Microsoft: AI ‘Business Agents’ Will Kill SaaS by 2030
When Microsoft CEO Satya Nadella first indicated that business Software as a Service (SaaS) applications are dead last December, it sent shockwaves through the enterprise software world. Now, Charles Lamanna, Microsoft’s corporate vice president leading business applications and platforms, is doubling down on this vision with a timeline and roadmap that is both ambitious and controversial.
In a recent conversation on the Madrona VC firm’s Founded and Funded podcast with Madrona Managing Director S. Somasegar, Lamanna did not mince words. He said traditional business applications will become the mainframes of the 2030s: still running, still consuming budgets, but ossified relics of a bygone era. The future, according to Microsoft’s leadership, belongs to AI agents.
The Anatomy of Disruption
To understand why Microsoft believes business applications are facing extinction, Lamanna breaks down what these applications have traditionally been: form-driven interfaces for data entry, static workflows that rigidly define business processes and relational databases storing structured information. It’s a model that, as Lamanna points out, has not fundamentally changed since the mainframe era, despite decades of technological evolution.
“If you go and look at a biz app that ran on a mainframe, it looks remarkably similar to a web-based biz app of today,” Lamanna said in the podcast. “That’s not going to be true in 10 years.”
The replacement? What Microsoft calls “business agents”: AI-powered entities featuring generative AI (GenAI) user interfaces that dynamically adapt to user needs, goal-oriented agents that find optimal paths rather than following predetermined workflows and vector databases designed for AI-native operations.
The Timeline: Revolution or Evolution?
Lamanna’s timeline is aggressive: The new patterns will be clearly codified within 6-18 months, with mainstream adoption by 2030. It’s a prediction that has industry watchers divided.
Rocky Lhotka, a Microsoft MVP and vice president of strategy at Xebia, expressed skepticism about the 2030 timeline. “Very forward-looking and optimistic in my view,” he told The New Stack, pointing out that businesses with significant capital investments — manufacturing, transportation, construction — cannot simply “throw out their existing employees, machines and other equipment and replace that with virtual agents.”
Meanwhile, Mary Jo Foley, editor in chief at Directions on Microsoft, indicated there is a less idealistic way for Microsoft to get to its goal.
Microsoft will fall back on its “existing playbook of making agents the next wave of paid add-ons” for Dynamics and Office apps. This would mean additional subscriptions on top of existing payments — a strategy that increases average revenue per user while gradually acclimating customers to the agent model, she said.
“Business apps as we know them are dead,” Foley told The New Stack. “That’s the new trendy message from Microsoft, Salesforce and other companies who are now focused on making ‘agentic’ a thing. But turning legacy ERP, CRM [customer relationship management], HRM and Office apps into ‘agent-native’ platforms — whatever that really ends up meaning — is going to be a long and painful process, if it ever really happens.”
The Real-World Implementation Challenge
However, the vision of agent-native platforms faces significant hurdles. As Foley points out, “Replacing forms and dashboards with natural language interfaces is one thing, but changing existing business workflows into a bunch of interconnected agents is another, especially when you have to worry about supporting and migrating large, legacy customers and workloads.”
Richard Campbell, founder of Campbell & Associates and longtime Microsoft MVP, offers a more nuanced view. He said it is not about replacing applications but reimagining them entirely.
Using the example of CRM systems, Campbell asked: “If you have an LLM [large language model] with access to your [Microsoft] Teams and email interactions with customers, isn’t it able to effectively act as the CRM on demand?”
This is about fundamentally rethinking what software even means in an AI-first world.
The Organizational Transformation
Lamanna’s vision is not just technological but also organizational. He predicts a fundamental restructuring of how companies operate. Instead of experts in narrow domains, workers will become generalists supported by expert AI agents.
As Lamanna explained, based on his own situation, “I have an agent which helps me with sales research. I’m not a salesperson, I’m an engineer, but I don’t have to go talk to a salesperson to get ready for a customer meeting.”
Moreover, traditional departmental boundaries will dissolve. “Maybe sales, marketing and customer support all become one role, and one person does all three,” Lamanna said.
And human-AI teams will emerge as the very definition of a team will change.
“A team is a group of people and AI agents,” Lamanna said. “That’s really how we need to start thinking about how we organize organizations and companies.”
