Brian Chesky's recent comments on making Airbnb "agent-friendly" and his call for an AI-native operating system might seem far removed from the daily grind of spreadsheet work. But they are, in fact, pointing directly at the same problem: we are asking AI to work inside tools that were built for a different era. The core insight here is that AI-native systems, whether for booking a vacation or managing a budget, demand fundamentally different architecture than the one we've been using for decades. That's why we believe the next step for spreadsheets isn't a smarter add-in or a better macro language. It's a complete rebuild from the ground up.
Consider what "agent-friendly" actually means. When Chesky talks about making Airbnb's platform navigable by AI agents, he's acknowledging that the future of user interaction won't always involve a human clicking buttons. The same logic applies to data management. If your spreadsheet is designed only for human eyes and human fingers, you're limiting what AI can do for you. This is the very tension that builders like those at Flow Engineering are grappling with, they're bringing AI agents into hardware design, a field where legacy tools are deeply entrenched. The lesson is that an agent can only be as effective as the environment it operates in. A traditional spreadsheet, with its rigid cell grid and manual formula dependencies, is a hostile environment for autonomous agents. An AI-native spreadsheet, by contrast, treats data as a fluid, queryable resource that agents can reason over without bumping into structural limitations.
This is where the practical shift matters for you. An AI-native spreadsheet doesn't just automate the tasks you already do; it changes what you can ask the tool to do. Instead of spending time cleaning data, writing VLOOKUPs, or formatting rows, you can focus on the strategic questions: "What happens if we shift our pricing model?" or "Which customer segments are showing early signs of churn?" The tool becomes a thinking partner rather than a manual labor device. This aligns with a broader trend we're seeing across the industry. The team at Restate recently secured significant funding for a durable execution engine built on its own storage layer, precisely because off-the-shelf databases weren't designed for the reliability that AI workflows demand. The message is consistent: legacy infrastructure, whether it's a database or a spreadsheet, creates friction that AI-native design eliminates.
The open question is whether the spreadsheet industry will move fast enough. We've seen why the smartest AI builders are stepping back from consumer apps, because building for the mass market often means dumbing down the capability to match the lowest common denominator of the existing tool. The spreadsheet market is huge, but it's also conservative. The specific consequence to watch is which company first ships a spreadsheet that treats AI agents as first-class citizens, not as afterthoughts. That will be the moment the category truly transforms.