Cognitive Business: Inside the Microsoft Cortana Intelligence Suite

Cognitive Business: Inside the Microsoft Cortana Intelligence Suite
This post was published on the now-closed HuffPost Contributor platform. Contributors control their own work and posted freely to our site. If you need to flag this entry as abusive, send us an email.

In this Cognitive Business interview we talk to Lance Olson, Director for Cortana Intelligence at Microsoft, where he works with organizations around the world in the pursuit of insight through the use of big data and advanced analytics.


Lance Olson, Director for Cortana Intelligence at Microsoft / Source: Microsoft

The Cortana Intelligence Suite helps the span of all organization types (from nonprofit to public to private) manage their business decisions through a suite of services centered around analytics and big data predictions. The suite supports cross-functions, from product development to operations. The Cortana Intelligence Suite, can for example, help businesses find the right market segmentation for a given product, understand customer needs, and find the problems that need fixing in areas such as manufacturing.

Interesting, right? Let’s dive into the Q&A.

How does Cortana Intelligence work, from an AI perspective?

First, we must think about the prediction(s) we want, and then we must follow a series of steps to get there. Let me explain through an example.

Let’s say Rolls Royce wants to predict - in real time - engine health and fuel usage (to assure maximum aircraft health). With identified prediction goals, the next steps would require us to:

  1. Identify the outcome you’re tracking, such as aircraft and engine health.
  2. Think of data needed to understand the health of the engine and acquire relevant data inputs - be it data that is streaming, batched (from relational systems), or weather data feeds.
  3. Combine data points to see what’s likely to happen in the engine, and feed the data into Cortana Intelligence to train the system to make predictions - this involves starting with defining what a healthy engine looks like and what an unhealthy engine looks like.
  4. Train the system to identify patterns of healthy and unhealthy engines based on examples; this way the system will learn when to create alerts when it finds anomalies and predicts malfunctions, failures, or inefficiencies.
  5. Continue training, re-training, and fine-tuning the AI systems to assure higher prediction accuracy rates over time.

To see an example of this in action, check out this Rolls Royce case.

What sets Cortana Intelligence apart from its competition?

We focus on simplicity that can be done out of the box. In a matter of minutes, a business decision maker can look at the business dashboards that they can play with, plug in data, and see what the solutions look like with their data online using the Cortana Intelligence Gallery. Other differentiators include:

  1. lower consulting footprint
  2. the breadth of the cloud Cortana Intelligence sits on; we’re in 34 regions worldwide and growing quickly, hold the record of having the most secure cloud in the world, offer sovereign clouds, and meet sovereign country compliance requirements
  3. 15+ years working on search (i.e., Bing) and related research; we understand how people search; we’ve spent a lot of time understanding image processing and the search process, which requires the ability to accurately learn what people mean from written and spoken communication.

How do you define AI vs Big Data?

Big data involves new patterns for processing data that differ from traditional methods in their volume (size of data), variety (unstructured and semi-structured data), or velocity (real-time streaming data). Artificial Intelligence is about using algorithms and data to create a result that takes on more human traits such as speech, vision, and language understanding – the ability to read a sentence and understand the intent of the person who created it. AI often uses big data to build better algorithms and learn, but AI can be applied to many use cases that operate on traditional data sources, such as looking at a customer table in a relational database and predicting the customer’s propensity to buy a particular product.

What trends and challenges are most impacting AI right now?

Three trends that are having a transformative impact on AI today are the growth and availability of data through IoT, the increase in compute capacity from maturing cloud architectures such as Azure, and the increasing sophistication of the resulting algorithms that are available to us. One challenge we face is that the tools needed to build AI-driven systems are still relatively immature and therefore the skillsets needed are in scarce supply. So while it is now possible to do amazing things, it is not as easy as it should be.

What are your predictions about the future of AI 2016 and beyond?

Cognitive services such as speech and language understanding will continue to become increasingly useful, especially in bot scenarios in which written and spoken conversations are the primary interface.

Artificial Intelligence tools like Cortana Intelligence will become usable by a much broader set of people, often embedded as intelligence within existing tools and applications. This will enable more people such as developers and business analysts to perform machine learning and artificial intelligence tasks without being formally trained as data scientists. Here’s an example of how we recently did this with Uber to enable face authentication for their drivers.

What makes a company ready to implement Cortana Intelligence?

Most organizations can see ways they could better achieve their goals if they had an increased ability to predict future outcomes, but having the desire to improve in this area is one of the key requirements.

How can a company prepare for Cortana Intelligence and AI in general?

Pick a few key areas where you’d like to focus and start with a relatively simple project. Becoming highly data and AI-driven is like running a long-distance race. We don’t usually just go out and have great success on the longer distances without building up. It is an iterative process where we solve smaller problems that build on each other, and increase our momentum with each project. Having the right data is critical to success in these projects so it is good to start by taking stock of your current data assets and identifying areas where you can use what you have or extend its value in interesting ways by combining it with new datasets.

What industries do you see AI benefiting first in your space, and why?

AI is likely to impact nearly every industry in the world much in the same way the computer has done over the past 50 years. That said, in the last couple years I’ve seen a lot of momentum in Retail (e.g., Lowe’s smart kitchen design center, Ziosk restaurant kiosks), Manufacturing (e.g., Rolls Royce connected engines, eSmart Systems, Schneider Electric smart power), Healthcare (e.g., Dartmouth Hitchcock connected care), and Banking and other Financial services (e.g., Tangerine, Quarterspot loan success predictions). These industries have significant amounts of data already available to them that they can use at relatively low costs to gain real efficiencies in how they run their operations, employ their people, improve their products, and anticipate the needs of others so they can better serve their customers.

What is your favorite AI use case?

I’m torn. My favorite use cases range between a health and a business case. I’ll tell you about both. In India, doctors are using Cortana Intelligence to predict Lasik eye surgery outcomes and recovery times, ensuring optimal results for surgery patients at a lower cost and higher success rate.

In the U.S., Lowe’s is using Cortana Intelligence to provide their kitchen remodeling customers an exciting experience that allows them to quickly decide on their dream kitchen and quickly bring it to life. Partnering with Pinterest, Lowe’s today is leveraging Cortana Intelligence in the HoloLens.

When people set out to remodel their kitchen, they find it hard to make decisions on what it should look like - especially when there are multiple stakeholders involved, as is often the case. What we found is that people will use Pinterest and they will pin images of dream kitchens that they want. Taking note, we partnered with Lowe’s and Pinterest, to show what’s possible if we enable customers to log into their Pinterest account once inside the store. Cortana Intelligence takes the visual information, extracts ideas from images pinned (on Pinterest), and identifies the customer’s style preferences. Cortana Intelligence then compares those preferences with the Lowe’s catalog and popular purchases. Through that process it produces a Lowe’s version of your dream kitchen.

Enter the HoloLens. Cortana Intelligence interacts with the HoloLens and creates the customer’s augmented kitchen. The customer can then click and style - multiple people can participate. The customer can see the design in an immersive and dynamic environment. Say they want to change the backsplash that was generated, they can do that. Cortana Intelligence recommends adjustments to the kitchen design, based on customer sentiment. When the customer is happy with the design, he can decide to have the design sent to him as well as an estimate of his kitchen remodel.

This use case has proven to be a win-win for the customer and Lowe’s. For the customer, they get their dream kitchen. For Lowe’s, they get insights into customer sentiment and they get to see a decrease in time to decide on a kitchen remodel, which results in revenue.

“Cognitive Business” is an interview series featuring awesome people in the Artificial Intelligence (AI) world. Written by Lolita Taub and written for business people.