Microsoft goes full agentic at Build 2025 with new AI tools, from Copilot to Discovery
'Today, at Build, we showed you how we are building the open agentic web. It is reshaping every layer of the stack, and our goal is to help every dev build apps and agents that empower people and orgs everywhere,' Nadella said in his post on X. Here is a list of some of the biggest announcements from this year's Microsoft Build.
'We are taking GitHub Copilot from being a pair programmer to peer programmer. You now have a full coding agent built right into GitHub. You can assign it issues – whether it's bug fixes, new features, or ongoing code maintenance. And it will complete these tasks autonomously,' said Nadella.
GitHub Copilot has unveiled a new AI-powered coding agent that has been designed to automate regular development tasks and augment productivity, enabling developers to prioritise complex work. The agent can be assigned tasks through GitHub issues or Copilot Chat in Visual Studio Code. Once assigned, the agent begins a secure development environment using GitHub Actions, clones the repository, and analyses the codebase. Throughout this process, the agent commits changes to a draft pull request and offers session logs for transparency.
It is capable of managing tasks such as bug fixes, feature additions, code refactoring, and documentation improvements, especially in well-tested codebases. It also integrates context from related discussions and complies with custom repository instructions to align with project standards. Also, it has security measures in place that ensure that the agent's contributions are reviewed carefully before they are integrated. It is currently available to Copilot Enterprise and Copilot Pro+ users across multiple platforms, including GitHub's website, mobile app, and command-line interface.
According to Nadella, Copilot can now learn your company's unique tone and language. 'It is all about taking that expertise you have as a firm and further amplifying it so everyone has access.'
Yet another major announcement was the introduction of Microsoft 365 Copilot Tuning, which is a low-code feature within Copilot Studio. This feature allows companies to customise AI models with their own data, process, and workflows. With this, now businesses will be able to create domain-specific AI agents without requiring any expertise in coding. Simply put, a legal firm can develop an agent that reflects its unique form and expertise, automating a range of tasks combining institutional knowledge with context specific to each client.
The tech giant claims that all data remains within Microsoft's secure 365 service boundary, ensuring privacy and compliance. Besides, the company also introduced multi-agent orchestration, a feature that allows multiple AI agents to work together and distribute tasks among themselves. This allows AI agents to manage complex, cross-functional tasks with ease. Essentially meaning, during an employee onboarding, AI agents from IT, HR, and marketing teams can work together to streamline the entire process.
This year's Build conference also saw significant enhancements to Azure AI Foundry, Microsoft's comprehensive platform for developing and deploying AI applications and agents. The platform now supports more than 10,000 models, including open-source, proprietary, and even task-specific models. It is allowing developers to select and fine-tune models based on their business needs. The Azure AI Foundry has expanded its model offerings to include Grok 3 and Grok 3 Mini.
'Foundry is the complete app platform for building apps and agents. We are adding support for more models from Grok, Hugging Face, Meta, Mistral, and more. Plus: agentic retrieval in Azure AI Search, Foundry Agent Service, integration with Copilot Studio, and more. And we are ensuring the tools you already use for identity, management, and security will now all extend to agents too,' Nadella said.
Some of the key features of Azure AI Foundry include the Azure AI Agent Service, allowing the creation of agents that are capable of automating complex workflows and tasks.
Microsoft also introduced NLWeb, which is an open-source project designed to simplify the integration of AI interfaces into websites. This new open project essentially allows developers to use natural language to interact with any website. Nadella likened it to HTML for the agentic web. Developers can now add a chat interface powered by their chosen AI model and proprietary data with just a few lines of code. This will also allow users to interact with a site's content using queries in natural language.
On a deeper level the NLWeb operates as a Model Context Protocol (MCP) server, allowing websites to make their content available to AI agents within the MCP ecosystem. NLWeb enhances users' experiences through AI-driven interactions by using existing structured data formats like Schema.org and RSS. This project aligns with the tech giant's broader vision for an open agentic web where decentralised AI agents can perform tasks on behalf of users.
'We're bringing together the full tech stack to help speed up science itself. Discovery uses agents to generate ideas, simulate results, and learn. A great example is this promising candidate for a coolant that doesn't rely on forever chemicals,' said Nadella. Microsoft Discovery was one of the prominent announcements at the conference. It is a new AI-powered platform developed to transform research and development processes.
Microsoft Discovery uses agentic AI to automate complex tasks, allowing scientists and engineers to speed up innovation by streamlining data analysis and experimentation. By integrating with existing tools and datasets, Discovery essentially allows for seamless collaboration and knowledge sharing across teams.
