Microsoft is creating a new engineering organisation called CoreAI – with the mission of building a stack for “both our first-party and third-party customers to build and run AI apps and agents” said CEO Satya Nadella.
Microsoft CoreAI, announced Monday, will be led by Meta’s former Global Head of Engineering Jay Parikh, who joined Microsoft in 2024 – and tap talent from the CTO’s office, its Developer Division, and beyond.
Parikh said in a LinkedIn post: “Model-forward applications will redefine every layer of the application stack… We’re helping customers and partners move beyond first-generation AI applications into a world of dynamic, agentic applications – applications designed to act on their behalf with memory, adaptability, and intelligence at their core.”
CoreAI gets influential CVPs reporting to Parikh
Microsoft’s Eric Boyd, CVP, AI Platform; Tim Bozarth, CVP, Developer Infrastructure; Julia Liuson, President, Developer Division; and Jason Taylor Deputy CTO, AI Infrastructure, will join CoreAI and report to Parikh.
Nadella in a short missive shared with Microsoft employees this morning, claimed that “our AI platform and tools will come together to create agents, and these agents will come together to change every SaaS application category, and building custom applications will be driven by software (i.e. ‘service as software’)... [CoreAI will] also build out GitHub Copilot, thus having a tight feedback loop between the leading AI-first product and the AI platform to motivate the stack and its roadmap.”
The creation of the new Microsoft organisation, CoreAI, comes after Nadella created waves in an interview with the BG2 podcast in which contended that business applications will “collapse in the agent era.
“If you think about it”, Nadella said in that December 12 interview, “they are essentially CRUD databases with a bunch of business logic… the business logic is all going to these agents. These agents are going to be multi- repo CRUD… so they're not going to discriminate between what the back-end is they're going to update multiple databases and all the logic will be in the AI tier.” (Effective agentic AI, in short, “kills” SaaS.)
(AINIRO’s Thomas Hansen suggested in the wake of that claim: “This implies that if you can extract meta data from your database, you can generate CRUD apps that propagates meta data from the structure of your database, through the HTTP layer, ending up as function declarations the LLM can execute. The process is based upon meta programming, in a homoiconic graph-based programming language…”)
This view may be prescient or it may be wildly ambitious. Critics claimed it hinted at a world in which data protection rules bespoke to discrete applications and their datasets are eaten by, for example, a Microsoft.
Others of a more optimistic vein suggested that Nadella simply meant that AI agents will “replace integration… allowing posting and retrieval transparent to underlying applications. This will open the marketplace to smaller niche applications replacing… mammoth ERP systems.”
Microsoft's CoreAI has work to do...
For all the hype around “Agentic AI”, agent interaction “remains a significant challenge, often leading to inefficiencies and conflicts” as Microsoft Research’s own Ryan White and Professor Chirag Shah of the University of Washington noted just weeks earlier in an ArXiv paper.
Among the potential architectural adjustments they proposed were the implementation of “caching solutions that store and execute agent workflows and reduce the need for calls to foundation models…” and the use of “hierarchical architectures that integrate small language models and large language models” for much improved scalability and efficiency.
That is perhaps the tip of the iceberg Nadella wants to own.
As he put it today: “We will build agentic applications with memory, entitlements, and action space that will inherit powerful model capabilities. And we will adapt these capabilities for enhanced performance and safety across roles, business processes, and industry domains. Further, how we build, deploy, and maintain code for these AI applications is also fundamentally changing and becoming agentic.
“This is leading to a new AI-first app stack — one with new UI/UX patterns, runtimes to build with agents, orchestrate multiple agents, and a reimagined management and observability layer” Microsoft’s CEO said.
