Everyday AI Podcast – An AI and ChatGPT Podcast - EP 529: Microsoft Build Updates: 5 new Copilot AI updates and how to use them

Episode Date: May 20, 2025

Microsoft legit just dropped a book of AI updates at the Build Conference.We're going to go over the 5 most impactful AI-powered Microsoft Copilot updates and how they will change the future of work. Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion: Have a question? Join the convo here.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:GitHub Copilot's Autonomous Coding Partner UpdateCopilot Tuning for Enterprise CustomizationIntroducing Agent Foundry on AzureMulti-Agent Orchestration in Copilot StudioComputer Use Automation in CopilotMCP Native Support in Microsoft SystemsTimestamps:00:00 "Everyday AI: Transform Your Business"06:42 AI Coding Assistant Evolution09:29 Copilot Tuning for Business Leaders10:56 Data Privacy Concerns in Cloud Use16:52 "AI Collaboration Among Tech Giants"20:48 "Multi-Agent Orchestration Cautions"22:59 "Multi-Agent Orchestration in Copilot Studio"25:27 OpenAI Copilot Access and Availability29:38 Copilot Pro: Versatile AI Agent35:13 Microsoft Embraces Open AI Collaboration36:57 "Security Concerns Slow AI Rollout"39:44 Subscribe & Review RequestKeywords:Microsoft Build 2025, AI updates, Copilot AI updates, GitHub Copilot, GitHub Copilot coding agent, Autonomous coding partner, Visual Studio Code, Multimodal understanding, Natural language prompts, MCP protocol, Model context protocol, Anthropic, Microsoft 365 Copilot, Business leaders, Copilot tuning, Organization's internal data, Low code model tuning, Task specific agents, Secure service boundary, Azure, Agent foundry, AI agent playground, Enterprise grade AI agents, Grok, Elon Musk, Microsoft Azure, Agent to agent protocol, A to A, Multi agent orchestration, Copilot Studio, Agents collaboration, Agentic memory, Automated validation tools, Computer use in Copilot, Desktop applications, Repetitive tasks, MCP native support, Windows 11, Future of work, Third party applications, Agentic web, Security and access controls.Send Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the Everyday Podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome, the assistant accelerates execution. It seems like Microsoft dropped an entire book of AI updates. Actually, if you look at their book of news that Microsoft usually rolls out around their big conferences,

it was almost 100 pages long. It was 80-some pages, mainly of AI updates announced at Microsoft to build 2025. And if you don't have time to go through and try to understand what's important and what's not, don't worry. I've broken down the five new co-pilot AI updates that I think are going to be very helpful for everyday business leaders. So you don't got to waste your time. Just stick with me for the next, I don't know, 25 or 30 minutes. And I'll tell you what's new, why it's important. and how you're going to use it to transform, potentially, your business. What's going on, y'all?

My name's Jordan Wilson, and I'm the host and welcome to Everyday AI. This is your daily live stream podcast and free daily newsletter, helping us all not just keep up with AI, but how we can use it to get ahead, to grow our companies and our careers. If that's what you're trying to do, you're in the right place. It starts here with the podcast and the live stream, but if you really want to leverage it and put it to work for you, You need to go to our website at your everyday AI.com.

Once there, please go sign up for the free daily newsletter. So not only do we recap the live stream and podcast, so some of the most important takeaways from today's episode in that newsletter, but we also keep you up to date with everything else that's going on in the world of AI, in the business world, right? So that's where you actually go and leverage it. And while you're there on our website, you can go check out by category more than 530.

episodes, go watch the videos, go listen to expert interviews from the world's smartest people in AI, all available for free on our website. All right, if you came here for the AI news, we'll put that the newsletter. All right, we already have enough to cover just with the AI news from the Microsoft Build 2025 conference. And thanks for joining live stream audience. Happy Tuesday to you too, Jackie, joining from LinkedIn, Michelle and Dr. Harvey Castro, Big Bogeyface and crew on the YouTube machine. Thanks for tuning in. I want to know what are you most excited for out of this Microsoft build conference?

Are you using Microsoft 365 copilot? I always like to know what is our audience using? What are you excited for? But I'm going to go ahead and give you the five things right now. Okay. The five new copilot AI updates that I think are the best to save you time. All right. and stick around and I'll explain each one.