These advancements are part of Microsoft's broader strategy to enhance AI integration in the workplace, encouraging companies to build tailored AI solutions and also supporting scientific breakthroughs through advanced AI technologies.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Economic Times
an hour ago
- Economic Times
US computer science degrees from top universities are leaving graduates jobless: Why is top coding education no longer enough?
Synopsis The once-reliable path to high-paying tech jobs via a computer science degree is facing challenges. AI advancements and industry layoffs contribute to rising unemployment among recent graduates. Experts advise focusing on specialized skills and a deeper understanding of physical sciences, as the demand shifts from basic coding to problem-solving and innovation. iStock A computer science degree no longer guarantees a high-paying job in America. The rise of Artificial Intelligence and tech layoffs contribute to this shift. Many graduates struggle to find employment despite booming enrollments. Experts advise focusing on specialized skills and problem-solving. (Image: iStock) The long-standing belief that earning a computer science degree from a top American university guarantees a high-paying job is facing growing challenges. Despite booming enrollments and soaring expectations, many recent graduates are struggling to find employment in a rapidly evolving tech landscape. According to a report from The New York Times, a combination of shifting industry demands and the rise of artificial intelligence is leaving new computer science graduates unemployed and frustrated. For over a decade, students like Manasi Mishra were inspired by industry leaders who touted computer science as the ultimate pathway to success. Growing up near Silicon Valley, Manasi absorbed the message loud and clear: 'If you just learned to code, work hard and get a computer science degree, you can get six figures for your starting salary,' she recalls. This promise pushed her to start coding websites as a child, take advanced computing courses in high school, and finally major in computer science at Purdue University. Technology giants reinforced these expectations. In 2012, Brad Smith, then a top Microsoft executive, highlighted that computer science graduates typically earned starting salaries above $100,000, with generous bonuses and stock grants. This promise fueled an education boom. According to the Computing Research Association, the number of undergraduate computer science majors in the US more than doubled from 2014 to over 170,000 last year. However, the job market has shifted drastically. The introduction of AI programming tools capable of generating and debugging code has reduced the need for junior software engineers. Simultaneously, major tech companies such as Amazon, Intel, Meta, and Microsoft have executed significant layoffs. The Federal Reserve Bank of New York reports that unemployment rates for recent college graduates in computer science and computer engineering stand at 6.1 percent and 7.5 percent respectively. These figures far exceed unemployment among recent graduates in fields like biology or art history, where the rate is around 3 percent. Many graduates face a grueling job search. Students from a range of institutions, including Maryland, Texas, Cornell, and Stanford, reported applying to hundreds or even thousands of jobs. Yet, months of effort often ended with no interviews or offers. The job search process has become emotionally taxing, with some graduates describing it as 'bleak,' 'disheartening,' or 'soul-crushing.' Several candidates said they felt 'gaslit' by the industry's earlier promises of easy success. A key factor behind the bleak job prospects is AI's increasing role in software development. Entry-level coding jobs, which traditionally served as a stepping stone for new graduates, are now most vulnerable to automation. Tools like CodeRabbit promise to debug code faster than human engineers, causing firms to rethink their hiring strategies. Graduates also face an AI 'doom loop' during applications. Many use AI tools to tailor résumés and autofill applications quickly, while employers employ AI-driven systems to automatically filter and reject candidates, removing human judgment from the hiring process. In this changing landscape, experts urge students to rethink their approach. Sameer Samat, Google's Android chief, highlights that a computer science degree alone no longer guarantees success. 'If all you want to do is learn Java or Python, you don't need a degree,' he said. Instead, passion, deep expertise, and problem-solving skills are what set candidates apart. Samat advises aspiring engineers to focus on becoming top experts in a niche they care about, whether that be system design, AI, or user experience. This specialized knowledge increasingly drives hiring decisions at major tech firms. Industry leaders like Nvidia CEO Jensen Huang suggest that the future of technology will demand a deeper understanding of physical sciences rather than just software coding. Huang describes a shift toward 'Reasoning AI' and 'Physical AI,' where machines will need to comprehend real-world physics such as forces, friction, and inertia. This perspective aligns with voices like Telegram's Pavel Durov and Elon Musk, who emphasize the importance of mathematics and physics as foundational tools for future innovation. Adding to the uncertainty is the financial burden many students face. Zoho founder Sridhar Vembu recently cautioned against taking on large education loans for foreign degrees, especially as job opportunities become scarce. He advocates for employer-funded training programs and broader acceptance of alternative credentials, arguing this approach could prevent students from being trapped in debt. The era when a computer science degree from a reputed US university was a sure ticket to success is coming to an end. With AI reshaping the industry and job markets contracting, graduates must adapt by developing specialized skills, embracing lifelong learning, and being open to alternative career pathways. For students and educators alike, the focus must shift from coding syntax to deep problem-solving and understanding the interplay between technology and the physical world.
