Dropbox and Box Could Become the Next Kings of Systems of Record

Today’s systems of record, Salesforce, Workday, and the like, were built for humans. They feature complex UIs layered on top of complex data processing, with a database buried underneath. This architecture is deeply unfriendly to AI agents. That’s why, even as agents can generate entire websites, slide decks, and videos in one shot, they still struggle to accurately update a CRM or HR system.

The file system is AI’s natural language LLMs were trained on file systems. Recent developments show that AI agents process file-based data far more naturally than they navigate traditional enterprise software. And file systems come with a built-in permissions model: subfolders inherit and restrict access from parent folders, naturally mirroring how companies work—CEO at the top, each org head scoped to their domain. This elegantly solves governance, safety, and guardrail challenges. With this foundation, I believe we could build AI-native systems of record for HR, sales, and beyond within months—not years. The obvious pushback: existing systems encode complex compliance and governance rules. True. But rules are rules—they can be implemented in AI-native ways too. Adoption won’t happen overnight, but as people experience how much easier and more efficient these new systems are, the shift will follow the classic pattern: slowly, then all at once.

This is where Box and Dropbox have a massive advantage They already own the hardest part—file-system-based storage deployed across enterprises. They already sit on every employee’s machine. All they need is an AI agent layer that consumes their existing data. The end state is striking in its simplicity: Agent as the interface customers interact with Their existing file-system storage as the database No bloated middleware in between. That’s the new SaaS and it’s not going away before LLM itself can diagest info realtime.

But why not Microsoft, Google, or Apple? They have OneDrive, Google Drive, and iCloud. The answer is simple: are they willing to cannibalize their existing cash cows to birth something new? I doubt it. That’s the classic incumbent’s dilemma. And honestly, the same risk applies to Box and Dropbox. It may well be an unknown startup that comes out of nowhere and finishes the last mile.