In this Q&A, Michelle introduces M12, considers what kinds of AI-powered solutions will drive the next wave of AI innovation, highlights some of M12’s recent investments, and explains how Microsoft Marketplace can help startups reach enterprise customers.

What is M12, and how does it differ from traditional venture capital firms? 

M12 is Microsoft’s venture fund. We’re an early-stage fund focused on finding innovative AI startups and technologies that can leverage the power of Microsoft. And with that, we’re looking at the big picture—not just the companies that are winning, but those that are driving the systems of the future.

In today’s venture environment, there’s a lot of capital fighting to invest in the top AI founders and startups, so investors need something extra to stand out and win deals.

At M12, we’re backed by the power of Microsoft, which means a few things: instant credibility with potential partners, as well as access to its ecosystem,  expertise and research, and global go-to-market (GTM) motion. But most importantly, the team is built for impact. We help ensure that our startups have a custom plan and a dedicated resource to help make that plan a reality.  

Our team is a unique mix of founders, researchers, and folks who have been in the trenches across product innovation, finance, go to market, and business development. While our skill sets may be diverse, we put our collective expertise toward a common goal: to help promising companies win.  

What kinds of AI-powered solutions or business models do you believe will define the next wave of innovation? 

The last few years we’ve seen a lot of experimentation—thousands of AI fueled products have been launched—some finding spectacular product market fit, reaching more than USD100 million annual recurring revenue (ARR) in less than a year of launch, while others stuck in the pilot stage with low adoption or questionable durability.

We believe we are now entering a new phase of AI that’s less about experimentation and more about putting into production and delivering measurable outcomes. Buyers are holding new technology accountable to real return on investment (ROI) on shorter time horizons—is your product saving the enterprise money, driving measurable efficiencies, or generating new revenue opportunities.

We see the next wave of application technology embedded deeper into enterprise workflows, training on proprietary data, more domain-aware agents that move beyond assistance to coordinating multi-step work across teams, and orchestration tools that go beyond copilots. We believe that early use cases of AI like software coding and customer support will continue to scale, but it’s been interesting to see momentum in fields that traditionally have been slower to adopt new technology such as medicine, law, and accounting.

At M12, we’ve been thinking deeply about the next evolution in models beyond text-based large language models (LLMs), particularly world models, and how data from the physical world brings about new innovations, particularly in science.

We are also focused on the foundational layers that make AI possible at scale— infrastructure, tooling, data systems, and how to make AI factories more efficient, sustainable, and performant. You can see that in our recent investments in nEye.AI and Neurophos, and you’ll continue to see us make bets in these areas.

On the business side, we’re seeing rapid change. New pricing models like outcome-based and consumption-based pricing, faster adoption cycles particularly in small and midsize business (SMB) and individual buyers, and a sharper focus on ROI are reshaping how value is created.

Achieve business goals with AI

Explore extensive guidance and resources from Microsoft

Get access

What are the key criteria you look for when evaluating businesses to invest in?

We look at a combination of factors and because we invest early in a company’s lifecycle, we focus deeply on the strength of the team, their conviction, and their long-term vision. The best teams are shipping quickly, adopting the latest models, listening closely to customers, and adjusting based on real-world feedback. We are finding that velocity is “the” moat for many AI companies. We also look at market size and opportunity and increasingly, a founder’s plan for how to differentiate in a crowded market, and how we can support the business.  

Ultimately, we’re looking for incredible teams that are learning quickly, innovating with customers, and building with durability in mind. 

What are some of your recent investments and what made them stand out?

M12 meets hundreds of startups each year and typically invests in 18 to 20 early-stage companies a year. We aim to balance our portfolio with a mix of companies that have nearer term adoption and traction and those with longer time horizons.

• Inception Labs is pioneering a diffusion large language model (LLM) approach, which has some performance advantages over transformers. Its early—a “foundation model” and frontier investment that we’re excited by.
• Outset is an AI-moderated research platform that is fundamentally reshaping how enterprises operate and understand their customers, which is exactly what we look for in teams and technologies—they already have an impressive list of customers, including Microsoft, Uber, and Hubspot. 
• Neurophos is tackling one of the biggest bottlenecks in AI infrastructure: the inability of GPUs to scale efficiently across cost, power, and footprint. The team has developed a manufacturable optical processing unit that compresses GPU-scale compute into a dramatically smaller, cooler, and more sustainable form, delivering up to 100× performance and energy efficiency compared to today’s leading systems.
• Entire is approaching the next general wave of developer platforms from first principles, architecting a new platform from the ground up for agent-to-human collaboration. We’re excited to invest in Entire and support a team that is inventing a new paradigm of how developers and AI work together.

Where do you see the most promising opportunities for AI to disrupt industries globally, and how is M12 positioning itself to lead in those areas?

The biggest opportunities for AI disruption are in industries where complexity, scale, and constraints intersect—areas like enterprise transformation (this could be system of records being unbundled or completely rethought, hyper-personalized employee agents and applications), next gen cybersecurity, tooling and infrastructure that needs to be rebuilt when agents are writing the majority of software code, and innovations brought on by world models deployment are also top of mind.

M12 is uniquely positioned to support founders tackling these challenges because we invest where Microsoft brings deep expertise—enterprise environments, technology, global scale, and mission-critical systems. Our investors are thesis driven and often will spend months getting to know a space, going deep, which builds credibility with founders before an investment.  

How are you seeing founders adapt their strategies to thrive in an AI-first world, and what advice would you give them?