Mint
an hour ago
- Mint
GitHub chief executive Thomas Dohmke to step down, cites plans to become a founder
Thomas Dohmke, the CEO of Microsoft-owned software development platform GitHub, has announced his decision to step down after nearly four years at the helm. Dohmke revealed on Monday that he plans to return to his entrepreneurial roots by becoming a founder once again. In a heartfelt message posted on X, Dohmke expressed gratitude towards GitHub employees, whom he affectionately calls 'Hubbers', describing his time with the company as 'the ride of a lifetime.' Reflecting on his journey, Dohmke shared that over a decade ago, he and his family moved from Germany to the United States following the sale of his startup to Microsoft. Since then, he has collaborated with a diverse community including GitHub staff, Microsoft employees, customers, partners, and open-source developers worldwide. He highlighted his progression from building mobile developer tools to leading GitHub through significant milestones such as its acquisition and the launch of AI-powered initiatives like Copilot. Despite his accomplishments, Dohmke said his passion for startups has drawn him back. He will remain at GitHub through the end of 2025 to support a smooth leadership transition. He expressed pride in the company's evolution as a remote-first organisation with a global footprint. Under his leadership, GitHub has strengthened its position with over one billion repositories and forks and a developer base exceeding 150 million. The platform has expanded its reach internationally and earned key certifications, while GitHub Advanced Security has revolutionised vulnerability management through AI, significantly reducing remediation times. Additionally, GitHub Actions has solidified its role as a leading continuous integration solution, processing three billion minutes monthly, a 64 per cent increase year-over-year. Dohmke also emphasised the transformative impact of GitHub Copilot, which has evolved from a simple autocompletion tool into a comprehensive AI coding assistant with over 20 million users. This innovation has been the largest shift in software development since the personal computer, he said. GitHub Copilot's advancements include partnerships with major AI firms and new features such as synchronous and asynchronous coding agents.
Time of India
3 hours ago
- Time of India
GitHub CEO Thomas Dohmke resigns: ‘I've decided to leave……to become..'
GitHub CEO Thomas Dohmke has resigned from his position, triggering a major restructuring for the company. After nearly four years at the helm, Dohmke is stepping down signaling his departure from both GitHub and Microsoft. Following his resignation, Microsoft is moving GitHub directly into its new CoreAI team. Reportedly, Microsoft will not replace Dohmke's CEO role, effectively ending GitHub's status as a separate entity and integrating it fully into its operations. 'Still, after all this time, my startup roots have begun tugging on me and I've decided to leave GitHub to become a founder again,' Dohmke said. 'I'll be staying through the end of 2025 to help guide the transition and am leaving with a deep sense of pride in everything we've built as a remote-first organization spread around the world,' he added. GitHub to merge in Microsoft AI engineering team led by Jay Parikh Dohmke mentioned that GitHub will now be a part of Microsoft's CoreAI team – a new engineering group led by former Meta executive Jay Parikh. It is focused on building AI platforms and tools for both internal use and for Microsoft's customers. The division includes Microsoft's platform, tools, and Dev Div teams. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Sheila révèle son compagnon : un visage bien connu ! Weight Loss Groove Undo According to Dohmke, GitHub Copilot 'is the leader of the most successful and thriving market in the age of AI, with over 20 million users and counting.' 'In just the last year, GitHub Copilot became the first multi-model solution at Microsoft, in partnership with Anthropic, Google, and OpenAI. We enabled Copilot Free for millions and introduced the synchronous agent mode in VS Code as well as the asynchronous coding agent native to GitHub,' he added. In a post on X (formerly Twitter), Dohmke thanked Satya Nadella 'for the ride of a lifetime'. 'Thank you to @satyanadella, Julia Liuson, so many countless people, and most importantly, thank you to all Hubbers for the ride of a lifetime,' he said. Lava Blaze Dragon 5G: Clean, Simple and Surprising AI Masterclass for Students. Upskill Young Ones Today!– Join Now
