Microsoft Build 2025 keynote: Everything announced, in 14 minutes

Good morning and welcome to Build

And here we are in 2025, building out this open agentic web at scale

And we're going from these few apps with, you know, vertically integrated stacks

to more of a platform that enables this open, scalable agentic web

More importantly, it's all about expanding that opportunity for developers across every layer

We have a bunch of new updates. We are rolling out at Bill, starting with Visual Studio

It is the most powerful IDE for .NET and C++, and we're making it even better

.NET 10 support, a live preview at design time, improvements to Git tooling

a new debugger for cross-platform apps, and much, much more. And when it comes to VS Code

just a couple of weeks ago, we shipped our 100th release in the open

It included improved multi-window support and made it easier to view stage

directly from within the editor. And GitHub continues to be the home for developers

GitHub Enterprise has tremendous momentum in the enterprise. And we're doubling down for developers building any applications

Trust, security, compliance, auditability, data residency are even more critical today. As GitHub Copilot has evolved inside VS Code, AI has become so central to how we code

And that's why we're open sourcing Copilot in VS Code. this. Starting today, we will integrate these AI-powered capabilities directly into the core

of VS Code, bringing them into the same open source repo that powers the world's most loved

dev tool. In fact, we're building app modernization right into agent mode, right? So Copilot now

is capable of upgrading frameworks like a Java 8 to Java 21 or .NET 6 to .NET 9

and migrate any on-premise app to the cloud. And the next thing we're introducing is an autonomous agent

for site reliability engineering or SRE. The SRE agent starts automatically triaging, root-causing, mitigating the issue

and then it logs the incident management report as a GitHub issue with all the repair items

And from there, you can even assign the repair items to GitHub Copilot

A full coding agent built right into GitHub, taking Copilot from being a pair programmer to a peer programmer

You can assign issues to Copilot, bug fixes, new features, code maintenance

and it'll complete these tasks autonomously. And today, I'm super excited that it's now available to all of you

I don't think since Teams launched, we've had an update of this level

and it really brings together chat, search, notebooks, create, and agents all into this one scaffolding that's intuitive, right

I always say this is the UI for AI And chat for example is grounded both on web data as well as your work data and that the game changer especially with pages Search works across all of your applications

right, whether it's Confluence or Google Drive or Jira or ServiceNow, not just M365 data

With notebooks, I can now create these heterogeneous collections of data, right? In fact

I can have chats and pages and any documents, emails, all in that collection

And then, in fact, I can get all these audio reviews or podcasts out of it

You know, I can use create, you know, to turn a PowerPoint into a new explainer video or generate an image

And when it comes to agents, we have a couple of special agents, like researcher, right

It's been perhaps the biggest game changer for me because it's synthesizing across both the web and enterprise sources, right

Applying deep chain of thought reasoning to any topic or any project

yst goes from raw data across multiple source files. I can just upload a bunch of Excel files

It will get the insights. It will do forecasts. It will do all the visualizations

All of the agents you build can now show up in Teams and in Copilot

and you can ask questions, assign action items, or kick off a workflow by just at-mentioning an agent in a chat or meeting

And with the Teams AI library, building multiplayer agents is easier than ever

It now supports MCP. And with just one line of code, you can even have it enable A2A

And you can add things like episodic or semantic memory by using Azure Search and a new retrieval system

which I'll talk about later. And as a developer, you can now publish. And this is the biggest thing, right

Now you can build an agent, you can publish your agent to the agent store

and have them discovered and distributed across both Copilot and Teams, providing you access to the hundreds of millions of users

and unlocking that opportunity. Today, we are introducing a new class of enterprise-grade agents you can build

using models fine-tuned on your company's data, workflows, and style. We call it co-pilot tuning

You know, co-pilot can now learn your company's unique tone and language

and soon it'll even go further understanding all of the company's specific expertise and knowledge

All you need to do is seed the training environment with a small set of references and kick off a training run

The customized model inherits the permissions of all the source control. And once integrated into the agent, it can deploy to authorized users

Our new model router will automatically choose the best OpenAI model for the job

No more sort of those manual model selections. An approach today, though, goes from having apps that you built or agents you build only bind to one model

to truly becoming multi-model. That's why today we are thrilled to announce

Grok from XAI is coming to Azure. When you have multiple models

what you need is a new capability in how you use these models. And now you can provision throughput once on Foundry

and you can use that provision throughput across multiple models including Grok That just a game changer in terms of how you think about models and model provisioning And the Foundry agent service lets you build declarative agents

in fact, just with a few lines of code just in the portal. For complex workflows

it supports multi-agent orchestration. And I'm excited to share that now the agent service

is generally available. And we're making it straightforward, for example, for you to connect