Founders are shipping and executing faster, focusing on deeper integrations with customers, and building community and brand awareness quickly as part of their go-to-market. Many enterprises today are running multiple proofs of concept across similar AI tools and we believe this next year customers will focus and consolidate. There will be a major shift towards deciding which solutions deliver real ROI and are ready for wide-scale production. 

Founders who succeed will be the ones who build deeply integrated, customized workflows that fit into how customers actually operate. We’re seeing startups pairing product innovation with services, like forward-deployed engineers, to accelerate adoption early on, getting feedback loops started quickly as well with access to unique data to make their products sticky and “must haves.”

We’re also seeing companies rethink how they operate internally. One of our portfolio companies, Allstacks, which was founded more than 5 years ago, radically shifted their go-to market strategy last year to become AI native, where AI tools and agents are integral parts of each operating function. It was critical work and a good example of how forward-thinking companies are reinventing their operations to match their product innovation.

Many AI native companies are operating incredibly leanly, even as they scale, due to the use of AI in all elements and functions of their business. 

How does M12 evaluate which AI startups have the potential to scale and make a lasting impact?

We spend a lot of time distinguishing between the hype cycle and the durability cycle. Markets can move quickly, we’re seeing that more than ever, but lasting companies are built deliberately over many years.

The startups with the greatest potential are those that design with constraints in mind, whether that’s power, capital efficiency, inference economics, or customer trust, and still find ways to deliver meaningful value.

Why is collaboration between corporations and startups essential for accelerating AI transformation globally?

AI transformation and adoption especially in the enterprise doesn’t happen in isolation. Startups bring speed, creativity, and new ideas. Corporations bring scale, trust, and real-world deployment environments.

When those strengths come together, we see not an additive but multiplier effect. Corporations like Microsoft can serve as a bridge between early-stage founders and global enterprise customers, helping startups move faster while deploying AI responsibly and at scale.  

For example, Microsoft Marketplace is a great way for startups to transact with enterprise customers—we have a portfolio development team that is solely focused on opportunities like this, helping our portfolio companies join the marketplace and get ready to sell to enterprises, get through procurement and generally navigate the opportunities to partner with Microsoft and our customers.

That collaboration is what can turn breakthrough technologies into lasting impact, and it’s why partnerships are so critical in this next chapter of AI. 

Ready to learn more? Discover resources and tools to accelerate your AI journey

• Learn more about M12, Microsoft’s venture fund
• Explore cloud solutions, AI apps, and agents on Microsoft Marketplace
• Get the Prompt Guide for Business Leaders to turn AI intent into clear, actionable decisions
• Discover how Microsoft AI brings intelligence into the flow of work

The post FYAI: Why startups will help accelerate global AI transformation appeared first on The Microsoft Cloud Blog.

In this edition of FYAI, a series where we spotlight AI trends with Microsoft leaders, we hear from Amanda Silver, Corporate Vice President and Head of Product, Apps, and Agents. Amanda’s leadership has shaped Microsoft’s evolution toward open-source collaboration, and she’s advancing a future where AI transforms how developers build, deploy, and iterate at scale to drive continuous innovation.

In this Q&A, Amanda shares why developer-led AI adoption matters, how agentic DevOps is redefining workflows, and what leaders can do today to maximize impact.  

How is the AI landscape changing how developer teams deliver the apps businesses run on?  

AI is collapsing handoffs across the software lifecycle. DevOps successfully united build, test, deploy, and operate, but the earlier phases—discovery, requirements, shared vision, and initial scaffolding—mostly sat outside that loop. Now copilots can turn natural language ideas into specs and scaffolds, and agents take on tests, upgrades, and runtime operations. The result is a single, faster cycle from idea to impact: lower cost to iterate, quicker transitions, and more freedom to refine until the product fits the business. Think of it like the shift to public cloud: before the public cloud, teams waited weeks to procure hardware and commit capital up front; with the cloud, environments spin up in seconds and you pay only for what you use. AI brings that same elasticity to product definition and delivery—removing friction at the front of the lifecycle and letting teams iterate based on real feedback. Put simply: cloud removed friction from infrastructure; AI removes friction from intent to implementation.

What are some examples of how AI is helping developers re-imagine their daily work?   

AI is turning software delivery into a true idea-to-operate system. For developers, that means less time spent on manual cleanup and more time focused on creative, high-impact work. Copilots and agents now handle the repetitive, often invisible tasks that used to pile up—like debugging, dependency upgrades, and security patches. Instead of waiting for a quarterly “tech debt sprint,” agentic DevOps lets teams pay down debt continuously, in the background.  

A great example is how agentic AI is accelerating migration and modernization. In the past, updating frameworks or moving to new platforms meant months of planning and manual fixes. Now, agents can automate .NET and Java upgrades, resolve breaking changes, and even orchestrate large-scale migrations—compressing timelines from months to hours. This isn’t just about speed; it’s about keeping codebases healthy and modern by default, so developers can focus on building new features and improving user experiences.  

The net effect: developers spend less time firefighting and more time innovating. Technical debt becomes a manageable, ongoing process—not a looming crisis. And as AI agents take on more of the routine work, teams can operate in a steadier flow state, with healthier code and faster delivery.  

What does that mean for apps? Are they getting better? And how does this impact the role a developer plays?

