Microsoft and the Rise of the Full-Stack Builder
Just a few years ago, the notion of “fusion dev teams” was a key part of Microsoft’s strategy to empower developers at all levels of expertise to collaborate to build applications.
The concept was simple: Combine business domain experts with professional developers to create applications that solved real business problems. Fusion dev team members would use tools like Microsoft’s low-code/no-code Power Platform, as well as pro-code tools like Visual Studio and Visual Studio Code.
That vision has evolved. Amanda Silver, corporate vice president (CVP) of product in Microsoft’s Developer Division, told The New Stack she is witnessing the emergence of something a bit different. She calls it the “full-stack builder” — a new paradigm where business experts can directly modify applications using natural language, without needing to learn technical platforms or programming concepts.
“The vision for that is evolving with the introduction of these agentic capabilities overall,” Silver said. “What we now see as being kind of more the mode is when you think about the new applications that are going to get built, you have to kind of think about the pre-AI way of building applications, and the post-AI way of building applications. And even further, the agentic way of building applications.”
The Limitations of Traditional Fusion Teams
With fusion dev teams, the model was to give business users the tools to build applications themselves, as they understood requirements better than anyone.
Ryan Cunningham, CVP of Microsoft’s Power Platform, describes the core problem that fusion teams were trying to solve: “It’s sort of traditionally taking all the expertise out of the head of a finance operator or HR person and putting it in the head of a software engineer and hoping that the right thing comes out. The other side is a slow and expensive process.”
The approach had definite benefits. Fusion teams could reduce the communication gap between business and IT, speed up development cycles for simple applications and free professional developers to work on more complex problems.
But there were some limitations, including platform constraints when business requirements grew complex, and integrating with existing systems, advanced business logic and delivering custom user experiences often required professional development.
Also, as business users became “citizen developers,” they were stuck within the constraints of particular platforms, limiting both their capabilities and the transferability of their skills, Silver said.
Enter the Full-Stack Builder
The emergence of AI agents is changing this equation fundamentally. Instead of training business users to think like developers, AI systems can understand business language and translate it into technical implementation.
“We’re starting to think about this and call this kind of like more of a full-stack builder,” Silver explained. “This notion that if you architect the engineering system and the application right, that you could actually have somebody who may not be as code-centric and is more an expert in the business domain or the usage of the application, can actually come in and describe to a tool like a GitHub Copilot how they want the application to change, or the user interface to change.”
In a prior interview with The New Stack, Cunningham’s description of fusion dev teams indicated that the full-stack builder paradigm is basically a continuation of fusion team principles but with enhanced capabilities. “A lot of our most successful customers are really embedding tech people with businesspeople. And using the Power Platform because it’s a place where both of them can work and be productive in the same toolkit,” he explained.
The key difference is that, rather than constraining business users to work within platform limitations, the full-stack builder approach requires engineering teams to create systems that can understand and respond to natural language business requirements.
“Based on the engineering system and all kinds of dimensions of scaffolding around how the application is built that we provide as context to the AI, we can actually make more progress more quickly on those kinds of changes to the application,” Silver said.
But the real breakthrough comes in how this solves the core fusion team challenge. “It’s hard to teach a businessperson how to build scalable, secure enterprise software. It’s hard to teach a software developer how to operate a business, but if I can put them both on the same toolkit, they can do amazing, magical things together,” Cunningham said.
“In my own work, I’ve seen how AI is collapsing the walls between roles,” wrote Amit Gupte, a Microsoft full-stack program manager, in a blog post entitled “The Rise of the Full-Stack Product Builder: AI’s Blueprint for the Future of Work.” “Product Managers, Engineers, and Data Scientists are no longer operating in silos. With AI, a single person can now ideate, prototype, and validate—tasks that once required a full cross-functional team. That’s not a threat—it’s an opportunity.”
Meanwhile, also in a blog post, Krishna Mehra, AI partner at Elevation Capital, wrote: “At the center of this paradigm shift is a new archetype: the Full-Stack Builder. These are individuals who take end-to-end ownership of projects, leveraging AI to move seamlessly from idea to execution without waiting on handoffs.”
He added: “The rise of full-stack builders is proving that smart problem solvers using AI can cover what used to take multiple specialists — without the friction. The new wave is leaner, faster, and more adaptable.”
