Amid a whirlwind of buzz, Microsoft reshuffles the entire dev team for #ArtificialIntelligence

Microsoft’s Chief Executive, Satya Nadella has declared a monumental reorganization of the companies engineering division which is realigning the company toward the creation of the frameworks which will host agentic AI. The new division is called “CoreAI – Platform and Tools” which has sent the prior AI platform team and the prior developer division (responsible for operations ranging from .NET to Visual Studio) and several other teams. As for what this group will be doing specifically, it's basically everything that's mission-critical to Microsoft in 2025, as Nadella tells it:

A new division will include the Dev Div, AI Platform, and some key teams from the Office of the CTO (including the AI Supercomputer, AI Agentic Runtimes, and Engineering Thrive), whose objective will be to deliver the end-to-end Copilot & AI stack for first- and third-party users to build and deploy AI apps and agents. This group will also scale out GitHub Copilot — in effect, they have a tight feedback loop between the leading AI-first product that sets the pace for the rest of the stack and the AI platform to drive motivation for the stack and its vision, roadmap.

To achieve all that, “Jay Parikh will lead this group as EVP.” Parikh joined Microsoft in October; prior to this role, he was the VP and global head of engineering at Meta. That in the blog post there is not a single word about .NET or Visual Studio and almost everything is about GitHub Copilot and anything and everything related to agentic AI spells out exactly how Nadella envisions the company’s future.

The so-called AI agents are applications that are endowed with quantified action spaces well-defined as the degree of freedom and working memory to perform subsets of what humans in clerical jobs do now. Overly optimistic decennialists think that these agents will just replace jobs outright while more moderate AI commentators think they will be just another force multiplier that added to people’s jobs.

On this front, Nadella doesn’t sound particularly conservative, either. His blog post begins with the following incredibly lofty set of propositions:

The fourth year of 2021 will be the start of the next innings of the AI platform shift and 2025 will be centred on model-forward applications that revolutionise each and every application type. Not since the change from client-server to web has the application stack at every level be changed. It is similar to GUI, internet servers, and native cloud-based databases all being thrown into the app stack at once. What has happened in the last 30 years is now being played in the next 3 years!

And those are very big promises and expectations at that. If we are to believe his words, he seems to believe that agentic AI will be at the center of this process and that over the next three years, Microsoft’s mission is to ensure that it is providing the software and services that will be central to this. Thus, agentic AI is portrayed as the next big point that is to turn the tide in the favor of the tech industry, and more globally – in the world of work.

This is viewed as more progress than from GP LLM chatbots such as ChatGPT that are not specifically specialized in anything and rely on informed users to get the best out of them and LS-FA-Deep-Learning-Apps reminiscent of Apple’s integrated features and platform /software offerings. Almost everyone knows that some form of generalized deployment of agentic AI is bound to happen, but Nadella’s vision for how this reorganization of software construction will radically reshape the industry toward a fully nuaged world typifies one of the most optimistic views on agentic AI at the very least.

CEO of Microsoft: Satya Nadella