Apps will get better because they become learning systems. With AI in the loop, teams shift from ship and hope to continuous observe → hypothesize → change → validate cycle centered on continuously refining product–market fit. AI can help synthesize telemetry (such as funnels, dropoffs, session replays, and qualitative feedback), surface where users struggle, propose changes (like copy, flow, component layout, and recommendations), and can even wire up feature flags or experiments to prove whether a change works. The effect is a dramatic reduction in time to learning—and faster convergence on what users value.  

PreAI versus PostAI user interaction  

• PreAI: Users navigate dense menus and deep information architectures, scanning screens to find the one control that does what they need. Every step risks a dead end, and context is easy to lose when switching pages or tools.
• PostAI: Users express intent in natural language (like text, speech, or multimodal). The app interprets that intent, keeps context, and routes the user to the right data, action, or workflow—often composing the UI on the fly (for example, drafting a form, filtering to the relevant records, and suggesting the next best action). Think of this as moving from “Where do I click?” to “Here’s what I need—do it with me.”  

What changes for developers  

• From page builders to experience composers. Devs design intent routers and orchestrations that connect models, agents, data, and services—so the app can respond intelligently to varied user goals without forcing rigid click paths.
• From manual analysis to AI-assisted product loops. Instead of hand rolled dashboards and ad hoc investigations, AI highlights opportunity areas, drafts experiment plans, and opens pull requests with proposed code and config changes. Developers review, constrain, and ship—with guardrails.
• From “debt sprints” to continuous modernization. Agents can keep the app current—upgrading frameworks (for example, .NET and Java), repairing dependency drift, patching vulnerabilities, and standardizing pipelines—while feature work continues. That turns tech debt into a managed, always on workload rather than a periodic fire drill.   

Bottom line: AI tightens the loop between what users want and what the app becomes. Developers spend less time on menu wiring and manual forensics, and more time defining intent, composing agentic flows, setting success metrics, and supervising safe, measurable change. Apps improve faster—not just because they’re smarter, but because teams can experiment, learn, and adapt as usage grows.  

Where do you see Microsoft standing out in a sea of AI competition?  

Microsoft’s biggest differentiator is our ability to connect AI agents to the systems, data, and workflows that power real business. We serve organizations with massive, complex codebases and deep operational requirements—and our tools are designed to meet them where they are. With GitHub, Visual Studio, and Azure AI Foundry, millions of developers can access the latest models and agentic capabilities directly in their daily workflow, backed by enterprise-grade security, governance, and responsible AI benchmarks.  

But what truly sets Microsoft apart is the breadth of integration. AI agents built on our platform can tap into a huge ecosystem of business apps, data sources, and operational systems—whether it’s enterprise resource planning (ERP), customer relationship manager (CRM), human resources (HR), finance, or custom line-of-business solutions. Through open standards like Model Connector Protocol (MCP) and Agent-to-Agent (A2A), agents can securely connect, orchestrate, and automate across these environments, making it possible to deliver outcomes that matter: automating workflows, modernizing legacy systems, and driving continuous improvement.  

Yina Arenas’s Agent Factory series shows how Microsoft is building the blueprint for safe, secure, and reliable AI agents—from rapid prototyping to production, observability, and real-world use cases. Our platform isn’t just about building agents; it’s about enabling them to work with the systems and data that organizations already rely on, so teams can move from experiments to enterprise-scale impact.  

At the end of the day, Microsoft’s advantage is not just scale—it’s the ability to make AI agents truly useful by connecting them to the heart of the business, with the tools and standards to do it safely and securely.  

When should developers decide which tasks to delegate to agents versus tackle themselves for maximum impact?  

As my colleague, David Fowler, put it: “Humans are the UI thread; agents are the background thread. Don’t block the UI!” Developers should focus on the creative, judgment-driven work—setting intent, making architectural decisions, and shaping the product experience. Agents excel at handling the repetitive, long-running, or cross-cutting tasks that can quietly run in the background: code health, dependency upgrades, telemetry triage, and even scaffolding out solutions to unblock the blank page.  

The key is to delegate anything that slows down your flow or distracts from high-impact work. If a task is routine, latency-tolerant, or easily reversible, let an agent handle it. If it requires deep context, product judgment, or could fundamentally change the direction of your app, keep it on the human “UI thread.” This way, developers stay responsive and focused, while agents continuously improve the codebase and operations in parallel.  

By striking the right balance, developers can minimize time spent on routine tasks and stay focused on the work that moves products and teams forward. 

Why are AI coding tools attracting so much investment and interest? Why reimagine the developer experience now?  

Because software development already generates the kind of rich, structured signals that AI thrives on. Code and diffs, pull request reviews, test results, build logs, runtime and performance telemetry, issue trackers, and deployment outcomes are all labeled, timestamped, and traceable. That makes the dev environment a natural proving ground for applied machine learning: models can learn from real work, be evaluated against objective checks (like tests, linters, and policies), and improve inside an existing feedback loop (such as Continuous Integration and Continuous Delivery (CI/CD), feature flags, and canaries). In short, we have the data, the instrumentation, and the validation built in.  

There’s also a cultural reason: developers automate away friction—from compilers and build systems to version control, CI/CD, containers, and infrastructure as code. Generative AI is the next step in that lineage. It shifts more work from hand authoring to specifying intent and supervising outcomes: copilots help with exploration and acceleration; agents handle continuous code health, upgrades, and safe, reversible changes. Investment flows here because better developer experience maps directly to throughput, quality, and time to value.  

And yes—the future starts with developers. As dev teams discover where AI delivers real support in their own workflow, those patterns spread to the rest of the business, accelerating how every function experiments, learns, and ships.  

