Microsoft Integrates Model Context Protocol into Windows, Paving Way for AI Agent Revolution
Microsoft has taken a major step forward in its vision of an AI-first future for Windows, officially announcing native support for the Model Context Protocol (MCP). Often dubbed the 'USB-C of AI apps,' MCP is poised to revolutionise how intelligent agents interact with applications and services on the Windows platform.
The move was unveiled alongside the launch of the Windows AI Foundry, an initiative designed to accelerate the development of AI-powered features and ecosystems within Windows. Together, these efforts mark a deeper commitment to transforming Windows into a platform where autonomous AI agents can seamlessly assist users with a broad range of tasks.
MCP, introduced by Anthropic in late 2023, is an open-source communication standard that enables AI apps to connect with each other and external services—much like how USB-C connects different hardware devices. By embracing MCP, Microsoft aims to allow developers to build agents that can interact directly with both web services and Windows system functions.
'We want Windows as a platform to be able to evolve to a place where we think agents are a part of the workload on the operating system, and agents are a part of how customers interact with their apps and devices on an ongoing basis,' said Pavan Davuluri, head of Windows, in an interview with The Verge.
The idea is simple but powerful: equip AI agents with the tools and protocols to interact intelligently with different components of the operating system. To support this, Microsoft is introducing new developer tools and an MCP registry that acts as a trusted directory of MCP servers. These servers will enable agents to tap into core Windows functionalities such as the file system, the Windows Subsystem for Linux, and window management tools.
In practice, this means AI assistants on Windows could soon go beyond traditional limitations. For example, during a private demonstration, Microsoft showcased how the AI assistant Perplexity could use MCP to access and query a user's files. Instead of manually selecting folders or uploading documents, users could simply ask, 'Find all the files related to my vacation in my documents folder,' and the agent would handle the task seamlessly.
This level of intelligent interaction could extend across the operating system, from simplifying workflows in Excel to streamlining system settings. Microsoft is also preparing Copilot Plus PCs to include an AI agent interface, allowing users to change settings using plain language commands.
However, Microsoft is not overlooking the potential security risks. Experts have flagged possible vulnerabilities with MCP, including token theft, server compromise, and prompt injection attacks. Acknowledging these concerns, Microsoft is limiting access to the initial rollout and making the preview version available only to select developers.
'I think we have a solid set of foundations and more importantly a solid architecture that gives us all the tools to start, to do this securely,' said David Weston, vice president of enterprise and OS security at Microsoft. 'We're going to put security first, and ultimately we're considering large language models as untrusted, as they can be trained on untrusted data and they can have cross-prompt injection.'
Despite these challenges, Microsoft's strategy reflects a strong belief in the future of "agentic computing"—a vision where AI agents are as central to digital interactions as the apps themselves. The company sees MCP as a foundational layer for this future, helping create a standardised, secure, and scalable framework for agents to thrive on Windows.
With the successful integration of MCP and the establishment of the Windows AI Foundry, Microsoft has sent a clear signal: Windows is not just ready for AI—it's being reimagined around it.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Hindustan Times
10 hours ago
- Hindustan Times
Thinking AI models collapse in face of complex problems, Apple researchers find
Just days ahead of the much-anticipated Worldwide Developer Conference (WWDC), Apple has released a study titled 'The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity', which saw researchers testing 'reasoning'; AI models such as Anthropic's Claude, OpenAI's o models, DeepSeek R1 and Google's Thinking models to see how far they can scale to replicate human reasoning. Spoiler alert — not as much, as the entire AI marketing pitch, would have you believe. Could this signal what may be in store for Apple's AI conversation ahead of the keynote? The study questions the current standard evaluation of Large Reasoning Models (LRMs) using established mathematical and coding benchmarks, arguing they suffer from data contamination and don't reveal insights into reasoning trace structure and quality. Instead, it proposes a controlled experimental testbed using algorithmic puzzle environments. The limitations of AI benchmarking, and need to evolve, is something we had written about earlier. 