But I'm not going to make you wait any longer. So number one, I think it is the GitHub, the GitHub co-pilot coding agent updates. Two, co-pilot tuning for Microsoft 365. Three would be the agent foundry. Four would be multi-agent orchestration inside co-pilot studio. Five would be computer use in copilot. And then I just realized this morning, oh, whoops, I accidentally. I only made it six, not five.

So I got a bonus one. Stick around if you want that one. So there's our top five. The GitHub co-pilot coding agent updates. Copilot tuning for Microsoft 365, the agent foundry for multi-agent orchestration in Copilot Studio. And five, computer use in co-pilots. All right.

Let's get straight into it. All right. So number one, the new updates to the GitHub co-pilot coding agent. So a big shift here. I think GitHub co-pilot has primarily been looked at as a coding assistant. And now it's a straight up autonomous coding partner. All right. So some of the bullet point here for the new update.

So GitHub co-pilot has now evolved into an agentic autonomous AI coding partner. It is embedded directly inside GitHub and it functions inside visual code or sorry, visual studio code as well. It's capable of testing, iterating, refining code and handling routine and specialized development tasks. And you don't just have to talk to it anymore in natural language prompts, right? So it also accepts screenshots and mockups as inputs, as now it is multimodal in its understanding. That one's a big one, right?

So, you know, it's always like, oh, a picture is worth a thousand words. Well, imagine prompting iteratively with an autonomous coding agent that has access to your data with images, right? That's a big step forward, right? Being able to show GitHub co-pilot a product mockup and just being like, yo, go build this for me. Pretty cool there. Also, right now, and this is what Microsoft is saying,

I haven't had, you know, obviously a chance to run all of these through yet because they were just announced a couple of hours ago. So, you know, the other big thing with GitHub co-pilot right now is it's acting more. and Microsoft is trying to turn it into more of a peer programmer that will suggest edits with you, right, which is cool. So whereas before, I think GitHub Copilot was really looked at as a coding assistant, helping with kind of last end coding tasks or, you know, kind of like debugging. So now it's different.

So now Microsoft is really trying to position, you know, GitHub Copilot. I would say more in line with what we've seen. over the last couple of days, really, right? So we've seen rumors, and good thing there's rumors out there on the Google, on the Google side with their Jewels programmer. We've seen rumors of that already floating around. Open AI just released a couple of days ago. Their codex, their codex updates.

So I would say it's more in line with a complete programming suite, right? Now GitHub co-pilot is. So if you remember a couple of months ago, we talked about on the show, Devin, kind of the, you know, which they were early to the game. Devin was in terms of their AI coding assistant. But I think that's where the separation is right now, right? I don't think GitHub, you know, what Microsoft is trying to do with it. They're not trying to turn it into a cursor or a windsurf or a lovable or a bolt where it's kind of like a vibe coder, right? This is an entire coding platform.

And also it logs all actions transparently into GitHub issues for team review. The other big thing, and we're going to talk about this a little more, is also now it supports MCP. So the protocol model context protocol created and made popular by Anthropic. And now everyone is supporting MCP. It's kind of the way for AI agents and just AI systems to talk to each other. Kind of like how the internet has an API. Well, AI has MCP. So, you know, Anthropic really popularized that protocol.

And now GitHub co-pilot supports it as well. So right now, at least the general availability for this begins on June 4th. So that's the other thing, right? And not just talking about GitHub co-pilot, but with Microsoft in general, with Microsoft 365 copilot. But it's a little tricky, right? It's a little tricky to understand when these things are going to be GA or generally available. So right now, general availability is pegged for June 4th of this year for Enterprise customers and ProPlus customers. All right.

Big update number two is copilot tuning. All right. This one's excited. All right. But it's not available for everyone right away. you have to have 5,000 co-pilot licenses or a Microsoft account team contact to take advantage of this. But I like this. So let's talk about what the heck copilot tuning is.

