The Opportunity Emergence: Box CEO Reveals Why AI Creates Entirely New Software Categories

"Now is the moment. There's a very very long list of things that software never did before that AI agents are perfectly primed to go do now."

— Aaron Levie, CEO of Box

The enterprise software landscape has fundamentally shifted, and the organizations that recognize this transformation are building the next generation of billion-dollar companies. Box CEO Aaron Levie's recent analysis at Y Combinator reveals why we're experiencing the most significant category creation opportunity in over a decade.

In a candid discussion about market timing and startup opportunities, Levie outlined why AI hasn't just improved existing software—it's made entirely new categories of problems solvable for the first time in enterprise software history.

The Solved Problem Landscape of 2022

Three years ago, the startup landscape presented a challenging reality. Levie conducted what he describes as a systematic analysis of core business functions—examining whether fundamental enterprise needs had been addressed by existing solutions.

The results were sobering for entrepreneurs.

"Fast forward to 2022, we've solved a lot of problems," Levie observed. "When I want food, it comes in 20 minutes from DoorDash. When I want to listen to music, it's on Spotify." The same pattern held across enterprise software: payroll, CRM, email, calendar—every core business function had established incumbents or well-funded startups attacking the space.

This created what Levie calls a "tough environment for startups" where founders were "really only able to do derivative things because the core functions had been solved." The market had reached a saturation point where competing with established players required fighting uphill battles against modern, well-executed solutions.

The fundamental challenge: Traditional software categories had been claimed by companies that understood how to build modern technology and were actively serving those markets.

AI as Category Creator

The emergence of AI agents changes this equation entirely. Levie argues that artificial intelligence doesn't just make existing software better—it enables solving problems that were previously impossible to address with software.

"There's a very very long list of things that software never did before that AI agents are perfectly primed to go do now," Levie explained. "And that's basically the opportunity set."

The distinction is critical. The opportunity isn't in building "CRM but with AI" because established players like Salesforce will integrate AI into their existing platforms. Instead, the opportunity lies in identifying categories of professional services or work that have no incumbent technology solutions—areas where AI agents can deliver value that was previously only possible through human expertise.

This creates what Levie describes as entirely new categories where startups can establish market leadership before incumbents recognize and respond to the opportunity.

The Economics of Intelligence-Driven Business Models

Traditional SaaS companies faced a fundamental limitation: revenue was capped by the number of human users who needed software licenses. If you sold legal software, your addressable market was limited by the number of lawyers at each prospective customer.

AI agents eliminate this constraint entirely.

"All of a sudden you can have AI agents that effectively contain the labor of that job function in the software itself," Levie noted. "You can go to a company and you can say, 'I know you only have three lawyers, but my agents could do the amount of work of basically unlimited lawyers.'"

This shift enables consumption-based business models where pricing aligns with value delivered rather than user count. A process that costs $5-10 per transaction with human labor might cost 10 cents with AI agents—allowing companies to charge $2 per transaction while creating massive value for customers and maintaining healthy profit margins.

The new economics reward companies that can build substantial software capabilities above the commodity AI layer. Success belongs to organizations that create workflow software, specialized context, and unique integrations that command premium pricing over raw AI inference costs.

The Architectural Intelligence Opportunity

This framework illuminates why certain emerging categories represent massive market opportunities. Consider the challenge of architectural understanding in software organizations—a problem affecting hundreds of thousands of professionals that traditional software has never adequately addressed.

Enterprise architecture tools were built for specialists. They produce complex diagrams and documentation that exclude the majority of stakeholders who need architectural context for effective decision-making. Business analysts struggle to scope requirements without understanding system boundaries. Product managers cannot assess technical feasibility independently. Developers inherit systems without adequate context about architectural decisions.

Until recently, there was no technological path to make architectural knowledge accessible across organizational roles. The problem required understanding unstructured information—documentation, conversations, decision records, code relationships—and translating complex technical concepts into role-appropriate explanations.

