We’re excited to announce that IBM® watsonx.data™ has recently launched an integrated vector database, based on the open source Milvus, in the data lakehouse. Now, IBM watsonx™ customers can unify, curate and prepare vectorized embeddings for their generative artificial intelligence (gen AI) applications at scale across their trusted, governed data. This enhances the relevance and precision of their AI workloads, including chatbots, personalized recommendation systems and image similarity search applications.

The emergence of gen AI has hastened businesses’ adoption of AI, with 80% of enterprises (link resides outside of ibm.com) either using or intending to use foundation models and embrace gen AI. Against this backdrop, organizations are elevating the importance of data as they rush to harness its potential to tailor and enhance the performance of gen AI models faster than their competitors.

However, data landscapes remain complex. The availability and quality of underlying data inherently determines the relevance and reliability of AI, and organizations are still dealing with fundamental data challenges to effectively scale AI.

82% of enterprises face data silos, inhibiting operations. Data volume is set to exponentially grow across diverse locations and formats, with varying quality. IDC predicts a 250% increase in stored data across on-premises and cloud storage by 2025, further increasing complexity.

IBM watsonx.data’s open data lakehouse architecture empowers organizations to access and unify data from multiple data sources without migrating or recataloging. This allows you to maximize value from existing data investments and connect enterprise data across clouds and on-premises environments for AI use, reducing duplication and streamlining operations like extract, transform and load. You can then share a single copy of data seamlessly across your organization through a single point of entry.

Trusted, governed data is essential for ensuring the accuracy and relevance of AI applications. One way to prepare data for AI is by creating vectorized embeddings for low-latency queries. This unlocks large volumes of enterprise data for gen AI and retrieval augmented generation (RAG) use cases at scale. According to Gartner (link resides outside of ibm.com), by 2026, more than 30% of enterprises will have adopted vector databases to ground their foundation models with relevant business data.

Watsonx.data’s embedded Milvus vector database enables you to store and query vectorized embeddings for RAG use cases. This helps ground AI applications in trusted data, enhancing the relevance and precision of your outputs.

Users can seamlessly connect to trusted data in watsonx.data from IBM® watsonx.ai™ or another AI tool. Grounded AI applications can also benefit from using smaller, fit-for-purpose models, which reduce inferencing latency and overall system costs.

Try watsonx.data yourself with a free trial. Or join our upcoming webinar (link resides outside of ibm.com) to learn how to vectorize your data for RAG at scale with watsonx.data.