Harnessing Data and AI for Business Value

Since helping to pioneer artificial intelligence (AI) technology in the 1950s, including coining the term “artificial intelligence,” AT&T has been shaping machine learning (ML) and AI technologies for decades. Today the company relies on its Chief Data Office (CDO) to set AT&T’s North Star for data, analytics, and AI excellence. The CDO, which is home to many of the world’s leading data scientists, developers, engineers, and researchers, empowers employees across AT&T to use data and AI to improve efficiencies and deliver better results for customers.

To accomplish its goal, CDO has a three-part mission:

• To harness, share and catalyze insights from the company’s massive data stores while transforming and modernizing AT&T’s data platforms, data supply chain and the data science ecosystem.
• To democratize data and AI across AT&T, stimulating wider-scale adoption of data-driven solutions by putting high-powered data analytics capabilities into the hands of every AT&T business leader and employee.
• To contribute back to and help lead and grow the broader data and AI community through open source and industry groups, academic engagement, and investment in its workforce.

Today, I’ll address that first mission, with future posts addressing those second two elements.

Harnessing Data and AI for Business Value

As one of the world’s leading modern communications and technology companies, AT&T carries more than 534.7 petabytes of data across its global network every day. To manage data at this scale, the CDO team has defined a common approach to how data is stored, managed, accessed, and shared across AT&T.  We established a “single version of truth” for each defined data product so people aren’t using different sets of data on the same projects and coming up with conflicting answers to the same questions. So we created a common data catalog for data findability, and implement data quality checks and security patterns across the data pipelines.  In addition, we established a data governance council that includes all core data user groups across the firm to get and stay aligned on this common approach to data.

This discipline enables our CDO team, hand in hand with our business partners across the firm, to harness this massive flow of data to help solve a diverse array of AT&T’s most technically challenging problems. Harnessing this collaboration with the business domain experts allows the CDO team to invent game changing AI solutions to improve the customer experience, increase internal efficiency, and identify business opportunities. Here are some examples of how data experts from CDO worked side-by-side with their business unit counterparts to create value for our customers:

• Avoiding network outages: Predictive models using the latest AI and statistical algorithms power the End-to-End Incident Management platform to scan more than 52 million different network records, devices, and customer circuits, and analyze over 1.2 trillion daily network alarms/alerts, to anticipate and avoid network service outages by detecting and exploiting patterns in these massive datasets – often in real-time.
• Blocking nuisance robocalls: A revolutionary network-level AI-based solution has blocked more than 6.5 billion robocalls. The solution uses sampling, predictive modeling, multivariate analysis and more to work 24/7, filtering through billions of daily records looking for patterns and suspicious qualities indicating likely robocallers. The solution then checks detected anomalies against internal and external rules and safeguards to avoid suspending legitimate automated calls.
• Preventing device fraud: A world-class, multi-stage AI-based fraud management tool evaluates millions of daily transactions across all AT&T sales channels in near real-time, inspecting each event within milliseconds against hundreds of rules to detect fraud patterns. CDO equipped the solution with an intuitive, flexible user interface to enable front-line members of the fraud team with minimal technical background to write, test, and deploy rules themselves, with little-to-no engineer involvement, to quickly address trends in the marketplace.
• Improving collections experience: An ecosystem of AI-based natural language applications, ML models and other technologies (including intelligent voice routing, Webforms, and Bots), power AT&T’s first-ever voice virtual assistant to handle roughly 400,000 delinquent payment calls per year. The fully automated self-service solution guides customers through a frictionless, step-by-step decision engine that negotiates flexible payment terms, capturing a higher percentage of delinquent account balances than ever before.
• Enabling climate risk planning: The telecom industry’s first hyperlocal climate risk visualization and planning solution predicts potential risks and impacts of environmental events on company infrastructure – up to 30 years in the future. CDO uses sophisticated ML and AI algorithms to analyze and predictively map millions of meteorological data points against hundreds of schematic layers tracking AT&T’s assets.

These examples show how a modern and connected approach to data lays the foundation for the transformative impact of innovative AI solutions to help solve challenging business problems.  We are seeing a tremendously positive impact on the business, and we have more to do on our data and AI journey ahead.

UP NEXT: Democratizing Data and AI