The Determinism Dilemma
However, not everyone is convinced this transformation will be smooth, or even desirable. Lhotka raises concerns about determinism and innovation.
“Today’s LLM models aren’t deterministic,” Lhotka pointed out, “but accounting and inventory and many other business concepts are very deterministic and have very discrete rules to ensure the software mirrors the real world. It isn’t clear to me, at least not yet, how well LLMs are going to bridge the gap between a virtual world where nondeterministic outcomes can be tolerated, and, for example, filling a truck with gravel, where nondeterminism might literally crush the truck.”
Lhotka also noted that there is a risk of “ossification” in a different form.
“If most business functions are run by agents, the result will be ossification,” he said. “Business innovation will cease, because LLMs don’t innovate. They aren’t creative.”
This could, paradoxically, create opportunities for “human-first” companies that can innovate while their AI-first competitors stagnate, Lhotka said.
The Industry Convergence
Industry convergence around open standards is key to Microsoft’s vision. Lamanna notes that protocols like Model Context Protocol (MCP) and Agent2Agent Protocol (A2A) are seeing adoption rates not seen since the early days of the web with HTML and HTTP.
“There is a tremendous amount of industry consolidating,” Madrona’s Somasegar said on the podcast. “In fact, the thing that surprised me is the case of Anthropic. They came out with MCP, and within a few months, pretty much anybody that mattered talked about how they are all-in on supporting MCP and came out with their own offerings. That level of industry consolidation around something is both, I think, exciting and fantastic.”
Indeed, “It’s probably like 30 years since we’ve had such an industry-wide convergence on an open standard,” Lamanna noted. Microsoft has adopted these standards and is contributing improvements back to the community.
Meanwhile, Brad Shimmin, an analyst at The Futurum Group, said he sees this convergence as potentially liberating for businesses.
“From the vantage point of business users and companies as a whole, ditching that yoke of complexity and lock-in must hold some appeal,” he told The New Stack.
However, “Will we do away with Microsoft Excel in favor of a chat interface capable of ingesting, processing and analyzing data behind the scenes through the magic of MCP- and A2A-infused agentic workflows?” Shimmin asked. “Will this do away with the need for practitioners and ISV partners to build software, such as plug-ins or extensions to software packages that simply no longer exist?”
The Path Forward: Three Keys to Success
For enterprises looking to navigate this transformation, Lamanna offers three critical success factors based on patterns observed across Microsoft’s customer base:
- Resource constraints: Successful companies deliberately create budget pressure to drive genuine productivity improvements rather than incremental changes.
- Democratization: “All of your users, no matter where they are, technical, nontechnical, need to be picking up and using these tools each and every day,” Lamanna insists. Companies that restrict AI to technical teams or pilot projects fail to transform. “Companies which are struggling are companies that don’t have AI in everybody’s hands every day,” he said.
- Focus: Rather than spreading efforts across hundreds of initiatives, successful companies “do five projects very well, with a lot of force and with continuous improvement in mind,” Lamanna said.
Apps Evolving or Agents Replacing?
Andrew Brust, CEO of Blue Badge Insights and Microsoft MVP, poses perhaps the most fundamental question: “Will agents replace apps … or will apps evolve into agents?”
The answer may be both — and neither. As Campbell said, we may be moving toward a world where “it makes it really hard to point at anything and call it an app. Suddenly, that’s an old idea.”
Instead of sovereign applications like ERP systems, we’re heading toward a landscape of data stores and dynamic interaction tools, where governance shifts from applications as gatekeepers to data itself being tagged for sensitivity and access privileges, Campbell said.
The Bottom Line
Microsoft’s vision of agent-native business platforms represents either the most significant transformation in enterprise software since the advent of the internet, or an overly optimistic prediction that underestimates the inertia of enterprise IT.
The question is no longer whether AI will transform business applications, but how quickly and completely will this transformation occur. Lamanna’s prediction that by 2030, agent-based systems will be the prevailing pattern may be optimistic, but the direction seems inevitable.
He warns that companies need to choose whether they want to “be the people that watch that happen” or “the people that do it to ourselves.” In a world where startups are already operating with AI agents as core team members, waiting for certainty may mean waiting too long.
Whether the transformation is completed by 2030 or takes another decade, the enterprise software landscape of 2035 will not look like today’s, and Microsoft is betting its business applications’ future on being the company that kills its own products before someone else does.