So this is now a version of low code model tuning using your organization's internal data and your workflows. So whereas right now, and I think a lot of people in very early reactions to co-pilot tuning are more talking about, you know, brand voice, right? And, you know, oh, let's let's have responses, you know, use our brand language. And yes, that's technically one thing you can do with co-pilot tuning, but it's about much more than that. It's really about tuning the co-pilot model to better work within the confines that you set, right? And that, you know, fine tuning a model can mean a lot of things. But, you know, to oversimplify it for our everyday business, leaders out there. It's when you essentially train a model and you give it examples of inputs and

outputs. So when you say, hey, when I ask you about quarterly reports, this is what it should look like, right? That's the very oversimplified version, you know, and whether that looks like how the outputs are structured or the tone of voice, the language that's used, where you're pulling that data from. So that's an example of tuning model tuning. So now you'll be able to tune co-pilot as a whole, which is really cool. And it allows businesses to create task-specific agents without writing code. And this is no data science required. So, you know, tuning is fully guided and accessible.

So low code. So what that means is you don't need code. You can use code, right? But you can just do it with natural language, natural language, and via the accessible, fully guided tour. In doing this, you're also able to create custom agents. for different industries or verticals within your business, like consulting, finance, legal, et cetera. Also, all of the data and the agents operate inside Microsoft 365's

Secure Service Boundary. Also, Microsoft does not use this. This is important because this is what people are going to be asking, right? Microsoft says they do not use this customer data to train their foundation model. So a lot of companies, you know, have been hesitant, you know, in years past. to essentially hand over the keys, you know, the keys of the castle, so to speak,

or your company's data, which I've always just been like, y'all, like, that's dumb. Let's be honest. Can I just call a spade a spade here? Companies that have been like, oh, I don't want to, you know, hand over my data to Microsoft or Google. Oh, my gosh. Okay. Well, do you use cloud storage? What's that?

Yes, you do. So it's more or less the same thing. right. But, you know, Microsoft does say when you're using this copilot tuning, they're not using the customer data to train their models. So when is this rolling out? Who gets access? How can you use it? Well, like I said, unfortunately, this isn't for really small, medium-sized businesses. This is more of an enterprise endeavor right now. But I do hope Microsoft scales back these requirements because I think let's just be honest here too, right? Enterprises are slow.

They are slow moving. So for so many bigger corporations to take advantage of this co-pilot tuning, they're too slow. There's way too much red tape. There's way too much bureaucracy. There's way too, you know, it's too slow. Too slow an engine. I would have liked to see Microsoft roll this out to,

maybe organizations with 500 or 1,000 copilot licenses, you know, to put this out to, hey, you got to have 5,000, that's usually just your pretty big enterprise companies, right? Or if you have a Microsoft account team contact. So we'll see. You know, Fred here. Good one, Fred. I like a good joke in the morning.

Fred says that he's only $4,99. licenses short. Yeah, I'm right there like 4,998, Fred. So yeah, I would like to see that world is out to smaller organizations, right? If they want to make this product better, this co-pilot tuning, which again, this is, this can be one of Microsoft's big differentiators, right? Because it's kind of a three and a half horse race, depending on how you look at things in the AI race right now, right? Between Google, Microsoft, OpenAI, and maybe Anthropic on the outside looking in. Meta is obviously trying to break into that conversation as well.

And Microsoft has the clear advantage within Microsoft 365 because Copilot has access to all of your data. And this co-pilot tuning is an extremely powerful piece. However, I don't think there's going to be very many organizations, unfortunately, taking advantage of this. So I think a misstep here for Microsoft. not rolling this out to even medium-sized organizations because this could be a standout killer feature. But I think the few organizations that have those 5,000 co-pilot licenses, for the most part, they're Titanic's.

