Microsoft is on the edge: Windows, Office? Naah. Let's talk about cloud, AI
At its dev conference, Redmond is all about Azure, bots, Azure and also Azure
Build At the Build 2017 developer conference today, Microsoft CEO Satya Nadella marked a Windows milestone – 500 million monthly active users – and proceeded to say very little about Windows or Office.
Instead he, along with Scott Guthrie, EVP of the Microsoft Cloud and Enterprise Group, and Harry Shum, EVP of Microsoft's Artificial Intelligence and Research group, spent most of their time on stage, in Seattle, talking about Azure cloud services, databases, and cross-platform development tools.
Arriving on stage to give his keynote address, Nadella in jest said that he thought it would be an awesome idea on such a sunny day "to bring everyone into a dark room to talk about cloud computing."
Office and Windows can wait.
Microsoft watchers may recall that its cloud-oriented businesses have been doing well enough to deserve the spotlight. In conjunction with the company's fiscal second quarter earnings report in January, the Windows and Office empire revealed that Azure revenue grew 93 per cent year-on-year.
During a pre-briefing for the press on Tuesday, Microsoft communications chief Frank Shaw described "a new worldview" for the company framed by the "Intelligent Edge" and the "Intelligent Cloud."
Nadella described this newborn weltanschauung as "a massive shift that is going to play out in the years to come."
He mused about a software-based personal assistant to illustrate his point. "Your personal digital assistant, by definition, will be available on all your devices," he said, to make the case that the centralized computing model, client and server, has become outmoded. Data and devices are dispersed.
In other words, all the data coming off connected devices requires both local and cloud computing resources. The revolution will not be centralized.
That could easily be taken as reheated Cisco frothing about the explosive growth of the Internet of Things and bringing processing smarts to the edge of the network. But Microsoft actually introduced a new service that fit its avowed vision.
Microsoft's bipolar worldview – the Intelligent Edge and the Intelligent Cloud – manifests itself in a novel "planet scale" database called Azure Cosmos DB. It's a distributed, multi-model database, based on the work of Microsoft Researcher Leslie Lamport, that promises to make data available locally, across Microsoft's 34 regions, while also maintaining a specified level of consistency across various instances of the data.
An Intelligent Meeting demonstration, featuring Cortana, showed how AI has the potential to exchange and coordinate data across multiple services. But "potential" requires developer work – it will take coding to create the Cortana Skills necessary to connect the dots and manage the sort of cross-application communication that knowledge workers accomplish today through application switching, copying, and pasting.
Conveniently, the Cortana Skills Kit is now in public preview, allowing developers to extend the capabilities of Microsoft's assistant software to devices like Harman Kardon's Invoke speaker.
Beyond code, it will take data associated with people and devices in an organization to make those connections. That's something Microsoft – with its Azure Active Directory, its Graph, and LinkedIn – has in abundance.
A demonstration of real-time image recognition to oversee a construction worksite showed how a capability like image recognition might be useful to corporate customers. Cameras spotted unauthorized people and located requested equipment on-site. It looked like something companies might actually find useful.
Artificial intelligence as a general term sounds like naive science fiction. But as employed by Microsoft, it refers to machine learning frameworks, natural language processing, computer vision, image recognition or the like.
"We believe AI is about amplifying human ingenuity," said Shum.
Microsoft's concern is convincing developers and corporate clients to build and adopt AI-driven applications using Microsoft cloud computing resources, rather than taking their business to AWS or Google Cloud Platform.
One way Microsoft hopes to achieve that is by offering cloud computing outside the cloud, on endpoints like IoT devices. The company previewed a service called Azure IoT Edge to run containerized functions locally. It's a way of reducing latency and increasing responsiveness, which matters for customers like Sandvik.
The Swedish industrial automation biz has been testing Azure IoT Edge to anticipate equipment failure in its workplace machines, in order to shut them down before components break, causing damage and delays.
It wouldn't be a tech keynote without, yup, oodles of artificial intelligence
Like its peers, Microsoft has become convinced that everything is better with AI inside. It added several new AI-infused offerings – Bing Custom Search, Custom Vision Service, Custom Decision Service and Video Indexer – bringing its Cognitive Services count to 29, if anyone's counting.