'We show that state-of-the-art LRMs (e.g., o3-mini, DeepSeek-R1, Claude-3.7-Sonnet-Thinking) still fail to develop generalizable problem-solving capabilities, with accuracy ultimately collapsing to zero beyond certain complexities across different environments,' the researcher paper points out. These findings are a stark warning to the industry — current LLMs are far from general-purpose reasoners. The emergence of Large Reasoning Models (LRMs), such as OpenAI's o1/o3, DeepSeek-R1, Claude 3.7 Sonnet Thinking, and Gemini Thinking, has been hailed as a significant advancement, potentially marking steps toward more general artificial intelligence. These models characteristically generate responses following detailed 'thinking processes', such as a long Chain-of-Thought sequence, before providing a final answer. While they have shown promising results on various reasoning benchmarks, the capability of benchmarks to judge rapidly evolving models, itself is in doubt. The researchers cite a comparison between non-thinking LLMs and their 'thinking' evolution. 'At low complexity, non-thinking models are more accurate and token-efficient. As complexity increases, reasoning models outperform but require more tokens—until both collapse beyond a critical threshold, with shorter traces,' they say. The illustrative example of the Claude 3.7 Sonnet and Claude 3.7 Sonnet Thinking illustrates how both models retain accuracy till complexity level three, after which the standard LLM sees a significant drop, something the thinking model too suffers from, a couple of levels later. At the same time, the thinking model is using significantly more tokens. This research attempted to challenge prevailing evaluation paradigms, which often rely on established mathematical and coding benchmarks, which are otherwise susceptible to data contamination. Such benchmarks also primarily focus on final answer accuracy, providing limited insight into the reasoning process itself, something that is the key differentiator for a 'thinking' model compared with a simpler large language model. To address these gaps, the study utilises controllable puzzle environments — Tower of Hanoi, Checker Jumping, River Crossing, and Blocks World — and these puzzles allow for precise manipulation of problem complexity while maintaining consistent logical structures and rules that must be explicitly followed. That structure theoretically opens a window, a glance at how these models attempt to 'think'. The findings from this controlled experimental setup reveal significant limitations in current frontier LRMs. One of the most striking observations is the complete accuracy collapse that occurs beyond certain complexity thresholds across all tested reasoning models. This is not a gradual degradation but a sharp drop to near-zero accuracy as problems become sufficiently difficult. 'The state-of-the-art LRMs (e.g., o3-mini, DeepSeek-R1, Claude-3.7-Sonnet-Thinking) still fail to develop generalizable problem-solving capabilities, with accuracy ultimately collapsing to zero beyond certain complexities across different environments,' note the researchers. These results inevitably challenge any notion that the LRMs truly possess generalisation problem-solving skills, required for planning tasks or multi-step processes. The study also identifies a counter-intuitive scaling limit in the models' reasoning effort (this is measured by the inference token usage during the 'thinking' phase), which sees these models initially spend more tokens, but as complexity increases, they actually reduce reasoning effort closer to the inevitable accuracy collapse. Researchers say that 'despite these claims and performance advancements, the fundamental benefits and limitations of LRMs remain insufficiently understood. Critical questions still persist: Are these models capable of generalizable reasoning, or are they leveraging different forms of pattern matching?,' they ask. There are further questions pertaining to performance scaling with increasing problem complexity, comparisons to the non-thinking standard LLM counterparts when provided with the same inference token compute, and around inherent limitations of current reasoning approaches, as well as improvements that might be necessary to advance toward more robust reasoning. Where do we go from here? The researchers make it clear that their test methodology too has limitations. 'While our puzzle environments enable controlled experimentation with fine-grained control over problem complexity, they represent a narrow slice of reasoning tasks and may not capture the diversity of real-world or knowledge intensive reasoning problems,' they say. They do add that the use of 'deterministic puzzle simulators assumes that reasoning can be perfectly validated' at every step, a validation that may not be feasible to such precision in less structured domains. That they say, would restrict validity of analysis to more reasoning. There is little argument that LRMs represent progress, particularly for the relevance of AI. Yet, this study highlights that not all reasoning models are capable of robust, generalisable reasoning, particularly in the face of increasing complexity. These findings, ahead of WWDC 2025, and from Apple's own researchers, may suggest that any AI reasoning announcements will likely be pragmatic. The focus areas could include specific use cases where current AI methodology is reliable (the research paper indicates lower to medium complexity, less reliance on flawless long-sequence execution) and potentially integrating neural models with traditional computing approaches to handle the complexities where LRMs currently fail. The era of Large Reasoning Models is here, but this 'Illusion of thinking' study is that AI with true reasoning, remains a mirage.
