Langflow, Cassandra: IBM’s DataStax Buy

The Greek Goddess Cassandra was known for foretelling the future, and now IBM is planning that its namesake open source database system and its associated tools will help its enterprise customers do the same — with AI.

On Tuesday, IBM announced plans to acquire NoSQL database provider DataStax. Terms of the deal were not disclosed.

IBM will fold DataStax technology, including the company’s flagship Cassandra open source NoSQL database system and its popular Langflow low code AI tool, into its own watsonx portfolio of generative AI products.

Founded in 2010, DataStax was built on Cassandra, though also created AstraDB cloud database service, Langflow, and DataStax Enterprise, a commercial version of Cassandra. The company is headquartered in Santa Clara, California.

Biggest Challenge for Generative AI

The biggest challenge businesses face today when spinning up their own generative AI applications is, in part, difficulties in managing data. IBM asserts that Cassandra’s vector capabilities and Langflow’s low-code development environment should both help in these regards.

Many of Cassandra’s current users — including companies in the software, retail, finance, and ecommerce sectors — are starting to use the database for AI duties as well.

DataStax’s hundreds of enterprise customers include FedEx, Capital One, The Home Depot and Verizon.

“As enterprises adopt AI, they often struggle to leverage and deliver the knowledge locked in their enterprise data estate to power their AI agents and large language models,” wrote DataStax Chairman and CEO Chet Kapoor in a blog post about the acquisition. “We have long said that there is no AI without data, and this vision will now be propelled as DataStax joins IBM.”

Datastax is not the only recent AI-related acquisition: MongoDB also purchased VoyageAI. For Subbu Iyer, CEO of fellow database systems provider Aerospike, these acquisitions point to how “the AI gold rush for data continues at a fevered pace.”

“For AI to realize its potential, organizations need massive amounts of real-time data to provide timely, accurate and meaningful results,” Iver wrote by e-mail.

DataStax’s AI Chops

Cassandra has built-in vector processing and RAG capabilities, both essential for LLM-based generative AI applications. Cassandra is open source under an Apache license, and has over 9,000 stars on GitHub.

The Python-based Langflow, with a vendor-agnostic API, can be used to build RAG and multi-agent AI applications. It will add flexible middleware capabilities to the IBM watsonx.ai, an end-to-end AI development studio, according to IBM.

While Cassandra was DataStax’s flagship offering, Langflow has proven to be quite a success too, attracting thousands of users and earning over 49,000 stars on GitHub.

“Paired with Astra DB — the near-zero latency database with powerful vector and knowledge graph capabilities — and IBM’s watsonx.data, LangFlow provides one of the fastest routes to building enterprise-ready generative AI applications,” wrote Ritika Gunnar
IBM general manager for data and AI in a blog post announcing the acquisition.

👁 Langflow workflow diagram

The watsonx AI platform already harnesses other open source technologies as Iceberg, Spark, Velox and Presto, and Meta’s Velox data execution engine.

IBM has made good use of open source-based companies in the past decade. In 2019, it acquired Red Hat, which offers a Linux and Kubernetes-based open source platform for the enterprise. And last year, IBM set off to acquire HashiCorp and its stack of cloud native automation tools. That deal is expected to close shortly.

DataStax also heavily participated in the Apache Pulsar and OpenSearch open source communities, which IBM plans to continue supporting.

The DataStax acquisition is expected to close in the second quarter of 2025, subject to closing conditions and regulatory approvals.