Digital Transformation & Maturity

AI for Everyone

Sponsored by: Microsoft


There’s a large amount of noise pertaining to AI and at times, it appears that there’s a lot more A than I.  At Microsoft, we strive to be a trusted advisor throughout this digital transformation.  For over 25 years, Microsoft has been preparing for this moment.  When Bill Gates created Microsoft Research, a refugee for the best and brightest to explore moon-shot ideas, he imagined creating technology that would enable computers to see, hear, talk and understand human beings.  At last, the time has come! Currently, there are over 7,000 researchers and developers focusing on AI and the progress the team has made is remarkable.  For example, Microsoft researchers developed AI technology which reached human parity for reading comprehension using the Standard Question Answering Dataset (SQuAD).  SQuAD applied a machine reading comprehension dataset comprised of questions from 100,000 Wikipedia articles.  The models surpassed human accuracy in answering the questions – a big win not only for Microsoft, but for the technology industry.  At Microsoft we want to make our AI capabilities available to everyone, so they can maximize their potential. Our mission is to democratize technology.

The telecommunications industry is a prime candidate to integrate AI across its businesses due to the substantial number of customers, the vast array of products offered, and the complex and overwhelming assortment of technical and business parameters. In the digital era, the telecom industry has shifted from basic phone and Internet service to a sector that is going high-tech and constantly evolving into digital agents, mobile, wearable and IoT/automated environment. Telcos’ AI requirements are complex and unique to each entity. AI products and services are not a “one size fits all” solution and currently, an entity cannot usher out an AI solution for many business customers at once. Therefore, here at Microsoft, we’re proactive, creative, and bring a unique outside-in perspective to empower our partners.

Microsoft is a Trusted Partner

Although service providers recognize that AI is unchartered territory– they understand if they’re not at the table, they’re on the menu.  Connectivity is at the very core of AI and networks continue to exponentially grow in scale and complexity – particularly around the advancements in IoT. Many telco customers are dissatisfied with their service because humans cannot manage this network expansion and the never ending corporate consolidations (buying up and bundling more and more product lines) – which adds additional complexity.  Currently, telcos are leveraging AI to process Big Data, improve operations, and increase revenue.  Specifically, telcos are integrating AI within the body of the company (physical assets – network systems, products and services) as well as leveraging AI to empower their customers (digital agents, bots, speech capabilities).

There’s a real opportunity for Microsoft to be a trusted partner in helping telcos craft their AI strategy, provide them with technical expertise, and with the breadth of our internal resources (e.g. Cognitive Services Lab).

Now is the time for telcos to infuse AI into their products/services to empower the enterprise and provide a unique experience for your customers.  There’s no doubt that to be successful at implementing AI, one must move at a rapid pace to maintain a competitive advantage and differentiate from fierce competitors in conjunction to thinking in a tactful and thoughtful manner.  It’s central to highlight that while moving at a rapid pace, one must be conscious about the positive and potentially negative implications AI can have on society. That’s why Brad Smith, President and Chief Legal Officer, and Harry Shum, Executive Vice President of Microsoft AI and Research Group published The Future Computed: Artificial Intelligence and its role in society, which shares the principles we oblige by when designing AI.

From the inception of Microsoft, we’ve built a business of being the trusted partner in democratizing technology. As we all go through this digital transformation – more than ever, we are in the best position to accelerate our customer’s AI efforts with Azure, our powerful cloud, big computing power, and the top tier talent continuously creating world class algorithms in deep learning and deep reinforcement learning. We know the road ahead will be filled questions, excitement and at times frustration, and it would be misleading to say we have the answers to all the questions that surround this transformative period. Nevertheless, we will listen and strive to meet your unmet, unarticulated needs in a way to improve your business and amplify the lives of your customers and their potential.

To understand more about the AI-enabled services and solutions that Microsoft is delivering to communications service providers, TM Forum Chief Analyst Mark Newman interviewed Worldwide Director of Industry Technology Strategy Rick Lievano.

MN: Where do you see the biggest “wins” for CSPs in adopting AI in the short, medium and long term?

RL: For the short term, we’re seeing CSPs exploring AI to transform customer care.  This may be in the form of customer facing automated virtual agents (bots), or virtual agents designed to assist Customer Service Representatives (CSRs) be more efficient.  Deflecting a single digit percentage of interactions to a low cost intelligent channel can result in millions of dollars in savings per year – and often, happier customers.

In the medium term, AI can help analysts and data scientists make sense of the tremendous volume of data CSPs have been collecting for years.  Whether it’s predictive maintenance of set-top-box or customer-premises-equipment product recommendations, next-best-action, churn prediction, or financial forecasting – AI can make it tremendously easier to find the needle in the haystack.

Although it may sound terrifying, in the longer term CSPs will see the biggest impact once they deeply embed AI across their mission-critical business processes and operations, enabling AI to more efficiently manage and streamline their networks, shape traffic, and move network workloads to where it makes most sense.  There is no shortage of scenarios where adopting AI can provide a huge benefit.  The greatest risk is not moving fast enough to deploy it.

MN: To what extent should CSPs look to build their own AI capabilities rather than rely on technology companies such as Microsoft?

RL: Don’t build plumbing! From my early days as an enterprise software developer, I learned that focusing on building solutions to meet the business need– and not building the underlying platform – is the best road to success.  The same applies to AI – why build core AI capabilities when companies like Microsoft invest billions of US dollars a year developing AI services? Plus the talent war for AI is fierce and it’s extremely difficult to acquire AI talent therefore, it’s difficult to build your own AI capabilities. Here at Microsoft, we have over 7,000 people focusing on AI, making us one of the largest talent pools in the world. Microsoft’s goal is to simplify and “democratize” AI so that anydeveloper can be an AI developer. Sophisticated AI capabilities for vision (e.g. image analysis, object detection, face recognition), speech (e.g. speech-to-text intent identification), and language (e.g. natural-language processing, key phrase extraction) are available via simple REST APIs. And expert AI developers can raise the hood to tweak, customize, and extend these capabilities via the Cognitive Toolkit (CNTK), TensorFlow, or their AI framework of choice.  In short, let Microsoft provide the AI plumbing while CSPs provide the real business innovations on top.  But for software nerds, it’s still a lot of fun to build plumbing.