In fact, Microsoft is counting. It says that some 568,000 developers from more than 60 countries have signed up to use Cognitive Services since the AI-oriented portfolio was introduced at Build two years ago. It also claims that 130,000 developers have signed up to use its Bot Framework since the software's introduction last year.
As point of comparison, Facebook at its F8 conference trumpeted 100,000 bot developers for its Messenger Platform and a similar number of bots.
And just as Facebook has had to temper its bot expectations and shift away from dialog-based interaction, the way Microsoft personnel talk about bots suggests that interactive, conversational software hasn't been quite as transformative as hoped.
"We really still believe that in a couple of years everything will want to be able to understand language and speech, said Lili Cheng, general manager of Microsoft FUSE Labs, during the press pre-briefing.
In the meantime, as Facebook has learned, menus offer clarity for bots that text- and speech-based interaction can't quite deliver yet.
"We continue to believe very strongly that [bots] are an exciting part of what we're offering," comms chief Shaw insisted. His message sounded more like reassurance than wild-eyed evangelism.
The Microsoft Bot Framework has become a bit more capable, with the addition of Adaptive Cards. Cards are rich media interfaces that work across multiple apps and platforms. Bot developers can also now publish bots that can access channels like Bing, Cortana, and Skype for Business and can implement payment requests via API.
For developers who wish to train their own neural networks, Shum previewed a service called Azure Batch AI Training that supports building models using frameworks like Caffe, Microsoft Cognitive Toolkit, and TensorFlow. This service is in private beta at the moment.
Visual Studio for the Mac reached general availability, offering developers a well-regarded IDE for creating cross-platform apps in C#, among other languages, in conjunction with Xamarin's tooling, .NET Core, and Azure.
Speaking of Azure, there's now software called Azure Cloud Shell, for authenticated, browser-based shell access to Azure resources. Azure can also manage MySQL and PostgreSQL databases as a service, under the names Azure Database for MySQL and Azure Database for PostgreSQL. And there's now a mobile app called Azure Mobile Portal for Android and iOS.
Kiss goodbye to that boring old Oracle stuff
In an effort to win customers away from the competition, Microsoft has introduced its Azure Database Migration Service to help move third-party and SQL Server databases into Azure SQL Database with minimal downtime.
As Guthrie observed during the keynote, Microsoft is hoping to entice Oracle customers to defect.
Azure Service Fabric, which lets customers run Windows- and Linux-based containers, can now handle containers spun up via Docker Compose. And with the arrival of Service Fabric's 5.6 runtime and 2.6 SDK, Windows Server containers have reached general availability.
Visual Studio 2017, through an extension, gained tools for serverless computing development, through Microsoft's event-driven managed programming services Azure Functions and Azure Logic. The capabilities include an Azure Functions Runtime preview, which lets Azure Functions run on-premises, and Azure Application Insights support for Azure Functions, which makes analytics and diagnostic capabilities available for Azure Functions.
There were a number of other Azure-related announcements not mentioned at the keynote that Guthrie touches on in a blog post published on Wednesday.
Microsoft opened its chat-based workspace, Teams, to all developers, enabling them to publish apps through the Office Store and to make those apps more visible via an improved app discovery process. It introduced a preview of Teams in November.
There was an impressive demonstration of the forthcoming Presentation Translator add-in, which brings real-time language translation APIs to PowerPoint. The presenter's speech was quickly and accurately translated and displayed on screen.
Microsoft Graph APIs have expanded to reach into OneNote, SharePoint site, and Planner.
The Redmond-based software giant plans to release new Insights APIs for Azure analytics. But they're not yet available, even as a technology preview. Two novel Microsoft Graph capabilities did see the light of day: delta queries, for listing data changes related to Graph, and custom data, which lets developers extend base Graph data types, like users or contacts.
Actionable Messages, previewed in web-based Outlook last fall, have made it to Office 365 users in Microsoft Teams and in Outlook 2016 for Windows, bringing additional integrations with third-party services like Salesforce.
And in happy ending of sorts, Office Store has gotten hitched to Microsoft AppSource, enabling organizations to manage their software in a more organized fashion. Along similar lines, there's now centralized deployment with Office for Mac and Office Online. It appears there's room for central control somewhere between the Intelligent Edge and the Intelligent Cloud.
Closing the day-one keynote, Shum recalled the words of computer science luminary Alan Kay: "The best way to predict the future is to invent it."
And after that, set aside some time to debug the unintended consequences. ®