Empowering developers with AI to deliver lasting impact 

We’re entering a new era of software delivery—and it’s agentic, adaptive, and deeply human-centered. With copilots and agents in the loop, developers are building systems that continually adapt to business needs. At Microsoft, we’re empowering developers to move from idea to impact faster by focusing on creativity, product vision, and building with trustworthy AI. 

In fact, Frontier Firms are already showing us what’s possible. They treat software as a dynamic system—refined through telemetry, experimentation, and AI-powered insight. And across all types of organizations, compelling AI use cases are emerging—from customer service to software engineering—setting the pace for what’s possible with the latest AI tooling. 

Ready to learn more? Discover resources and tools to accelerate your AI journey: 

• Learn three skill-building insights that help Frontier Firms drive innovation.
• Get started with GitHub Copilot.
• Build your first production-grade agent in under an hour with Azure AI Foundry.
• Simplify development and meet evolving business needs with Microsoft Cloud solutions. 

The post FYAI: Why developers will lead AI transformation across the enterprise appeared first on The Microsoft Cloud Blog.

In this edition of FYAI, a series where we dive deep on AI trends with Microsoft leaders, we hear from Juan M. Lavista Ferres, Corporate Vice President and Chief Data Scientist for Microsoft’s AI for Good Lab, who is leading Microsoft’s efforts to create a collaborative ecosystem and a team of dedicated data scientists and researchers, domain experts, and organizations worldwide that drives progress toward addressing some of the world’s most pressing challenges. 

In this Q&A, Juan shares his insights on the growing AI for good movement, how Microsoft partners with local organizations to help solve region-specific challenges, and how to measure and scale AI projects to drive impact. 

You’re Microsoft’s Chief Data Scientist and Director of the Microsoft AI for Good Lab. What does that actually look like day-to-day?

My journey at Microsoft started all the way back in 2009 when I was running randomized controlled experiments as part of Microsoft EXP. Looking back, so much of my work then informed what eventually matured into our official AI for Good program. One of our first AI projects focused on better understanding the causes of sudden infant death syndrome. The research we produced with our collaborators helped understand some of the causes—and in so doing, reinforced our conviction that AI can be an indispensable tool in improving people’s lives.

This also illuminated a gap that clearly needed to be filled. The majority of AI expertise in the world is concentrated in the tech and financial sectors, and too many organizations working on solving important societal problems don’t have the resources or technological expertise to apply AI to their work. We help fill this gap by providing AI expertise and compute capacity to bring cutting-edge technology to our partner organizations.

I have a strong bias for action, so each day I am focused on how we can leverage AI to help the most people, as fast as possible. This means I work with our team to conduct applied research, seek out and engage diverse new partners, and guide the build out and deployment of AI solutions. In many ways, this is my dream job. I get to find ways to use data, apply AI, and improve the world, and do so with some of the most gifted researchers and scientists.

Our Lab was one of the first in the AI for Good space, which I’m proud to say has become a growing movement. The United Nations, for instance, hosts an annual AI for Good conference which we participated in this July in Geneva. It’s no surprise that this is catching on. For many of the problems we’re working on, AI isn’t just a solution, it’s the only solution.

Who is Microsoft partnering with to solve region-specific or community-based challenges?

Partnership is at the heart of the work we do. We believe that in order to make the most of AI’s potential to benefit everyone, we must leverage local, on-the-ground expertise to co-create solutions tailored for the communities with whom we’re working. We have teams in Microsoft offices in the United Arab Emirates, South America, and Kenya because it’s vitally important for us to work alongside local researchers.

One example of the importance of partnership in practice is our work to detect historical flooding and improve disaster response globally. Floods are among the deadliest forms of extreme weather, affecting millions of people and causing over USD40 billion in damage each year. To address this, we developed a model using 10 years of Synthetic Aperture Radar (SAR) data, which can detect floods through clouds, at night, and in remote regions. This is the first model of its kind. We worked closely with partners in Kenya, Ethiopia, Spain, and the United Nations to enable real-time response. In Ethiopia, we used the model to detect nearly three times more flood-prone areas than existing datasets. It confirmed known flood zones and revealed new ones, offering a more complete view of flood risk across the country. After devastating floods struck Kenya in May 2024, we used this tool to estimate that 75,000 hectares—about 2% of all cropland—were flooded or at risk.

In many cases, the value proposition we’re bringing to the table is not only increased accuracy but also speed. If we can use AI to, for example, measure deforestation in the Amazon in a matter of days rather than months, governments in the area can respond much faster.

How do you measure the impact of AI for Good projects?

The definition of ‘impact’ is unique to each project, but overarchingly, we assess whether and to what extent people’s lives have been improved through our work. The key to achieving this is being guided by subject matter experts who have a robust understanding of the challenges to be solved and the data available to solve them. Our goal is always to support our partners and advance their missions, whether that means forecasting malnutrition with greater precision or improving how we track giraffe populations to support conservation efforts.

We also want to expand the benefits of this technology to communities that otherwise might not be able to take advantage of it. For example, we’re working on a tool to diagnose retinopathy of prematurity in low-resource settings where pediatric ophthalmologists aren’t available. Retinopathy of prematurity is an increasingly common condition, thanks to improvements in maternal and infant healthcare that saves the lives of prematurely born babies around the globe. But if this condition isn’t diagnosed and treated, it can lead to lifelong blindness. We’ve developed algorithms that demonstrate how AI can help detect this condition using videos captured on a smartphone. This has the potential to empower healthcare professionals to intervene early in an infant’s life and preserve their eyesight.

