Sponsored Content by Datastax

Apache Cassandra: The Secret Weapon for AI

The quality of artificial intelligence is closely tied to the data used to power it. Learn how four companies build on the Apache Cassandra open source database to deliver big AI value to their customers.


By Bryan Kirschner, Vice President, Strategy, DataStax

There’s a lot required of companies that build a business around artificial intelligence (AI). While it’s getting easier every day, they often rely on a variety of development frameworks and tools, properly trained machine learning models, significant compute resources, talented data scientists and application developers.

But there’s one very common denominator, without which AI can’t succeed: Data. Lots of it.

And all this data must be underpinned by a database that can handle the scale, performance, security and unique requirements of AI, including large learning models (LLMs) like GPT-4.

When you think of an AI success story, Netflix should be one of the first companies that comes to mind. Its ability to harness data at scale, in real time, enables the streaming content leader to predict, with often jarringly good accuracy, what titles its hundreds of millions of users might want to watch next. Other AI leaders today include Uber, Apple, and FedEx — each having built the ability to incorporate real-time AI into their businesses to better engage and delight their customers.

While these early AI leaders may be from different industries with different business models – they do have one thing in common: they rely on the open source Apache Cassandra® database.

Cassandra has long been known for its ability to handle data at massive scale, with high performance, speed, and security – and so it’s no surprise that it’s become a standard choice for AI. And recently, vector search capabilities were introduced into Cassandra so it can handle large learning models (LLMs) and generative AI workloads even better.

Fortunately, the secret of Cassandra to drive AI is not relegated to Fortune 100 companies with sizable resources.  Here we’ll look at how four companies—SupPlant, Bud Financial, Uniphore, and Alpha Ori—have built AI-driven offerings that accomplish everything from helping farmers improve crop yields to eliminating tons of pollution from maritime operations, all on the foundation of DataStax’s Astra DB: the easy-to use, database-as-a-service built on Cassandra.

AI to help farmers understand the language of crops

If plants spoke a language, SupPlant’s mission is to function as a translator for farmers.

The Afula, Israel-based company has been getting a lot of attention (it made Time’s “Best Inventions” list) with its real-time AI platform that helps optimize crop outputs. 

Using a stream of real-time data pouring in from sensors that monitor plants’ trunks, stems, and fruit; soil moisture and salinity; and temperature and weather, the company uses AI to translate this information, combined with crop growth models and weather forecasts, into irrigation recommendations and other insights about a farmer’s crops.

Data is a primary ingredient in SupPlant’s success, says CTO Revital Kremer.

“Real-time data is the key to unlocking the potential of precision agriculture,” she says. This requires a high-throughput, reliable data store to manage all SupPlant’s sources, including time-series IoT data, satellite imagery, weather data, soil characteristics, and more.” 

SupPlant relies on the Cassandra-based Astra DB as its primary database solution. SupPlant’s Astra DB environment incorporates 1.5 billion data points collected via its proprietary sensors over 2,000 crop seasons starting in 2016. Farmers using SupPlant’s solutions have reduced water usage by ~30% and enhanced crop yields by 5-10%, the company says.

AI for a competitive edge in banking

Inroads by nimble, technologically advanced fintechs has been a recurring theme in the financial services arena, with larger institutions hampered by regulatory constraints and inertia. Open banking, with all its newly available up customer data, only accelerated the wave of disruption.

Bud Financial is a disruptor, but its transactional AI offering is also a key competitive weapon available to larger institutions looking for a way to quickly understand customers’ creditworthiness and make highly relevant recommendations. 

Bud has developed a platform that uses a broad range of real-time data — often from non-traditional sources far beyond traditional credit bureau data—to gain deep insights into borrower income, spending, and creditworthiness. 

This, in turn, helps financial institutions make data-driven decisions to drive highly personalized and engaging experiences for customers and simplify lending and onboarding processes, while at the same time providing opportunities for banks to find efficiencies and reduce costs.

The real-time streams of transactional data and massive volumes of newly available open banking data that Bud employs requires a highly dependable, fast, and scalable data architecture; Bud relies on Cassandra, via Astra DB, for its reliability and capacity to handle high write volumes.

AI to help sales reps “read the room”

Gauging prospects’ interest in buying is always a challenge for enterprise sales teams. The pandemic and the consequent move to virtual meetings has only made it harder. 

AI to the rescue. Silicon Valley-based Uniphore employs AI to essentially “read the room” during virtual sales conversations, capturing a range of visual and audio clues that can easily escape human detection. 

It works like this: during an online video call with a sales prospect, Uniphore’s AI platform captures and processes 200 data points on the participant’s face at a rate of 24 frames per second, monitoring changing emotion and engagement levels through a variety of metrics, including voice tone, frown lines, and head position. 

It also incorporates natural language processing to assess word choices. Multiple AI models, built with Tensorflow and ONNX  and running on Nvidia Triton AI-inference servers, stream data via the open-source Apache Kafka streaming platform to its foundational datastore: Astra DB. The data is then read in real-time from Astra and processed into a sentiment analysis—all in real time—for sales teams. Uniphore’s platform makes the information available as a real-time sentiment analysis during a call, or as a detailed post-call analysis.

Sales reps get a data-driven approach to understanding a customer’s interest in certain value propositions versus others, when they are successful in satisfying a customer’s objections, or what a customer’s reaction is to a particular slide in their deck. By quickly helping to connect all the dots on what is or isn’t resonating with customers, representatives can build stronger rapport and close more deals.

AI for smoother sailing

With all the modern advances in ecommerce and next-day deliveries, 90% of world trade still requires the use of the most ancient of long-distance transportation methods: really big boats. The maritime transport industry is a key part of world economies, but it’s a resource-intensive way of moving goods, with a lot of opportunities for efficiency improvement.

Singapore-based Alpha Ori addressed this opportunity by developing an AI-based platform that, via onboard servers, collects more than 5,000 data points every 30 seconds from a broad variety of navigation, cargo, and engine control systems on a vessel. The data includes everything from ship speed and fuel consumption, to propeller shaft torque measurements, to wind speed.

Through the use of ML models and AI, the Alpha Ori platform then makes recommendations to improve vessel performance, increase operational efficiency, and reduce carbon emissions to meet sustainability imperatives.

The results are impressive. The 200 ships that were using its platform in 2021 reduced fuel costs by an aggregate $5 million and reduced carbon emissions by 30,000 metric tons, Alpha Ori says.   

The company built its offerings on Astra DB to reduce the operational burden of managing an enterprise grade datastore and to ensure the reliability of its data architecture. 

Astra DB for real-time predictive and generative AI 

The DataStax Astra DB managed cloud service built on the massively scalable, highly performant Cassandra database is the ideal data engine for a broad range of real-time AI initiatives – from predictive AI use cases like ecommerce recommendations to game-changing autonomous agents powered by generative AI and LLMs. Astra DB is the only database service today that harnesses the power of Cassandra in a simple, affordable, pay-as-you-go cloud service that is fully AI-ready.

Learn more about how DataStax enables real-time AI.