The tech giant launched a wide range of innovations in platforms, tools, cloud services and other offerings that should help its Azure public cloud continue setting the pace in a wide range of solution segments. At the same time, the new services should enrich its deep stack of on-premises and hybrid-cloud platforms, tools and apps.
In Wikibon’s perspective, Microsoft is a clear leader in driving cloud-based artificial intelligence and machine learning applications into the hands of users in both the business and consumer worlds. The vendor clearly played its strong suit in the Monday announcements by rolling out a wide range of AI/ML-related feature enhancements across its Dynamics, Office 365 and other application solutions for business and consumer markets.
Microsoft’s investments in core AI/ML platforms and tooling are now table stakes to compete with Amazon Web Services, Google Cloud Platform, IBM Cloud and others in the hotly competitive cloud arena. Carrying forward initiatives that had been introduced or extended in the past several years’ Build events, Microsoft made several key solution announcements related to its core AI offerings that are clearly targeted at the new generation of AI/ML application developers:
- Enhanced AI DevOps automation: Microsoft announced enhancements to its automated AI/ML capabilities and tooling to ensure reproducibility, auditability and automation within a standardized end-to-end AI/ML DevOps workflow.
- Skill-tailored AI app-development tooling: Microsoft announced code-free visual AI/ML development interface and new AI/ML notebooks for developers who prefer a code-first experience.
- Rich prebuilt AI application services: The vendor launched new AI-infused cognitive cloud services to accelerate customer development of vision, speech, language, search, recommendations, personalization, handwriting and forms recognition and technical-domain real-time speech transcription.
- Reusable AI-driven search results: The company previewed new cloud services that enable AI-driven search results to be reused in the development of intelligent apps that embed ML models or BI visualizations.
- Curated AI modeling and training data: Microsoft previewed an Azure service that provides curated data to help customers improve AI/ML model accuracy while speeding upfront data preparation.
- AI inferencing acceleration: Microsoft launched new hardware-accelerated models optimized to run on FPGAs as well as runtime support for the Open Neural Network Exchange specification for executing Nvidia Corp.’s TensorRT and Intel Corp.’s nGraph software platforms for high-speed AI/ML inferencing on those vendors’ respective chipsets.
However, the most important announcements on Build’s first day focused not so much on the underlying AI/ML but on the capabilities needed to tailor and deploy these applications for the new world of intelligent edge apps in “internet of things,” mobile and embedded environments. The principal Build announcements that distinguish Microsoft with edge application developers include:
- Developing write-once, run-everywhere AI from cloud to edge: Microsoft Azure SQL Database Edge, currently in preview, helps developers build AI apps that run anywhere from the core Azure clouds all the way down to edge devices operating in cloud connected or fully disconnected edge scenarios. It enables write-once, run-anywhere AI apps through the same programming abstractions implemented in Microsoft’s core cloud data platform, Azure SQL Database and the on-premises SQL Server. The solution’s runtime incorporates the same data streaming, time-series data, in-database machine learning, and graph processing technologies that are implemented in the Azure cloud, supporting a full range of sophisticated low-latency AI apps. Any AI/ML developer who builds applications that are optimized for the Azure cloud should join the Early Adopter Program to access the preview of Azure SQL Database Edge. This offering can free AI app developers from having to learn new tools and languages in order to build intelligent edge solutions. It also allows enterprise DevOps professionals to use the same application management and security tools to control AI workloads on Arm and x64-based interactive devices and edge gateways through the same centralized control plane that they use with those running in the Azure core and IoT clouds.
- Modeling AI-enriched edge applications for transparent cross-edge deployment: Another key announcement in Microsoft’s intelligent edge strategy was IoT Plug and Play. This specifies a new open modeling language for connecting IoT devices to the cloud at scale. The language enables new AI and other software to be written once for deployment to all supported partner-certified devices, rather than specifically for each connected device. Under this announcement, Microsoft is providing customers a large ecosystem of IoT Plug and Play partner-certified devices that support fast cloud-to-edge connectivity. It is also providing new tools for building AI-enriched mixed-reality and gaming applications for mass-market IoT and other application environments.
- Extending graph-based context throughout distributed edge apps: Microsoft extended the graph-computing technology that it announced two years ago at Build in support of a fully device-agnostic edge computing strategy. The vendor announced that Microsoft Graph — which operates as a connective thread for contextual intelligence across cloud, IoT/edge, enterprise and other applications — may now incorporate users’ own business data for other, non-Microsoft platforms and apps. Alongside the new Fluid Framework and Edge browser enhancements, Microsoft Graph provides a development abstraction plane for building seamless, real-time and interactive experiences in multidevice world.
- Autolearning edge-application business logic: Microsoft announced a new “Semantic Machines” capability that can automatically learn how to map people’s words to the computational steps needed to carry out requested tasks. This technology can free developers from the need to script to the last detail every possible skill that might be required of an edge-based intelligent apps. It also enables those agents, bots and other apps to dynamically evolve their controlling logic through local application of adaptive AI to fresh chat, interaction and sensor data. The technology will be made available through Microsoft’s Cortana and Bot Framework in order to drive fluidly contextual and intelligent interactions from edge apps.
- Simplifying edge business-logic governance via distributed blockchain tech: We shouldn’t overlook the importance of Microsoft’s newly announced Azure Blockchain Service to its edge application-development strategy. Building on the blockchain application-development workbench that was announced at last year’s Build, this offering, currently in public preview, simplifies deployment and governance of business logic that’s persisted in distributed Ethereum community blockchains. It includes a Visual Studio tool for creation, compilation and automated deployment of smart contracts, application code and other business logic that’s written to a immutable blockchain running on Azure cloud and managed using Azure DevOps services. The new Azure Blockchain Dev Kit connects business processes managed in myriad application and data platforms running in Azure as well as in on-premises platforms, hybrid clouds, IoT meshes, serverless environments and other complex deployments.
- Building and tuning fully autonomous edge apps: Microsoft announced a program that offers developers of autonomous edge apps the opportunity to work with its experts building such intelligent agents using its AI, cloud, IoT and robotics tools. The program includes access to a new category of “machine teaching” tools that allow domain experts who aren’t data scientists to build AI business logic that drives autonomous systems. It also includes access to simulation technologies, such as Microsoft’s AirSim, that enable autonomous devices to learn in controlled, realistic environments.
Developers of cloud-to-edge apps with AI at their core should strongly consider Microsoft’s Azure-centric development environment for their most sophisticated projects.
However, what was lacking from the first day’s announcement was any comprehensive Microsoft approach to one of the hottest edge-app segments: robotic process automation. Though the new Form Recognizer and Semantic Machines services have obvious RPA applications, none of the Microsoft keynoters or other announcements spelled out a comprehensive approach to addressing that intelligent-edge opportunity.
Also, Microsoft’s new autonomous AI developer engagement program lacks the visionary scope and product depth that characterize rival AWS’ approach, rolled out last fall, to this growing edge-application developer focus.
Clearly, these are significant gaps in Microsoft’s portfolio coverage of intelligent edge from a development standpoint. Nevertheless, Wikibon has every confidence that Microsoft will address them in the coming year through its usual mixture of cutting-edge research and development, strategic acquisitions and astute partnering activities.