The more accurate and useful our AI tools are, the greater their downstream impact will be. This is no small feat. AI technology is evolving and changing rapidly, and our researchers work tirelessly to stay at the forefront.

Where do you see the biggest need—and opportunity—to scale this work?

In order to really scale the benefits of AI, we need to ensure that people are connected not just to AI but to digital technology in general, which is still out of reach for 2.6 billion people around the world.

Our team has invested heavily in developing disaster response capabilities using geospatial machine learning and satellite imagery. We plan to continue supporting first responders in the immediate aftermath of natural disasters in the United States and across the globe. We’ve also seen tremendous benefits of using AI for biodiversity conservation, such as in the Amazon and in Sub-Saharan Africa.

As a non-native English speaker from Uruguay, I use AI to ensure my grammar in documents like research papers is pristine and error-free. A vast majority of academic publications and journals are in English. Large language models have made it easier for myself and countless others to contribute their research, which is a tide that lifts all boats.

What’s a project you’re most excited about right now—and why?

At the end of 2024, we announced Project SPARROW, an AI-powered edge computing solution designed to autonomously collect biodiversity data in the most remote corners of the planet. These devices can run for long periods of time on solar power without disrupting the ecosystems in which they’re embedded, which is one of the aspects of the project I’m most excited about. Protecting vital ecosystems benefits people, animals, and our planet. Our goal is to have SPARROW operating on every continent by the end of the year.

In June 2025, I traveled to Nairobi to connect with our AI for Good Lab in Kenya and our partners on the ground there. I was also invited to give a TED talk focused on SPARROW and the opportunities it presents to increase biodiversity. I highlighted earlier the importance of partnerships with local organizations, and I view this as a chance to uplift our collaboration.

Using AI to make new discoveries and improve people’s lives 

When researchers, conservationists, nonprofits, non-governmental organizations (NGOs), and academic institutions are tasked with doing more with less, technology offers a way forward. AI transforms the way we can use data to make new discoveries and improve lives.  

Microsoft is investing not only in solving today’s challenges but also paving the way for a brighter tomorrow. Our work helps showcase the potential benefits of AI, pushing the bounds of what was previously possible. By applying advanced AI to real-world datasets, the AI for Good Lab helps unlock insights that empower communities and accelerate progress on urgent global issues. 

AI for Good Lab

harness the transformative power of AI to tackle global challenges and improve lives.

Explore projects

Want to learn more? Here are a few resources

• Explore the AI for Good Lab’s projects.
• Begin your AI journey with AI resources and solutions from Microsoft.
• Check out our recently announced AI Economy Institute, Microsoft’s corporate think tank dedicated to advancing independent research and developing actionable solutions that can help societies worldwide positively adapt to the economic and social transformations brought by AI.

The post FYAI: Explore the Microsoft AI for Good Lab with Juan M. Lavista Ferres appeared first on The Microsoft Cloud Blog.

From streamlining enterprise workflows to accelerating scientific discovery, AI agents are transforming how developers build and how businesses operate.

• 15 million developers are using GitHub Copilot, using features like agent mode and code review to handle repetitive tasks, allowing them to focus on the fun, creative parts of software development.
• Hundreds of thousands of customers are using Microsoft 365 Copilot to assist with research, brainstorming, and solution development, allowing increased for efficiency.
• More than 230,000 organizations—including 90% of the Fortune 500—have used Microsoft Copilot Studio to build AI agents and automations to improve productivity and scale business quickly.
• More than 11,000 AI models are now available through Azure AI Foundry, including Microsoft-hosted and partner-hosted models. This extensive library of AI models provides unparalleled resources for organizations to innovate and scale their AI-powered solutions. 

In this edition of FYAI, a series where we dive deep on AI trends with Microsoft leaders, we hear from Yina Arenas, Vice President of Product, Azure AI Foundry, who is leading the work at Microsoft to empower every developer to shape the future with generative AI using breakthrough models and enterprise AI agents.

In this Q&A, Yina shares her insights on the shifting AI landscape, including why businesses are getting stuck in the “proof of concept” phase and how Azure AI Foundry can meet organizations where they are and take their AI projects to the next level.

What shifts in the AI landscape are you seeing that are fundamentally changing how people—and organizations—build and scale AI?

We’re seeing a profound shift from AI as a research experiment to AI as a core business capability. What’s exciting—and challenging—is that organizations are no longer just asking, “Can we build this?” but “How do we build this responsibly, at scale, and with real impact?” That shift requires new tools, new mindsets, and new ways of working across teams. At Microsoft, we’re focused on making AI more accessible and inclusive—so that everyone, from developers to domain experts, can contribute to building solutions that matter. It’s not just about the tech—it’s about empowering people to solve real problems with AI.

Why is it still so hard for businesses to move from experimentation to production with AI—and what needs to change to unlock that next wave of value?

Many organizations get stuck in the “proof of concept” phase because the leap to production is complex. It’s not just about selecting the right model—it’s about integrating it into systems, ensuring it’s secure and responsible, and aligning it with business goals. What’s missing is a cohesive, end-to-end approach that brings together the right tools, governance, and collaboration in a developer-friendly environment. That’s where Azure AI Foundry comes in—it’s designed to help teams not only move faster but do so thoughtfully by providing a cohesive end-to-end platform and offering traceability across prompts, models, and runtime behavior. We’re making it easier and less complex for developers to build apps while also giving business decision makers the ability to see how these apps perform, measure their ROI, and meet compliance requirements. To unlock the next wave of value, we need to make AI development more collaborative, transparent, and outcome-driven.