They're big, they're slow, they're not agile. They're not going to be able to start tuning, even though it is low code. I don't think that they're going to be able to take advantage of this in a way that Microsoft is probably envisioning. be just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI Assistant, now live in the Adobe Firefly app, the all-in-one creative AI studio. Powered by Adobe's Creative Agent, Firefly AI Assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows, drawing on 60-plus pro-grade tools

across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in the driver's as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adopi.com. All right. Number three, update. Agent Foundry. So this is essentially an AI agent playground. So it is powered by Azure, Microsoft Azure, but the AI Foundry agent

service is now generally available for designing, deploying, and scaling, enterprise. grade AI agents. So this is a ton of models. So right now, depending on what you're looking at, anywhere from about 2000 to 11,000 model. So we're talking about different models from, and some of these are new. So GROC as an example is new, you know, XAI CEO, or I don't know if he's technically the CEO anymore, what he calls himself, but Elon Musk joined Microsoft CEO, Saddianadella to talk a little bit about the agent foundry and now having GROC 3 available inside Microsoft's new agent foundry, as well as GROC 3.5, whenever that does release any day now, right? We've been here in any day now for a couple of weeks. So, you know, from those models, proprietary models, you know,

GPT, Hopen AI's models, GPD4-1, you have DeepSeek, you know, Microsoft or sorry, Azure's safe version of deep seek, right? That's worth noting, the one that they've tuned and made safer, entropic models, meta's llama. So, you know, there's literally thousands of different models that you can now use as the base for your agents in the agent foundry. Also, there is a real-time model leaderboard that helps you pick the best performing models for your tasks. So like I said, you can use proprietary or the bring your own models, some more open, source, open weight to build into those workflows. And this also is accessible inside Copilot Studio. So you can have both, you know, high code or pro code tools as well as those

low code offerings as well inside Copilot Studio. This also adds enterprise grade security features, including Entra Agent ID from Microsoft. And this one is pretty cool. And more on this here in a second is it does support multi-agent workflows and also protocols from the bigger players. This is great. I can't believe it took AI for all the big companies to like kumbaya together, right? Because now inside the agent boundary from Microsoft, you have support for Google's A2A or agent-to-agent protocol that allows agents to talk to each other, as well as Anthropics

kind of open model context protocol or MCP. So like I said, how this works and I'll do an dedicated episode

on MCP soon, especially now since there's wide availability and access to the MCP or to the model contact protocol, is this just allows

right, your agents to be able to speak to other platforms outside in this example of Microsoft's agent foundry. And what this means, it's big, right? Yes, it's great that there's now this, you know, this Azure AI foundry with, you know,

you can use thousands of different models, a model leaderboard, very cool, right? But I think just as important as all those other things is the agent to agent or A2A protocol, as well as the MCP, the model protects, sorry, the model context protocol. Joe is saying, good morning, all loving this build coverage. Hope we have a co-pilot specific episode to look forward to in the near future. Well, hey, Joe, you're looking at it right now. And I believe it was you, Joe, yesterday that said you would want to see a Microsoft build or a co-pilot episode. So here we go.

Glad to have you join us live. All right. Next, number four. And this was leading from number three straight into number four, which is huge. So multi-agent orchestration in co-pilot studio. So, you know, what's interestingly enough, I was working with IBM a couple of weeks ago. And we saw a lot of similar updates from IBM inside of their Watson X. So, you know, which I don't know if this is one of those, you know, big tech is just kind of copying each other.

Or if this is just the way that work is flowing, but we'll see. But right now we have the co-pilots multi-agent orchestration inside of co-pilot studio. So now it supports multiple agents working together as a team with native support, like we said, for A to A and MCP to collaborate and divide tasks. That's really cool. So Microsoft says that agents can discover other agents. They can negotiate tasks and collaborate securely with governance as well. So I'm going to be curious on how that all works, right, on how agents are going to decide on how to split up tasks, right, and work within the guardrails that you said. But, I mean, sounds super futuristic and exciting.

We'll see once it's, you know, generally available and once people are using this, how this works. So what can you use this for? I mean, my gosh, this is everything, right? This is what we've been talking about for, you. years on, you know, the future of work is when you can have multiple specialized agents with different models, you know, working with your live data, working with each other, right? So obviously there's, there's huge upside potential here. There's a lot of downside as well, right? When you talk about the old adage or the saying that if you're one percent off, right,

eventually you're going to be completely off course, right? That's when a single human is, is 1% off. Or if you're working with one, you know, large language model, but think when you have multiple agents working together, that 1% off. If it's not perfectly aligned, that is a compounding factor, right? You're going to be much further off base.

