-wide supply of trusted, real-time data.

  • Cloud native: Spend more time building value when working with a Kafka service powered by the Kora Engine, including GBps+ elastic scaling, infinite storage, a 99.99% uptime SLA, highly predictable/low latency performance, and more.

  • Complete: Deploy new use cases quickly, securely, and reliably when working with a complete data streaming platform with 120+ connectors, built-in stream processing with serverless Apache Flink (preview), enterprise-grade security controls, the industry’s only fully managed governance suite for Kafka, pre-built monitoring options, and more.

  • Everywhere: Maintain deployment flexibility whether running in the cloud, across clouds, on-premises, or in a hybrid environment. Confluent is available wherever your applications reside with clusters that sync in real time across environments to create a globally consistent central nervous system of real-time data for the business.

Here are some of the reasons why businesses are putting Confluent’s data streaming platform at the center of their AI strategy:

  • Establish a dynamic, real-time knowledge repository for the business

    With Confluent’s immutable and event-driven architecture, enterprises can consolidate their operational and analytical data from disparate sources to construct a unified source of truth for all their data. This empowers your teams to excel in model building and training, driving unparalleled levels of sophistication and accuracy across a range of applications.

  • Integrate real-time context at query execution

    Native stream processing with Confluent’s serverless Flink offering enables organizations to transform and optimize the treatment of raw data, at the time of generation, and turn them into actionable insights using real-time enrichment, while also dynamically updating your vector databases, like Pinecone, to meet the specific needs of your GenAI applications.

  • Experiment, scale, and innovate with greater agility

    Confluent’s decoupled architecture eliminates point-to-point connections and communication bottlenecks, making it easy for one or more downstream consumers to read the most up-to-date version of exactly the data they want, when they want it. Decoupling your data science tools and production AI applications facilitates a more streamlined approach to testing and building processes, easing the path of innovation as new AI applications and models become accessible.

  • Craft governed, secure, and trustworthy AI data

    Equip your teams with transparent insights into data origin, flow, transformations, and utilization with robust data lineage, quality, and traceability measures included within our Stream Governance suite. This fosters a climate of trust and security essential for responsible AI deployment.

Data streaming for AI: 1) Connect to data sources across any environment, 2) Create AI-ready data streams with Apache Flink (preview), and 3) Share governed, trusted data with Pinecone to fuel downstream AI applications.

How developers benefit from Confluent’s Pinecone connector

Our Pinecone Sink Connector (preview) allows organizations to easily access their high-value data streams from within Pinecone to power any GenAI use case. The connector takes data from Confluent Cloud, converts it to vectors using embedding models, and then stores these embeddings within the Pinecone serverless vector database.

The following features are supported for this connector:

  • Streaming UPSERT operations from a list of topics into the specified Pinecone index and namespace.

  • Avro, JSON Schema, Protobuf, or JSON (schemaless) input data formats. Schema Registry must be enabled to use a Schema Registry-based format (for example, Avro, JSON_SR (JSON Schema), or Protobuf).

  • Skip vector embeddings generated using Azure OpenAI.

  • Multiple tasks. More tasks may improve performance.

  • At-least-once delivery semantics.

Please note that the Pinecone Sink Connector (preview) can only be run on Confluent Cloud in these AWS regions, supports Azure OpenAI only, does not support bytes format, and does not support schema contexts.

Once you download the Pinecone Sink Connector (preview), be sure to leverage the Quick Start Guide to configure your integration.

Start your free trials of Pinecone serverless and Confluent Cloud

Get started today with Confluent and Pinecone serverless, the fastest way to bring highly contextualized GenAI applications to market.

Not yet a Confluent customer? Start your free trial of Confluent Cloud. New sign-ups receive $400 to spend during their first 30 days. No credit card required.

Table of Contents
    • Greg Murphy is the Staff Product Marketing Manager focused on developing and evangelizing Confluent’s technology partner program. He helps customers better understand how Confluent’s data streaming platform fits within the larger partner ecosystem. Prior to Confluent, Greg held product marketing and product management roles at Salesforce and Google Cloud.

    Get Started with Confluent Cloud

    Apache Kafka® reinvented for the cloud — start free, launch fast

    Start your free trial with Pinecone

    Did you like this blog post? Share it now