How does Azure AI Foundry help bridge that gap—and how is it different from other approaches out there?

Azure AI Foundry is built to meet organizations where they are—whether they’re just starting or scaling AI across the enterprise. It brings together the best of Microsoft’s AI capabilities from foundational models to orchestration and monitoring in a unified platform. What sets Azure AI Foundry apart is not only that it’s built on decades of world-class research but that it’s built with humans at the center, so whether you’re a data scientist, product manager, engineer, or business leader, our AI solutions work for you. It also bakes in responsible AI from the start by integrating tools, from testing to monitoring to governance, that support the entire life cycle.

Who is Azure AI Foundry built for, and how does it support cross-functional teams—from data scientists to decision-makers—to build together?

Azure AI Foundry is designed for anyone looking to take their AI projects to the next level—whether you’re part of a big enterprise, a startup, or a software development company. It offers access to the leading frontier models, integrates orchestration frameworks, supports open protocols for multi-agent collaboration, and provides native observability tooling—all within a secure, governed environment. Whether it’s optimizing call centers, analyzing data, improving product searches, or automating workflows, Azure AI Foundry pulls everything—models, tools, and agents—into one user-friendly platform. With tools like GitHub, Visual Studio, and Copilot Studio, Azure AI Foundry makes it easy for developers, data scientists, IT pros, and decision-makers to shorten the journey from idea to production.

Azure AI Foundry

Design, customize, and manage AI apps and agents at scale.

Get started today

Where are you seeing Azure AI Foundry already making an impact—and what kinds of transformation are customers unlocking?

As the central hub for building, orchestrating, and managing AI solutions, Azure AI Foundry remains the centerpiece of our AI platform strategy. It is now used by developers at more than 70,000 enterprises and software development companies—including Atomicwork, Epic, Fujitsu, Gainsight, H&R Block, and LG Electronics—to design, customize, and manage their AI apps and agents. And just six months in, more than 10,000 organizations have used Azure AI Foundry Agent Service to build, deploy, and scale their agents. Developers are designing agents that act, reason, take initiative, and deliver measurable business outcomes.

Heineken, for example, used Azure AI Foundry to build a multi-agent platform called “Hoppy” that helps employees access data and tools across the company in their native language. Their implementation has already saved thousands of hours, reducing tasks that once took 20 minutes to just 20 seconds.

Fujitsu evaluated Azure AI Foundry Agent Service to automate sales proposal creation. This boosted productivity by 67%, letting their teams to focus on customer engagement. The AI agent integrates with existing Microsoft tools familiar to around 38,000 employees, retrieves dispersed knowledge, and lays the foundation for broader AI-powered innovation.

Draftwise, a digital native offering an AI-powered contract drafting and review platform, is using cutting edge models in Azure AI Foundry (Cohere multimodal and AOAI reasoning) to help streamline the contract drafting process by integrating with a lawyer’s document storage system.

What excites you most about what’s next—for Azure AI Foundry, and for how people can reimagine the way they work and create with AI?

What excites me most about what’s next for Azure AI Foundry is how it’s unlocking a new era of creativity and empowerment—not just for developers, but for everyone. We’re moving beyond the idea of AI as a tool you use to AI as a copilot you build with. Azure AI Foundry is helping people imagine and create agents that understand their goals, adapt to their workflows, and evolve with their needs.

That shift—from writing code to orchestrating intelligence—is profound. It means that a product manager, a marketer, or a frontline worker can shape how AI works for them, without needing to be a machine learning expert. It’s about putting the power of AI into the hands of the many, not the few.

And what’s most inspiring is that we’re just getting started. The agents people are building today are solving real problems—automating complex processes, accelerating insights, and freeing up time for more meaningful work. But the agents of tomorrow? They’ll be collaborators in creativity, partners in problem-solving, and catalysts for innovation we haven’t even dreamed of yet.

That’s the future I see—and it’s being built right now, by people who are reimagining what’s possible with AI.

Design, customize, and manage AI apps and agents at scale

Through leaders like Yina Arenas, Microsoft’s vision for the future of AI is both inspiring and deeply human-centered. With platforms like Azure AI Foundry, we’re entering a new era where AI becomes not just a tool, but a true collaborator—empowering everyone, regardless of technical expertise, to innovate and solve real-world problems. With Azure AI Foundry, the potential of AI is being unlocked by developers everywhere, sparking a wave of transformation and boundless possibilities.

Interested in learning more? Here are a few resources:

• Build your first production-grade AI agent in under an hour: Azure AI Foundry
• Learn how Azure AI Foundry is supporting open Agent2Agent (A2A) protocol
• Read Azure AI Foundry Agent Service documentation
• Empower your team to grow their AI skills
• FYAI: How agents will transform business and daily work

The post FYAI: How to leverage AI to reimagine cross-functional collaboration with Yina Arenas appeared first on The Microsoft Cloud Blog.

From the world’s biggest companies using agents to automate business processes that run tens of thousands of times a day to each of us now having the power to quickly create custom AI assistants using plain language—agents are reshaping work for both organizations and individuals.  

In this edition, we hear from Charles Lamanna, Corporate Vice President of Business and Industry Copilot, who is spearheading the work at Microsoft to bring AI agents to organizations. In this Q&A, Charles shares his insights on how we arrived at this moment in AI transformation, agents in the workplace for the organization and the individual, why customers that are AI-first are thriving, and where agents will start to show up outside of work.  