So as exciting as multi-agent orchestration inside Copilot Studio is, you have to also understand the huge downsides of this multi-agent orchestration. This is nothing against Microsoft either, right? I'm going to say the same thing for Google and, you know, open AI as well. And this is more of a word of caution to people out there, especially early adopters, you know, the AI leaders out there listening to this podcast. You have to take extreme care when working in multi-agent environment. So I actually, let me go ahead and grab the episode.

And this might be one. I don't often replay episodes, you know, maybe when I'm traveling or if I'm at conferences or, you know, in the rare occurrence where I'm just literally sick and can't talk, right, which isn't very often. Had a great show recently. I'm trying to find it here on our website because, you know, talking about multi-agent environments, like I said, this is the future. This is the future of work.

So let's see. Where was it? There we go. Episode 471. If you want to go listen to that with Babak from Cognizant, the CTO of AI at Cognizant. We talked about it in episode 471, inside multi-agent AI, rethinking enterprise decisions. That one's essential listening, especially as we talk about multi-agent orchestration. Like I said, huge upsides, if you get it right, if you have your data governance right, throw out human in the loop.

I hate that. You need to have expertise. in the loop with the right humans on the oversight as well. But I mean, what you can do with this, you can automate work literally everywhere, right? Your entire business systems, your marketing, HR, IT, right? So some examples here, you know, one agent could draft documents while another agent might create content from those boring text-based documents. Another agent could schedule meetings to talk about said documents, right?

So when you talk about agents getting together and deciding how to break up the task, yeah, the dork in me smiles on what's capable. So the other thing inside this multi-agent orchestration inside co-pilot studio is agentic memory, right, which is huge, being able to remember and have that context from agent to agent. So again, this is from Microsoft. We'll see how this plays out in reality, but also automated validation tools. for compliance and high quality performance. And like we talked about, the multi-agent orchestration,

just like the AI Foundry, does support A2A from Google and the model context protocol from Anthropics. So right now, this is not, we didn't even get a general availability date for this, which is probably a smart move as much as I don't like it, right? And I'd love to just see this unleashed on the world. but right now, this is in private preview right now. All right. Next and maybe last, depending on if you guys want the number six or not, because I made a mistake, computer use in co-pilot.

So right now, this enables agents to automate websites and desktop applications via natural language. So agents can interact like humans, clicking buttons, typing, and navigating interfaces. and this adopts automatically to UI changes in real life tasks. So what can you use this for? Well, really anything that you would do on your Microsoft computer, right? Anything repetitive. I think, you know, going through invoices is going to be a big one. Customer supports data entry, right?

Anything that is Uber repetitive, mundane, and is fairly simple, right? Hey, open this PDF of an invoice. We're going to grab the information from these columns and put them into Excel, right? That's kind of the lowest hanging fruit that I think we'll see with computer use, right? Going across multiple applications on a desktop, doing really just mundane data entry, I think will be one of the first and probably best early on instances of computer use in co-pilot. But also there is that reasoning engine, right? When we talk about Microsoft's ever evolving, right, that's the reality,

they're ever evolving relationship with Open AI, but it does have access to reasoning, reasoning models, so models that can think, right, that can plan ahead, that can, you know, try to understand a task step by step, right? When it does require a little bit of logic or a little bit of planning, I think that's huge. So let's talk about availability. So right now it is only available in. in the Microsoft 365 copilot frontier program for select users. And here's a big one.

So the eligibility on this one is by the number of copilot studio messages that an organization sends. So if you're not in the co-pilot frontier program, if your organization sends 500,000 co-pilot studio messages, then you should be, you should have availability for computer use and it is currently live for qualified enterprise customers under the program. But if you want to see an example of this, if you have co-pilot pro on the web, you can actually go use this right now, right? So I actually just did this. I was playing around with it a little bit yesterday. So let me actually stop sharing my screen. Well, actually, if you want the number six, like I said, I thought this was five.

I'm like, oh, yeah, here's the five things. And then as I read over that book of news a couple of times, I'm like, wait, number six is actually just as big. So if you want that number six, just go ahead and comment for our live stream audience. Just comment more or I can wrap it up. But as you guys decide, if you want to see one more or not, I will show you all an example of this live. We'll see how this goes. There we go. So I think it should be on my screen for our live stream audience.

