The value of data fabric for AI projects

To help drive new artificial intelligence applications, one company converted a data gathering effort that could only be completed once every six months into one that happens daily. Another reduced the number of data sets needed by more than 90%, leading to better results, faster.

This is the power a data fabric can bring.

In the rush to implement artificial intelligence applications, many organizations try to piece together the data required for a given use case, resulting in numerous disjointed data stores. A data fabric presents a better option, using a platform approach that supports multiple AI use cases across different functions.

Doing so confirms that data is certified, with appropriate governance, definitions, quality, and traceability with proper ontology, says Anil Nagaraj, Principal, Cloud Engineering, Data Engineering and Analytics with PwC US.

Customers prove the power of Microsoft Fabric

A large manufacturer was looking to boost revenue from its maintenance business by using AI to help identify areas for profitability improvement. That involved tracking data from dozens of sources, from manuals and equipment inspections to usage reports, and more. Much of the data was unstructured, including lots of images.

It took the company six months to piece together enough data to feed the AI engine. That was clearly not sustainable, given the data would need to be constantly updated. “It needed to be industrialized,” Nagaraj says.

PwC did just that using Microsoft Fabric. “You bring in data once, create a data set, and certify it with Microsoft Purview,” Nagaraj says. Additionally, the new Microsoft Fabric IQ helps define relationships between the data sources, making the collective data more meaningful. “Predictions on profit margins become more consistent and precise.”

Importantly, the six-month effort the company put in did not go to waste. PwC used its own intelligent data modernizer to harvest that data and accompanying logic to help inform its AI models. Now, the system is automated, with constantly updated data feeds to drive real-time AI models.

The other example involves a large technology company with a 10-year-old data ecosystem supporting all its business operations, from sales and finance to product intelligence. The system encompassed some 5K+ data sets that had sprung up over time. Besides being unwieldy and slow, the sheer volume made updates challenging because the consequences were difficult to predict.

Here again, to get the data AI-ready, PwC used Microsoft Fabric to dramatically reduce the footprint of the data infrastructure, reducing the datasets down to several hundred. Now, all business functions benefit from real-time data, enabling more effective decision-making.

Better data yields answers faster

“In the age of AI, people don’t want to wait for yesterday’s business results to make decisions,” Nagaraj says. “If I launch a product, I want to know how it’s doing now, not wait until tomorrow.”

Toward that end, both companies now benefit from modern features such as AI-driven dashboards, query engines and real-time capabilities. Microsoft Power BI, for example, lets users ask complex questions in plain English.

Microsoft Fabric also means far less data proliferation with real-time data. Data is stored once, certified, and available in real time for reuse across any number of use cases—no need to duplicate data sets.

A sound architecture delivers business value

While a sound data foundation is crucial, achieving positive results from an AI project also requires proper architecture.

That’s where PwC can help, by architecting a solution that combines the right technologies with attention to data governance and access rights. That includes structuring workspaces for the needs of different business units and giving each employee the appropriate level of access.

PwC also brings experience with proven tech-enabled solutions and frameworks around topics such as data quality, ingestion, and propagation. The frameworks help companies build data foundations more quickly, resulting in faster AI deployment and time-to-value from AI applications.

But all the technology is secondary to what Nagaraj says is PwC’s primary goal: “We are laser-focused on delivering business value.”

Learn more about how PwC can help you employ AI to reshape what’s possible for your business. Visit PwC.com.