MN: Do you have many CSP customers today buying AI solutions from you?

RL: CSP Tier-1 customers in every continent have production applications running on Microsoft AI. We have great AI stories with innovative customers like Telefonica, Three UK, Tele2, Portugal’s MEO and many others.  Another interesting perspective is the significant number of telecom ISVs that have embraced the Microsoft AI platform, including Amdocs, Wysdom.AI, Apttus (a leader in configure-price-quote), and others. These ISVs evaluated every major AI vendor and chose Microsoft AI as the underlying platform to build their solutions.

MN: Does Microsoft have any strong thoughts as to whether AI is best deployed in the cloud or on-premise? 

RL: The cloud is an essential pillar in AI’s recent resurgence.  The raw compute power, elasticity, and economic benefits of the cloud make AI possible at scale. Being able to easily provision a thousand cores to run an AI workload – and then quickly dispose of them – is truly transformative.  Generally speaking, AI was built for the cloud.  However, many scenarios are perfectly suitable for on-premise deployments, typically those consisting of workloads where there are regulatory concerns, the model is clearly defined, or data requirements are modest – edge capabilities included (more on that below).  We find that many AI solutions are hybrid and consist of components running across multi-cloud, on-premise, and edge devices.

How important is collaboration and partnerships to AI? Where should CSPs play in this collaborative space?

As a nascent field, modern AI is only barely out of the lab.  Although AI has been around for decades, recent breakthroughs in cloud, big data, and algorithms have ushered its renaissance. For any emerging technology, collaboration and partnerships across both vendors and customers is integral for success and wide adoption. This is why you’re seeing such a high degree of coopetition with vendors like Microsoft, Google, Facebook, and AWS. This collaboration extends to CSPs, where we’ve partnered with many Tier-1’s to not simply provide a turn-key solution, but ideate, iterate, and build something truly innovative.  Telefonica AURA is an excellent example, as is the Tele2 Call Together translation service they’ve recently launched. CSPs can also turn to TM Forum to collaborate with other CSPs and suppliers on AI, join relevant projects, and actively participate in events and catalysts.

From a technology standpoint, Microsoft leverages OSS for many of its AI projects, including the Cognitive Toolkit (CNTK).  And from an ethics and educational perspective, Partnership on AI is a technology industry consortium focused on establishing best practices for artificial intelligence systems and to educate the public about AI.

How is Microsoft today using AI in its own products and services? Does this provide any lessons for telcos?

We’re infusing intelligence throughout our own products to enable millions of people to realize the benefit of AI today and advance our AI platform. If you use products like Office 365, Skype, Bing, Xbox, or Cortana, your experience is being enhanced thanks to Microsoft AI. These innovations initially born in our products are now being funnelled into our AI services, so that any developer can take full advantage of these capabilities.  Closer to the telco-sphere, our Azure cloud relies heavily on AI to efficiently manage its global, hyperscale network. We’ve had years (decades) to finetune and have picked up many scars along the way.  Recent experiments in the social chatbot space have taught us valuable lessons and, undeterred, we’ve continued to tackle the big problems.  The lesson is simple – experiment and expect some failures.  In time, persistence pays off and not trying is the real risk.

There is lots of interest today in edge computing or – as the telecoms community calls it, multi-access edge computing. Will AI have an important role to play at the edge?

Every device has a potential role to play in AI, from the largest cloud server to the lowliest Raspberry Pi. Earlier I mentioned how AI is going hybrid, where multi-clouds from various providers, as well as an on-premise cloud, can perform some AI processing within a larger workflow.  Modern mobile SOCs now include embedded AI capabilities, enabling new scenarios where AI happens at the edge.  We are already working to exploit the AI capabilities present in Huawei’s new Neural Processing Unit (NPU) hardware, enabling faster and more accurate translations while offline. The new Microsoft Seeing AI app leverages AI capabilities in the underlying iOS and hardware to provide a better experience when describing the world around the low vision user. In the IoT world, you can build advanced analytics, machine learning, and artificial intelligence in the cloud and deploy to physical devices using Azure IoT Edge.

For More Information

Microsoft AI

Microsoft Cognitive Services

Sponsor Profile

About Microsoft Corporation

Our mission is to empower every person and every organization on the planet to achieve more.  The telecommunications industry is striving to innovate with new services and improve their customer experience, business insights, and operational efficiency. Microsoft offers enterprise-class platforms and solutions that together with a rich partner ecosystem, help to support the mission-critical operations of today’s communications service providers (CSPs).  Please follow us on Twitter @MSFTTelco or visit for more information and customer stories.


    About The Author

    Director of Business Development, Telecom AI, Microsoft

    Nate Yohannes is a Director of Business Development at Microsoft focusing on Artificial Intelligence in global telecom. Prior to Microsoft, Mr. Yohannes was appointed by The White House to serve in President Obama's administration as the Senior Advisor to the Chief Investment and Innovation Officer, US Small Business Administration. As the Senior Advisor, Mr. Yohannes assisted with managing the Small Business Investment Company, a $25 billion private equity/venture fund and the SBIR program, a $2.5 billion per year grant for high growth businesses engaged in federal R&D that has the potential for commercialization.

    Leave A Reply

    Back to top