Disrupt or be disrupted. That phrase describes successful companies in today’s highly competitive and brutally fast-paced marketplace. Intel IT is leading the digital transformation of Intel’s business processes by enabling data-driven decision making with predictive and advanced analytics services as well as delivering insights and solutions to improve how Intel designs, manufactures, sells, and delivers its products.
In a nutshell, to maintain a competitive edge, Intel must make more products much faster with fewer people, and sell those products to a larger number of customers. I’m leading a team of 130 engineers that, through a competency center for big data and machine learning, is helping Intel achieve those goals through analytics. Our solutions are highly automated. We’re embedding machine-learning algorithms in production processes to automate and improve complex tasks that are helping increase sales, reduce product time to market, and improve product quality while reducing costs.
Advanced Analytics as a Disruptor
We define a “disruptor” as the “ability of an algorithm to perform a cognitive task at a sufficient level to free up humans to perform a higher-value task that machines cannot perform.” In that sense, advanced analytics is the ultimate disruptor. Building on nine years of investment in advanced analytics, we are now working toward this goal, although the journey is by no means finished. We already have developed analytics solutions that can perform the following tasks:
- Process huge amounts of unstructured marketing data in mere hours
- Comb through billions of manufacturing sensor data points to help increase yield and tool availability
- Increase the accuracy of silicon validation processes
These are just a few examples of how advanced analytics can positively disrupt the business and increase business velocity.
By nature, analytics solutions are scalable. Once we’ve created an algorithm that can perform a task, it can be duplicated indefinitely without additional costs. As we build more and more advanced analytics solutions, they have the power to optimize the “collective brain” of the entire enterprise. We will be able to start connecting the dots across the various aspects of the business, coordinating thousands of algorithms to better inform decision making and further increase the pace of innovation.
Anatomy of Intel’s Advanced Analytics Team
Several years ago, as it became obvious that analytics was going to be a strategic competency for Intel, we made a conscious decision to coalesce Intel IT’s expertise in advanced analytics into a big data and machine-learning competency center, as opposed to spreading these talented people throughout the organization. The expertise hub we have created represents an incredible depth of knowledge—our data scientists possess not only skills in big data analysis and machine learning, but also in product development and business processes. This rich background enables our team to understand business needs and quickly develop analytics solutions that address those needs.
We run the team like an innovative company, enabling us to adapt to rapidly changing business needs and priorities. At the top of the organization is a lean office that combines portfolio management, business development, and research and development. This office oversees the work of six “startups” (Design, Manufacturing, Sales, Internet of Things (IoT), Healthcare, and Deep Learning) and can help quickly grow new startups to address new business needs or a new analytics domain.
Using Advanced Analytics to Improve How Intel’s Business Runs and Grows
We are using advanced analytics in two fundamental ways at Intel.
First, we are disrupting pivotal internal processes, replacing the old way of doing business with a new, far better way. If you consider how ride-sharing has disrupted the legacy approach to hiring a taxi, you can get a glimpse of how advanced analytics can optimize a business process to such a degree that the old way of doing things becomes irrelevant. We are focusing on the most mission-critical processes that can provide the most competitive advantage. For example, we are using analytics to make our Assembly, Test, and Manufacturing facilities smarter; re-engineer our business-to-business (B2B) customer experience, and digitize our supply chain. Applying such solutions at scale will provide Intel with the business agility and velocity it needs to succeed in the modern marketplace.
Second, we’re taking an active part in the work of Intel’s business units and other relevant divisions such as product design and software development, helping them build artificial intelligence (AI) products. We provide the subject-matter expertise to help them build relevant hardware and software.
For example, our analytics solutions are helping Intel’s design teams develop hardware faster. The design teams want to build hardware for tomorrow’s machine-learning workloads, and the analytics benchmarks we’ve developed help the design teams evaluate different hardware alternatives. We also engage with Intel’s customers to help them maximize the benefit from their Intel hardware and machine-learning solutions. We have created tools that help customers implement machine learning in their environment, such as the Intel® Deep Learning SDK. We also provide reference architectures and implementation guidance.
Recipe for Advanced Analytics Success
While Intel IT has been working with advanced analytics since 2008, we remember what it was like when we were just getting started. And, when we talk with Intel customers, many of them are just embarking on their journey with big data, machine learning, and analytics.
We got where we are today by following a three-step process:
- Getting our feet wet. For the first two years, we chose opportunistic, “low-hanging” projects that could quickly prove the value of big data, machine learning, and analytics without a lot of up-front investment. During those two years, our mantra was “5/6/10.” That is, take five people highly skilled in analytics, give them six months, and they could produce USD 10 million or more in value. All too often, we see companies put a lot of investment into the platform and then figure out what to do with it. In contrast, at this stage we didn’t spend a single dime on an analytics platform—our focus was purely on conveying the message that analytics could create value.
- Growing at scale. For the next two to three years, we built a roadmap, established themes for our analytics projects, and built an analytics platform tightly coupled to business need. This approach kept us from spreading our resources too thin and enabled us to increase our impact by an order of magnitude—each project created at least USD 100 million in value. We established three focus areas and themes:
- For Design, we focused on reducing time to market.
- For Manufacturing, we focused on reducing cost while improving product quality.
- For Sales, we focused on increasing sales by optimizing the sales efforts through personalization.
After a certain period of time, processes and solutions become scalable. For example, our analytics R&D office can now easily launch one or two new startups each year.
- Blazing a path for disruption. Building on our successes over the past decade, we are now working on advanced analytics projects that can create a billion dollars of value through high-end machine learning and a clear line of sight to AI. We are now focusing on: a) building algorithms that can perform tasks currently only performed by humans, and b) optimizing existing business processes to the highest possible degree of efficiency through real-time, complex machine-learning algorithms.
Sharing Our Advanced Analytics Story
As described in our recently published 2016-2017 Intel IT Annual Performance Report, “Accelerating the Pace of Business through IT Innovation,” Intel IT is using advanced analytics to transform Intel and create business value. We expect to continue to leverage predictive analytics to empower Intel’s business units and to increase Intel IT’s agility.