AI agents make this category possible for the first time. They can process unstructured architectural information and provide context-aware intelligence tailored to each stakeholder's expertise level and decision-making needs.

This represents precisely the type of new category Levie describes: architectural intelligence for non-specialists. It's not improved architecture documentation—it's making architectural understanding accessible to organizational roles that were previously excluded from technical decision-making.

Why Companies Won't Build This Internally

Levie addressed a common concern about AI-enabled categories: won't companies simply build their own agents instead of purchasing software solutions?

His response centers on the "core vs. context" framework developed by Geoffrey Moore. Every organization must distinguish between capabilities that are core to their competitive advantage versus contextual functions necessary for operation.

"Just because you can do something, the vast majority of the world doesn't end up doing something," Levie observed. Most companies want to focus resources on their differentiating capabilities rather than building custom solutions for standard business processes.

For most organizations, architectural understanding represents a contextual function. While essential for making informed decisions, building custom AI systems to democratize architectural knowledge isn't core to their primary value creation.

Companies prefer purchasing reliable solutions from specialized vendors rather than maintaining custom software that requires ongoing development, debugging, and support for non-core functions.

Strategic Principles for Category Creation

Levie offered specific guidance for founders looking to capitalize on AI-enabled category creation:

Ensure Fundamental Market Transformation "Make sure that you're going after a market where AI fundamentally changes the very economics or actual process of that thing." Success requires targeting problems where AI doesn't just improve existing solutions but enables entirely new approaches.

Build Substantial Software Above AI The most successful companies will create significant workflow software, specialized integrations, and unique capabilities above commodity AI services. Raw AI inference becomes a small component of the total value delivered to customers.

Target Market Tailwinds "Go after markets where AI like fundamentally changes like the very economics or or or you know, actual process of that thing." Choose categories where AI creates structural advantages rather than marginal improvements.

Focus on Enterprise Context Functions The highest-value opportunities exist in areas that are essential to business operations but not core competitive differentiators for most companies.

The Deflationary Advantage

Software entrepreneurs benefit from what Levie calls "deflationary economics on the supply side." Core infrastructure costs—compute, storage, AI inference—decrease over time while customer willingness to pay remains stable.

"You don't have to basically raise your prices in perpetuity like many other industries. You can actually just get more efficiency gains over time," Levie explained.

This creates sustainable competitive advantages for companies that build substantial capabilities above commodity infrastructure layers. Customer pricing remains stable while underlying costs decrease, improving unit economics over time.

The key is avoiding pricing that customers perceive as "offensive" while building enough value above raw infrastructure to maintain healthy margins as underlying costs decline.

Aligning Vision with Market Reality

Levie's analysis validates what many forward-thinking entrepreneurs are already experiencing: this moment represents a fundamental shift in enterprise software possibilities. AI hasn't just made existing solutions better—it's created entirely new categories of problems that software can solve for the first time.

"There will be 100 startups that get created between last year and in three years from now that all become 5, 10, 20 billion dollar companies because they're able to find the next set of categories," Levie predicted.

This perspective aligns with our own observations in the architectural intelligence space. The problems we're seeing—business analysts unable to scope requirements without architectural context, product managers making roadmap decisions without technical constraint visibility, developers inheriting systems without understanding—these aren't issues that traditional enterprise architecture tools were designed to solve.

The companies defining the next decade are identifying problems that were previously solvable only through human expertise and creating AI-powered solutions that deliver that expertise as software. Success belongs to founders who can identify new categories in their domains—areas where AI agents can solve problems that traditional software never addressed.

The opportunity exists now because established incumbents are still figuring out how to integrate AI into their current products, creating space for new companies to define entirely new categories before the giants respond.

The question for enterprise software founders isn't whether AI will transform their industry—it's whether they'll identify and capture the new categories that AI makes possible for the first time in software history.

---

Aaron Levie's complete discussion at Y Combinator provides extensive insights into AI business models, competitive strategy, and market timing. The full conversation is available through Y Combinator's content channels.