Foundry to your container app or functions, and deploy any open source model into AKS

whether it's in the cloud or in hybrid mode with Arc. And you can now take a model

fine-tune it or post-train it in Foundry, and then drop it right into Copilot Studio so that

you can now use that post-train model to automate a workflow or build an agent. This healthcare

agent orchestrator that Stanford used is now available to everyone in Foundry. It's pretty awesome

We now have new observability features coming to Foundry to help you

monitor and manage AI in production. You can track the impact, quality, safety, as well as

cost all in one place. With EntraID, agents now get their own identity permissions

policies, access controls. The agents you build in Foundry and Copilot Studio

show up automatically in an agent directory in Entra. We're also partnering with ServiceNow and Workday

to bring automated provisioning and management to their agents via Entra. And when it comes to data governance

Purview now integrates with Foundry, right? So when you write an agent automatically because of Purview

you can ensure end-to-end data protection. Another massive safety consideration. And on the security side, Defender now integrates with Foundry

So that means your agents are also protected just like an endpoint would be from threats like wallet abuse or credential theft by Defender

Now, we want to bring the power of this app server and app building capability to the edge and clients as well with Foundry Local

which we are announcing today. It includes a fast, high-performance runtime, models, agents as a service

and a CLI for local app development. And yes, it's fully supported on Windows and the Mac

We're excited to announce the Windows AI Foundry. Windows AI Foundry is what we used, in fact

ourselves internally to build features. And now we're extending this platform to support the full dev life cycle, right

Not just on co-pilot PCs, but across CPUs, GPUs, NPUs, all and in the cloud, right

So you can build your application and have them run across all of that silicon

And Foundry Local is built into Windows AI Foundry, so you can tap into this rich catalog

of these pre-optimized open source models that you can run locally on your device

we're announcing native support for MCP in Windows. Windows will now include several built-in MCP servers

like file systems, settings, app actions, as well as windowing. And we're adding native MCP registry

that lets MCP-compatible clients discover the secure MCP servers that have been vetted by us for security performance all while keeping you in control We first announced Bash on Ubuntu on Windows nearly 10 years ago It subsequently became

what we obviously call today WSL. Today we are making WSL fully open source

And so we're announcing today, and you all should go check out the code in the GitHub repo

So NLWeb, it is a way for anyone who has a website or an API already to very easily make their website or their API an agentic application

We're integrating Cosmos DB directly into Foundry. So that means any agent can store and retrieve things like conversational history

and soon they'll be able to do, also use Cosmos for all their RAG application needs

And we're taking it further with Azure Databricks, connecting your data in Genie spaces

or in AI functions to Foundry. The other very cool capability is now

inside of a Postgres SQL query, you can have LLM directly, LLM responses directly integrated

We're bringing Cosmos DB to Fabric too, right? because AI apps need more than just structured data

They need semi-structured data, whether it's text, images, audio. And with Cosmos and Fabric and your data instantly available alongside SQL

you can now unify your entire data estate and make it ready for AI

And there's a lot more. In fact, we're even building our digital twin builder right into Fabric

Now you can very easily take digital twins with no code, low code

As you can see here, you can map the data from your physical assets and systems super fast

We're also announcing shortcut transformations in one leg. You can think of this as AI-driven ETL

You can apply all these pre-built AI-powered transformations, audio-to-text or sentiment ysis when data is coming in

summarization, all powered by Foundry straight into Fabric. So in fact, the largest GB200-based supercomputer is going to be Azure

And so we are very, very excited about scaling this and making it available to all of you as developers

We are bringing together the entire stack I talked about today and saying, look, let's apply it to science

And the scientific workflow, the scientific process, That's our ambition with Microsoft Discovery, which we are announcing today

It understands the nuanced knowledge in the scientific domain, right, from public domain as well as your own data if you're a biopharma company

Discovery is built on Foundry bringing advanced agents, highly specialized in R&D, not just for reasoning but for conducting research itself

It's great to see how companies across every industry are already using Discovery to accelerate their R&D

and I can't wait to see this in the hands of more R&D labs all over and what they can do

So that was a quick, comprehensive, whatever you want to call it, walk through the full stack

and how we're creating new opportunity for you across the agentic web

We're taking really a systems approach, a platform approach, which you can expect from Microsoft across every layer of the stack