Intel on Wednesday released its 2018-2019 IT performance report, giving an inside look at how the semiconductor business is transforming its IT operations to serve as a strategic part of the overall business. The report details the maturity of Intel's DevOps strategy and its ongoing efforts to scale Agile and DevOps practices. It also shows the progress Intel has made bringing machine learning into its operations.
A large part of driving change within IT involves creating the right culture, Intel's chief data officer Aziz Safa told ZDNet.
"Twenty years ago, we would not make a major change in the enterprise for years," he said. "Whereas today you are seeing new offerings come very fast. How do you incorporate these changes -- that can have a large impact on the productivity of the organizations you support, as well as the people who work for you -- and do it fast, without impacting your day-to-day production? That takes quite a bit of adjusting the thinking to create that agility you need."
Intel established its Corporate Data Office in early 2017 to serve as a "one-stop data shop" for the business. According to the report, the office has generated $1.25 billion in business value by using AI and advanced analytics in areas including sales, marketing, manufacturing and product design.
With the launch of the data office, Intel in 2017 set about to strengthen its IT basics in part with Agile and DevOps practices. Intel created a task force that trained more than 4,700 employees on Agile methodology and created more than 500 Agile Persistent Teams (APTs). At the end of 2017, the report says, 97 percent of these APTs met the criteria for Level-3 Agile maturity (Project Excellence).
In 2018, Intel's CDO focused on empowering APTs and adopted an industry-standard framework for scaling Agile practices to an enterprise/program level. They trained hundreds of employees on this framework and another 3,000 employees on Agile practices in general.
"Our new scaling framework has created a significant pull in the organization to fully commit to the Agile and DevOps transformation," the report says.
In 2019, the goal is to work on automating the development cycle with an initial on automating testing. Currently, about 22 percent of testing is automated. Intel wants to achieve 60 percent testing automation by the end of the year with the ultimate goal of a 50 percent reduction in time to delivery.
Faster delivery also requires the interoperability of data from different platforms, Safa said.
"We do have a lot of suites of applications what we interoperate with, and we want to make sure we do that very efficiently," he said. "In the enterprise historically, you see quite a bit of data silos. It's a problem when you're introducing new systems and applications."
To that end, Intel is working on channeling data puddles into relevant data lakes, where trusted, connected, and secure "gold copy" data is defined and integrated properly. The goal is to make this data available across all Intel lines of business and business processes.
Once this data is more accessible, it becomes easier to apply AI and machine learning to it. The performance report gives a few specific examples of how Intel has applied AI and advanced analytics already.
For instance, Intel has optimized its parts inventory with an intelligent, automated system that has reduced time to decision from six months to one week and increased savings by $58 million.
"We met the challenge of disparate stocking methods and systems with big data technologies and advanced analytical algorithms," the report says. "Our solution has provided seamless cross-functionality across organizations and suppliers, both internal and external. Forecast modeling consolidates demand signals and an automated forecasting process now measures and improves forecast accuracy."
In the area of product validation, according to the report, AI has resulted in 50 percent less debug duplication and 70 percent less regression testing. It's increased the number of issues identified by 50 percent at half the budget.
In the area of sales, Intel launched a Sales Assist pilot program to simplify account management and recommend actions for potential customers. Applying intelligent features in sales can be critical for a major business like Intel, Safa said, given just how many customers it interacts with on a daily basis.
The Sales Assist pilot program was deployed globally in 2018 and made a $46 million positive impact on sales that year.