The Technical Foundation: Architecture for Business Users
The full-stack builder model requires significant upfront investment in what Silver calls “engineering systems and context.” This is not just about adding AI to existing applications — it requires rethinking how applications are architected to support natural language modification.
Applications must be designed so that AI agents can safely modify components without breaking other parts of the system. This requires clear boundaries, well-defined interfaces and comprehensive testing frameworks. The agents also need context about why certain decisions are made, not just how they are implemented in code.
Dynamic Workflows Replace Static Processes
One of the most significant changes in the full-stack builder model is that the new model enables dynamic, AI-driven workflows that can adapt to business requirements.
“Now, with agentic capabilities, a lot of those things are much easier to create,” Silver said. “Even just creating the workflow application is much easier to model with agentic capabilities, and further, some of the aspects of the workflow that used to require human intervention now can be completed using agentic capabilities.”
This means business experts can describe processes in their own terms in natural language, and AI agents can handle the complex logic that previously required either simplified workflows or developer intervention.
Real-World Applications
Silver emphasizes that the full-stack builder model scales across different levels of application complexity.
“When you think about these new kinds of applications that are getting built, there are different levels of sophistication for those different kinds of applications,” she said. “Some require just simple prompts. Some require much more sophisticated, agentic application authoring to complete the capability.”
Silver’s team has implemented this approach internally. They have built “agentic applications that allow us to go and listen to all of those different channels and aggregate and get a much better sense of what is the reaction to those kinds of capabilities that we introduced,” she said.
This demonstrates how business requirements (such as understanding customer sentiment) can be translated directly into technical implementation without requiring business users to learn development skills, she said.
The Democratization Beyond Traditional IT
The implications extend beyond traditional enterprise IT departments. Silver argues that the full-stack builder model represents a democratization of technical capability.
“We’re bringing more people who have maybe less formal technical background and making them more capable of building various different kinds of solutions,” she explained. “Then the other way that it’s evolving is that we’re really evolving the engineering systems and the context that we build up in the way that the application is described, that allows us to use more natural language prompts to make changes to the application over time.”
Cunningham said this helps to expand software development opportunities.
“There’s just a lot of software, particularly inside of companies and organizations, there’s a lot of places where we just haven’t put the effort in over the last couple decades. Like it just hasn’t made sense to put a traditional full-stack dev team on an internal invoicing tool … Now it’s possible to go do really professional-grade software development and innovation on the cutting edge for sort of the long tail of scenarios that were not previously getting served by software engineers, and really do it with people that actually understand intimately what they need.”
This means the boundary between business and technical roles will blur, but in a different way than traditional fusion teams attempted. Instead of making business users think like developers, the full-stack builder model makes technical systems understand business language.
Challenges and Implementation Considerations
Despite the promise, implementing the full-stack builder model isn’t without challenges. One is that building systems that can safely respond to natural language business requirements is more complex than traditional application development. Engineering teams need new skills in AI integration, natural language processing and context management.
Other challenges include ensuring quality and consistency of applications, and governance and control, as traditional IT governance models assume that technical changes go through developer review processes. The full-stack builder model requires new governance frameworks that can maintain control while enabling business user autonomy. There are also change management and security implications to consider.
“Giving more people better tools will breed more creativity, will breed more creation,” Cunningham said. “It will come with a new set of challenges, no doubt, but I think it’s actually just a really optimistic time to be in tech. And I think we can invite a lot more people in, in a good way if we do it right.”
Looking Forward: The Future of Business-Developer Collaboration
Silver said she envisions the full-stack builder model as part of a broader transformation in how software gets built. Rather than replacing developers, it changes the nature of developer work and expands the pool of people who can contribute to application development.
The evolution from fusion teams to full-stack builders represents more than a technological shift. By making systems understand business language rather than requiring business users to learn technical languages, organizations break the constraints of traditional development processes.
Yet, professional developers will need new skills in AI system design, natural language processing and context management, as their role shifts from writing application code to designing systems that can generate and modify code based on business requirements.
The fusion dev team concept promised to bridge the gap between business and technology. The full-stack builder model delivers on that promise by eliminating the gap entirely.
