Confluent wants to deliver fresh, real-time data streams for AI

Data streaming major Confluent today announced a new “Data Streaming for AI” initiative to provide native integrations with leading players in the AI and vector database space, allowing its customers to tap the freshest contextual data from across their businesses.

Such realtime data is vital for use cases where speed is of the essence in business processes, such as fraud detection.

The first partners to have joined the initiative are MongoDB, Pinecone, Rockset, Weaviate and Zilliz. Confluent says this is just the beginning of their effort to help any business take advantage of real-time AI and more team-ups and innovations will be announced in the future.

“Continuously enriched, trustworthy data streams are key to building next-gen AI applications that are accurate and have the rich, real-time context modern use cases demand,” said Jay Kreps, CEO and co-founder of Confluent, in a press release. “We want to make it easier for every company to build powerful AI applications and are leveraging our expansive ecosystem of partners and data streaming expertise to help achieve that.”

The Mountain View, California-based company also announced a new generative AI assistant for its platform.

Role of real-time data in AI development

While the AI ecosystem has accelerated at an unprecedented rate over the past few years, many organizations are still relying on historical data — integrated with slow, batch-based point-to-point pipelines — to train their models.

This approach is typically suited for predictive AI, but when it comes to real-time AI use cases such as fraud detection or personalized recommendations, batch-based data doesn’t cut it. Its stale and low-fidelity nature affects the model’s ability to respond just in time with the most accurate, relevant and helpful information.

To fix this particular roadblock in AI innovation, Confluent is teaming up with multiple ecosystem players under its “Data Streaming for AI” initiative.

As part of this, the company explains, Confluent Cloud’s fully managed contextual data streams can be accessed directly within vector databases of MongoDB, Pinecone, Rockset, Weaviate and Zilliz, making it easier to use real-time data from different sources for AI-powered applications.

The Confluent platform acts as the shared source of real-time truth for all operational and analytical data, no matter where it lives, while the partner platforms help mobilize that fresh, contextual data for next-gen AI apps.

“Through our partnerships, we offer pre-built connectors and native integrations to popular data sources and AI technologies like vector databases so developers can effectively incorporate real-time data with MLOps pipelines, data augmentation workflows and generative AI inference chains,” Andrew Sellers, head of technology strategy at Confluent, told VentureBeat.

Building upon strategic partnerships

In addition to the vector databases, Confluent is building on its strategic partnership agreements with Google Cloud and Microsoft Azure to develop integrations, proof of concepts (POCs) and go-to-market efforts around AI.

For instance, it plans to leverage Google Cloud’s generative AI capabilities to improve business insights and operational efficiencies for retail and financial services customers. Meanwhile, Microsoft’s Azure Open AI service and Azure Data Platform will be used to create a Copilot template that will enable AI assistants to perform business transactions and provide real-time updates, benefiting industries such as airlines and transportation.

Notably, Confluent has also built some production-ready architectures around specific business outcomes in partnership with service partners Allata and iLink. These offerings speed up the whole process of developing, testing deploying and tuning AI applications, it said. 

According to Sellers, as of now, multiple enterprises are leveraging Confluent’s platform for real-time AI, including Spain-based bank EVO Banco.

“We helped (EVO Banco) build an advanced fraud detection system that combines real-time monitoring, advanced authentication methods, and real-time analytics. Through our platform, we receive data from multiple sources, including ATMs, online payments and mobile banking and immediately send it to be processed so transactions can be analyzed for fraudulent activity using a machine learning model trained with historical data,” he explained.

The technology strategy head noted that this is just the start. The company plans to expand the connectivity of its streaming platform with more partnerships through its Connect with Confluent program and will have more to share in the future. These engagements will empower more enterprises to accelerate their AI projects.

AI within Confluent

To make usage easier for its customers, Confluent also announced it is adding an AI assistant to its platform. The chatbot will work alongside developers, providing them with required information or automating certain tasks for them. 

For instance, a user could prompt the assistant in natural language to tell about their most expensive environment last month or ask for an API request to produce messages to their orders topic. The answers or the code generated will all be specific to the users’ deployment, Confluent said.

The bot will be available to Confluent Cloud customers in 2024.