Time of India
12 hours ago
- Time of India
For some recent graduates, the AI job apocalypse may already be here
HighlightsUnemployment for recent college graduates has risen to 5.8%, with a notable increase in job displacement due to advancements in artificial intelligence, particularly in technical fields like finance and computer science. Many companies are adopting an 'AI-first' approach, with some executives reporting a halt in hiring for lower-level positions as artificial intelligence tools can now perform tasks that previously required human employees. Dario Amodei, Chief Executive Officer of Anthropic, has predicted that artificial intelligence could eliminate half of all entry-level white-collar jobs within the next five years. This month, millions of young people will graduate from college and look for work in industries that have little use for their skills, view them as expensive and expendable, and are rapidly phasing out their jobs in favour of artificial is the troubling conclusion of my conversations over the past several months with economists, corporate executives and young job seekers, many of whom pointed to an emerging crisis for entry-level workers that appears to be fuelled, at least in part, by rapid advances in AI capabilities. You can see hints of this in the economic data. Unemployment for recent college graduates has jumped to an unusually high 5.8% in recent months, and the Federal Reserve Bank of New York recently warned that the employment situation for these workers had 'deteriorated noticeably.' Oxford Economics, a research firm that studies labour markets, found that unemployment for recent graduates was heavily concentrated in technical fields like finance and computer science, where AI has made faster gains. 'There are signs that entry-level positions are being displaced by artificial intelligence at higher rates,' the firm wrote in a recent report. But I'm convinced that what's showing up in the economic data is only the tip of the iceberg. In interview after interview, I'm hearing that firms are making rapid progress toward automating entry-level work and that AI companies are racing to build 'virtual workers' that can replace junior employees at a fraction of the cost. Corporate attitudes toward automation are changing, too — some firms have encouraged managers to become 'AI-first,' testing whether a given task can be done by AI before hiring a human to do it. One tech executive recently told me his company had stopped hiring anything below an L5 software engineer — a mid-level title typically given to programmers with three to seven years of experience — because lower-level tasks could now be done by AI coding tools. Another told me that his startup now employed a single data scientist to do the kinds of tasks that required a team of 75 people at his previous company. Anecdotes like these don't add up to mass joblessness, of course. Most economists believe there are multiple factors behind the rise in unemployment for college graduates, including a hiring slowdown by big tech companies and broader uncertainty about President Donald Trump's economic policies. But among people who pay close attention to what's happening in AI, alarms are starting to go off. 'This is something I'm hearing about left and right,' said Molly Kinder, a fellow at the Brookings Institution, a public policy think tank, who studies the impact of AI on workers. 'Employers are saying, 'These tools are so good that I no longer need marketing analysts, finance analysts and research assistants.'' Using AI to automate white-collar jobs has been a dream among executives for years. (I heard them fantasising about it in Davos back in 2019.) But until recently, the technology simply wasn't good enough. You could use AI to automate some routine back-office tasks — and many companies did — but when it came to the more complex and technical parts of many jobs, AI couldn't hold a candle to humans. That is starting to change, especially in fields, such as software engineering, where there are clear markers of success and failure. (Such as: Does the code work or not?) In these fields, AI systems can be trained using a trial-and-error process known as reinforcement learning to perform complex sequences of actions on their own. Eventually, they can become competent at carrying out tasks that would take human workers hours or days to complete. This approach was on display last week at an event held by Anthropic, the AI company that makes the Claude chatbot. The company claims that its most powerful model, Claude Opus 4, can now code for 'several hours' without stopping — a tantalising possibility if you're a company accustomed to paying six-figure engineer salaries for that kind of productivity. AI companies are starting with software engineering and other technical fields because that's where the low-hanging fruit is. (And, perhaps, because that's where their own labour costs are highest.) But these companies believe the same techniques will soon be used to automate work in dozens of occupations, ranging from consulting to finance to marketing. Dario Amodei, Anthropic's CEO, recently predicted that AI could eliminate half of all entry-level white-collar jobs within five years. That timeline could be wildly off, if firms outside tech adopt AI more slowly than many Silicon Valley companies have, or if it's harder than expected to automate jobs in more creative and open-ended occupations where training data is scarce.
Time of India
16 hours ago
- Time of India
Microsoft CEO Satya Nadella to Computer Science students: All of us are going to be more ...
Microsoft CEO Satya Nadella Microsoft CEO Satya Nadella has shared an advice for aspiring tech professionals. The CEO recently appeared in a conversation with tech YouTuber Sajjaad Khade where he urged students to focus on building strong fundamentals in computational thinking , even as AI shapes the software industry. He stressed that despite AI's growing role in coding, the ability to break down problems logically and design systematic solutions remains essential. However, he warned that even with AI, success still depends on giving clear, structured instructions—a skill that blends technical knowledge with systems thinking. 'The path to being that software architect gets speeded up,' Nadella said, adding that soon, 'All of us are going to be more software architects.' Getting fundamentals of software is important: Satyla Nadella During the conversation , Khade asked Nadella: 'In this world of AI, if I'm a beginner, just want to break into tech, what's your number one piece of advice?' by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Eat 1 Teaspoon Every Night, See What Happens A Week Later [Video] getfittoday Undo Nadella replied saying: 'Just getting real fundamentals of software (if you're a software engineer), I think matters a lot.' Adding further, he said 'To me, having the ability to think computationally (is important).' Nadella explained using his own example saying he was able to fix a bug by just assigning it to the CoPilot coding agent. 'Except I was thinking about it, it was a pretty cool issue, right? The issue was I did a filter, which was basically a percentile.. Creating a feature. But then I said, "Oh man, this is, like, you know, I could, you know, recount what is a SQL, right?' When Satya Nadella revealed that up to 30% of Microsoft's code is now written by AI During a conversation with Meta CEO Mark Zuckerberg earlier this year, Microsoft CEO Satya Nadella revealed that AI now writes up to 30% of Microsoft's code base. 'I'd say maybe 20%, 30% of the code that is inside of our repos today and some of our projects are probably all written by software,' Nadella then said. What to Expect at WWDC 2025: iOS 19, AI, M4 Macs & More