Let’s learn more from Charles about the transformative potential of agents and where this technology is headed next. 

What is a good analogy to describe the AI moment we are in right now? 

AI can feel abstract, but to understand the moment we’re in, it’s helpful to think about other turning points in history where technology became a force multiplier for people.  

Take the tractor, for example. Two hundred years ago, nearly everyone in America farmed for a living. Now, it’s less than 2%, because tractors completely changed the game, freeing people up to innovate elsewhere. Or clothing—not that long ago, most people owned a few outfits. Then mechanized looms came along and clothing became abundant. 

Steve Jobs famously called computers a “bicycle for the mind” because just as a bicycle helps humans move far more efficiently, computers amplified what our minds could achieve. Stretching that analogy, if computers are the bicycle, AI is the jetpack of the mind. It’s not just speeding us up, it’s lifting us to entirely new heights.  

If computers are the bicycle, AI is the jetpack of the mind. It’s not just speeding us up, it’s lifting us to entirely new heights.”

—Charles Lamanna

We’re at an exciting time where AI puts deep expertise directly at everyone’s fingertips, breaking down barriers to knowledge. Just as tractors transformed farming and mechanized looms changed clothing production, AI agents are transforming fields like law, medicine, and software development. They’re the tractors for lawyers, the mechanized looms for doctors, and combustion engines for developers.  

Why is the shift to an “AI-first mindset” so important for businesses?  

Adopting an AI-first mindset is crucial because it fundamentally transforms the way businesses operate. AI isn’t just a novelty—it’s a core capability that is necessary to stay relevant.  

AI business resources

Help your organization achieve its transformation goals

Discover resources

Businesses embracing AI-first thinking reach new levels of scale that are just not possible without the speed and power of AI. For example, global supply chains are enormously complex. AI can quickly process vast amounts of data, predict trends, and take actions in real time—tasks that would take people weeks or months can now happen in hours or minutes. AI agents can generate reports or even actively make informed decisions under human oversight.  

A core component for an AI-first company is agents—think of them as the new apps. They can execute core tasks with and on behalf of people, unifying business data, apps, email, chat platforms, and more. Agents can range from simple to advanced, doing everything from addressing customer service inquiries, to doing the heavy lifting of data analysis.  

Imagine replicating this efficiency across every element of a company: human resources, logistics, sales, finance, and research and development. You can see how profound this transformation is across the business.   

When do you think AI agents will go from being an add-on to the main driver of business? 

Right now, we’re seeing the beginnings of a major transformation where AI agents are shifting from being used as helpful “add-ons” to becoming the core drivers of business. We can think about this AI journey in a few stages.  

Initially, we’re seeing AI augment people in an organization—everyone has a powerful AI assistant that deeply understands their specific work, makes them more productive, and makes daily tasks easier.  

As organizations move further along this journey, we’ll see a big jump. AI agents will evolve to become key team members, capable of autonomously managing complex workflows and boosting efficiency considerably. People will set high-level strategies, provide direction and manage these agents. And this will continue to evolve.   

I think companies that have adopted an AI-first mindset are already working towards that future as more advanced agent capabilities are added.  

Who do you think is doing a good job of working with AI and agents? 

It’s exciting to see companies adopting AI agents in a big way. 

Take Estée Lauder Companies, for instance. They have 80 years’ worth of valuable consumer data from surveys, clinical trials, promotions, and product usage. With an agent called ConsumerIQ, built using Microsoft Copilot Studio, employees can instantly tap into insights that used to take hours of manual research. They can ask, “What are the latest trends for mascara use among Gen Z?” and within seconds, the agent will collect, summarize, and deliver the answer. 

Dow spends billions annually on freight shipping and receives thousands of invoices daily. They built agents that analyze and detect anomalies and have already uncovered billing errors—like catching a $30,000 charge that should’ve been $5,000.  

At Microsoft, we’re also using agents, and one example is on Azure.com. With more than 400 product and service pages, customers struggled to find the information they needed. The team built an AI assistant using Copilot Studio, and visitors who used the AI assistant showed 70% more pages visited per session and a 21.5% increase in conversion rates.  

Where do you see agents changing our daily lives beyond the workplace?  

We’re going to see agents show up more in our daily lives, and I think people will be excited about it. Imagine having a personal agent that handles annoying tasks no one enjoys—sorting emails, paying bills, or making appointments.  

In healthcare, agents can become personal health assistants, proactively monitoring your wellness and reminding you about medications. This could make life easier for those managing chronic health problems. 

On top of all this, anyone will be able to create agents for their own use without any coding experience. It won’t be something that only developers or coders are trusted to do.  

Looking ahead, the possibilities are limitless. Imagine a future where agents effortlessly handle life’s complexities, freeing you up to spend more time on the people and activities that bring you joy and meaning. That’s the future we’re building towards, and it’s closer than you might think. 

Learn more about Microsoft’s innovation with AI agents 

AI agents are an exciting space for business leaders to explore, from being able to assist with document creation at an individual worker’s level to taking on the most intensive and critical business processes at a company. Agents will vary in complexity, and they’ll come from many different sources—from the agents built by Microsoft or our partner ecosystem to the custom agents tailored to take on your exact challenge. With this rise of AI agents, we will see even more ways for AI deliver on the promise of real business value.

Interested in learning more?