So yeah, if you're using copilot's, the paid version on the web, which I know not a lot of people are, but let me just say this. Low key, it's really improved over the last couple of months, right? So this is not if you're using kind of the 365 version of co-pilot, right, for your business. This is if you're using copilot pro, very different. So this is the, you know, there's a free version, but also a $20 a month subscription. But if you're using a work account in your work is, you know, has Microsoft 365, you're not going to see this similar interface. But if it doesn't or if you're using a personal account or a co-pilot pro, you will see this.

So it is more of a chat, GBT-esque or Gemini-esque layout. But if you go into co-pilot and you click on action, okay, so this, this is, is actually the computer using agent is live. I didn't even know this until I'm trying to think the first time I discovered it. I know that they had previewed this in their wave two announcements. Right. So this is the official unveiling at Microsoft build of wave of wave two. But they did announce the wave two announcements in April and the actions were coming soon inside co-pilot. But it's actually live right now in Copilot Pro, but this is essentially a computer-using agent. And it's a little tricky because by default, it says, you know, actions are in preview and can make mistakes.

Please monitor their work closely. And, you know, it has these like pre-built kind of computer-using automations. Similarly, and this is like Open AIs operator, right, which is a big deal. But it's just called actions inside Copilot Pro where it says hand off every day, And it's part of their labs. So you have to sign up for labs. But at first, I'm like, okay, it's just showing like things to book, right? Like or order flowers or book a trip, you know, via Expedia.

And I'm like, I don't want to do any of those things. So it's a little unintuitive, but you don't actually need to use those pre-built actions. It can do anything. So probably a mistake here on Microsoft's part for this kind of computer using actions to not make it a little more clear. but you don't even need to do that. So I could just say something and say like, you know, please browse your everyday AI.com and find the latest five episodes on the episode page. Okay, so I can click go on that.

And again, for our podcast audience, I am just using the $20 a month co-pilot pro. But you'll see here, I mean, it's a live computer using agent. which right now, you know, you have to, I mean, operator from Open AI is the only other mainstream one that's at least easy to use. Anthropic has their computer using agent, but it's not web-based and it's actually extremely clunky and very hard to use. You have to be a little technical. But here we go. I mean, it's a little slow, right?

But that's okay. But you can see now I have a literal computer using agent. It's going to your everyday. A.I.com and I can kind of see on the right hand side what's happening. And then on the left hand side or the other part of the panel, I see in this virtual environment, the computer using agents, going. So right now it says it's navigating to episodes section. So it notices that at the top of the screen, there's a big kind of button there that says episode. So it's going and it's clicking there. What's funny here is it blurs out all the faces from.

the, from our website, which is fine. Right. So at any time, I can also take control or anything like that. But that's just kind of like an extra one there for people that maybe don't want to wait for access. Maybe you already have access to this computer using agent right now. And it already responded accurately. By the way, it did give me the latest five episodes by going to my website, navigating to the episodes page. So pretty cool there. All right. So let's see. Did anyone want number six? All right. Marie said it. Joe said it. Fred said it. All right. A couple of people wanted it. So here we go.

Number six, I kind of already mentioned it in other aspects. But here's why this is maybe just as big as some of these other updates that we already talked about. So MCP native support. So here's why this is big. So it's now model. context protocol for agent access to tool system and enterprise data. Like I said, the MCP protocol was designed by Anthropic. But here's why it's big. It's not just in the AI foundry. It's not just in copilot studio with the multi-agents. It is literally supported almost everywhere, including GitHub co-pilot, Azure dynamics,

co-pilot studio, Microsoft 365 apps, and even with. Windows 11. That's why this is huge. MCP is supported by default inside of Windows. Native MCP support. And I can't emphasize enough how big that is, right, for the future of multi-system agentic AI. Right.