Here are a few recommended resources:

• Register for the Business Applications Launch Event
• Refactor your skills at Microsoft Build 2025
• Read AI agents — what they are, and how they’ll change the way we work
• Begin your AI learning journey with the AI learning hub 

The post FYAI: How agents will transform business and daily work with Business and Industry Copilot Corporate Vice President Charles Lamanna appeared first on The Microsoft Cloud Blog.

Beyond the tools that empower businesses to shape their future with AI in a rapidly evolving market, our leaders at Microsoft are shaping our own organization with this technology. In this series, FYAI, we’ll highlight leaders from around Microsoft that are driving forces in our AI strategy for their unique perspective on our AI transformation; for your AI information, if you will.

Insights from Sarah Bird, Chief Product Officer (CPO) of Responsible AI

In this edition, we hear from Sarah Bird, Microsoft’s Chief Product Officer (CPO) of Responsible AI, ahead of her appearance at South by Southwest (SXSW) where she’ll be discussing the evolving safety practices for generative AI.  

In this Q&A session, Sarah shares her insights on various aspects of responsible AI, including her journey and dedication to responsible AI, her role as Chief Product Officer, the importance of integrating responsible AI early in the development process, and her insights on future AI breakthroughs and their safety implications. 

Let’s explore Sarah Bird’s experiences and perspectives on the evolving landscape of AI and discover how Microsoft is building trustworthy AI systems. 

FYAI: Responsible AI with Sarah Bird

 Who influenced you to pursue a career in responsible AI?

“For me, it’s less about who influenced me to pursue this career and more about who I’m helping every day through my work. AI is one of the most empowering technologies we have, but we can’t unlock its full potential without solving for responsible AI. That’s what makes this work so important—it’s about ensuring AI is safe and beneficial for everyone. And to do that, we have to work across boundaries. It reminds me of my grad school days—responsible AI is the ultimate group project, bringing together technology, society, and law to tackle these complex challenges in a meaningful way.” 

What does the role of chief product officer, responsible AI, actually mean? Tell us what your day-to-day looks like. 

“No two days are the same, and that’s what keeps me energized. At the core, my team is focused on three key things: spotting new risks, figuring out how to tackle them—especially when they’re things we’ve never seen before—and making sure our solutions are scalable so others can apply them easily. That framework guides us, but the reality is, AI is evolving fast. So a big part of our work is staying nimble—triaging issues in real-time, applying what we learn in practice, and adapting quickly to test and deploy new systems. It’s a mix of strategy and problem-solving, which is what makes it exciting.”

Where are you noticing gaps in companies’ implementation of AI safety practices?

“It’s been really inspiring to see how much more mature customers are getting with their responsible AI roadmaps and deployment. There’s real progress happening. That said, people are still learning, and the level of maturity varies across industries—some are further along than others. If there’s one thing I could shout from the rooftops, it’s that responsible AI can’t be an afterthought. It needs to be built into the entire development process from the start, not just bolted on at the end. It’s about putting all the pieces together to create a complete, responsible AI lifecycle.”

When do you think the next AI breakthrough is going to happen and what does that mean for safeguards?

“As an engineer, I’m focused on problem-solving rather than predicting when the next big breakthrough will happen. But I will say—it’s an exciting journey, especially with the pace of innovation. And while we still need another major leap before we can talk about the reality of what’s next, what’s really exciting about this space is that the breakthrough isn’t just the technology itself—it’s how we apply it. The real magic happens at the intersection of tech and people, and figuring out how to bridge that responsibly is what makes this work so fascinating.”

Why do you feel safety and innovation go hand in hand? 

“A goal of ours as a company is to help people do more with AI. We are constantly pushing the boundaries of what’s possible and doing so in a safe, trusted way. As I’ve said, safety is not just a ‘nice to have’ bolted on at the end of a project, but a critical piece of developing high-quality AI systems. I look at safety issues as a measure of quality – is your AI performing as well as it should be? We can’t innovate and drive meaningful progress if we don’t solve for this.” 

2025 AI Decision Brief

Gain insights from thought leaders at Microsoft to advance AI and drive consistent AI value in your org

Get the full report

Learn more about Microsoft’s responsible AI work 

At Microsoft, we’re committed to the responsible advancement and use of AI. Our approach is guided by principles that ensure AI development maximizes benefits and minimizes potential harms. We incorporate responsible AI practices from the beginning by training our employees to evaluate risks and collaborating with experts to review and test technologies. 

We believe that advancing safe, secure, and trustworthy AI requires a mix of industry commitments, policies, and global governance. Responsible AI is an ongoing journey that involves continuous learning and collaboration.

Sarah Bird is at the forefront of ensuring that AI technologies are developed and deployed responsibly, and her team is dedicated to building tools that test AI systems rigorously to ensure they work as intended and are safe, inclusive, and beneficial for everyone. As she highlights, by integrating responsible AI practices from the start, we can unlock the full potential of AI while maintaining the highest standards of safety and innovation. 

Want to learn more?  

• Are you attending SXSW? Check out Sarah’s panel on Wednesday, March 12: Building Trustworthy AI: Evolving Safety Practices for GenAI. 
• Visit our Understanding AI at Microsoft page to catch up on the latest content from our Building AI Responsibly blog series.
• Read Grow Your Business with AI You Can Trust to discover how safe and secure AI business practices can enhance performance and effectively drive impact.

The post FYAI: The role of responsible AI with Microsoft CPO Sarah Bird appeared first on The Microsoft Cloud Blog.