So what that means in theory in the future, right, third parties could access your Windows system because it has MCP built in by default. So this uses HTTP-based communication between agents and enterprise tools. It also does add security with signed code, scope permissions, and embedded registries. So you can actually see what's happening within the MCP protocol that is available natively now inside Windows. This might be a slower rollout. So Microsoft said there will be a development.

preview of the secure MCP in Windows after the Build 2025 conference. They said that future updates will enforce default security and access controls. And also, this does position MCP as the infrastructure layer for the agent-driven internet. I mean, the fact that I was not expecting this one at all, the fact that Microsoft is now supporting MCP natively inside Windows 11. That's big for the future of work. What this means is in the future, right? It is going to be so much easier to work with other third party applications

that also support the MCP protocol. So think of like how an API allows two unrelated internet websites to talk to each other, right? MCP essentially does that for agents. And there's other protocols. Microsoft even came out with their version of their own protocol as well. And then you have the, you have the Google version with A2A. I'm trying to remember what Microsoft's version. I think it was NIL web is what it was.

Yeah. So NIL web is kind of Microsoft's version of this. But what I like here is Microsoft didn't just say, okay, we're using NIA web and nothing else. Again, very exciting that MCP will hopefully, right, this is, there's no general availability date yet. So this might be one of those. You might have to wait a couple of quarters. But this does signal that Microsoft is really open to working with the rest of the agentic web. and other AI players.

So think of this as like speaking a language, right? And six months ago, for the most part, you had all of these AI tools right before Anthropic created the MCP and it obviously popularized very quickly, right? I can't remember, you know, aside from Gen A.I. and large language models, any time a protocol like this was so widely and quickly adopted across huge enterprise organizations, right?

We've seen, you know, Google pick up MCP. We've seen OpenAI pick it up. Now we see Microsoft making it native. So that's just enormous, right? So in the future, this does open up a ton of possibilities inside, you know, Windows 11 and probably Windows 12 and beyond, you know, in terms of Microsoft opening up your operating system to other large language models that support MCP as well as just the agentic web, right?

This is really big for the future of work. And I'm excited to see when this does roll out. Yeah, I would guess this is going to be, this is going to be a while because the security thing is huge. I have already seen a lot of, you know, dark web MCP type. kind of protocols pop up, right? So people can build these MCP protocols and you can incorporate them, you know, in various AI applications.

And like anything else, there's already been some bad ones out there that are, you know, trying to steal data or do some malicious things. So, you know, it obviously makes sense that Microsoft is going to have a slower rollout of this because it is an extremely popular and very robust way to work. work in the future. All right. I hope that was helpful. For those of you that said, oh, like, what else? Can we do more co-pilot stuff? Yeah. Well, recently, if you want, go listen to episode 492. So that's when we had VP of AI agents, Ray Smith, from Microsoft on the show to talk about what's new inside Microsoft co-pilot studio. So I talked about Microsoft co-pilot studio a little bit

here. So go listen to episode 492 if you want to know more about agents. inside of Copilot Studio, as well as I did also show you a little bit of kind of the web version of Microsoft Copilot. But a couple of months ago, Microsoft really made a lot of things free inside Copilot on the web. So I went over that in episode 479. So if you want to know more about, you know, how that's changed. Go check that out as well. All right. I think that's a wrap, y'all.

All right. So I hope this was helpful as we went over the five, well, actually six different co-pilot AI updates, how to use them, how I think that they're going to potentially change the future of work for organizations that are Windows or PC based. And this is big. And don't worry, we're going to be doing the exact same thing tomorrow for Google because within a couple of hours, Google is going to. to kick off their I.O. conference. It is AI week. We're going to have some probably smaller announcements from Anthropic Clawed later in the week as well. There's been some rumors that Open AI may release something as well. So we'll obviously keep you in the loop. So I hope this was helpful. If so, don't be greedy. Don't keep everyday AI as your little secrets. All right, please, if this is helpful, repost this on social media. Share this with someone. You're going to be the smartest person in AI when you share this with someone. But also, if you're listening on the podcast, we appreciate your support.

If you could, click follow or subscribe to the show on Spotify or Apple Podcasts. And if you could, leave us a review or rating and super appreciate that. We do a lot of work to stay up to date and to keep you up to date as well. So it only takes you a couple of seconds to show the love back. Thank you for tuning in. Make sure if you haven't, go to your everyday AI.com. Sign up for that free daily newsletter. I'll see you tomorrow.

for the Google I.O. Recap, and I'll see you every day for more everyday AI. Thanks, y'all. Meet Firefly AI Assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution, Stay in control with the ability to step in and refine at any time. See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI.

Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.