Google wants Drive to stop being storage and become an AI work layer

Google wants to stop you from ever opening a folder again. The company is repositioning Google Drive from passive storage into what it calls an "active knowledge base"—a system where documents are not retrieved but interrogated. The mechanism is a bundle of AI Overviews and upgraded Gemini capabilities across Sheets and Slides, pitched as the end of manual document hunting.

The timing is conspicuous. After two years of watching Microsoft 365 build Copilot into every seam of Office, Google is finally shipping its response. The update reaches AI Pro, AI Ultra, and Gemini Alpha business subscribers in the US today. Everyone else—roughly a billion free and standard-tier Drive users—gets to watch from the outside.

The architecture matters more than the marketing. On paper, Gemini inside Drive promises something beyond semantic search: cross-document synthesis, where the model ingests multiple spreadsheets and presentations and returns a distilled answer rather than a file path. That would distinguish it from the Gemini ecosystem integrations already live in Docs, which since 2024 have produced uneven results ranging from useful summarization to confidently wrong synthesis.

Google's specific claim on Sheets performance—that 100-cell tasks will run nine times faster than manual entry—deserves scrutiny. It is a synthetic benchmark, not a measured workflow, and synthetic benchmarks have a habit of evaporating under real-world conditions with messy formatting, merged cells, and inconsistent headers. The productivity pitch is plausible; the multiplier is not yet proven.

The strategic logic is clearer than the technical one. By embedding reasoning directly where documents live, Google creates friction against migration. Switching to Microsoft becomes costlier than subscription pricing when your institutional knowledge is wrapped in a query layer that competitors cannot easily replicate.

The moat here is data proximity, not model quality. Standalone LLMs choke on context windows and hallucinate across document boundaries; Gemini has native access to the files themselves. That pipeline—raw spreadsheet to structured reasoning without intermediate export—is genuinely difficult for third-party tools to match, even with superior base models.

Whether this constitutes an architectural shift or rebranded search depends on implementation depth. The official announcement emphasizes "advanced reasoning," but the actual interface shown resembles conversational query over existing index structures. True synthesis would require the model to maintain state across sessions, reconcile conflicting data sources, and surface uncertainty rather than flattening it into confident output. Early Gemini-in-Docs behavior suggests the latter tendency persists.

For enterprise administrators, the tier gating creates a familiar calculus. The AI Alpha and business plans that unlock these features sit well above standard Workspace pricing. The bet is that knowledge-work productivity gains justify the premium; the risk is that limited rollout prevents network effects from validating the "active knowledge base" concept before competitors close the gap.

The competitive pressure is asymmetric. Microsoft's Copilot strategy has targeted similar functionality but with broader initial availability and deeper integration into Excel's formula engine. Google's counter relies on Sheets' collaborative ubiquity and the sheer gravitational pull of existing Drive archives. Neither approach has definitively solved the trust problem: users still verify AI-generated numbers manually, which partially negates the speed advantage.

What remains unaddressed is the governance layer. Cross-document synthesis amplifies existing risks—proprietary data bleeding into training loops, access controls failing at model boundaries, audit trails dissolving into prompt histories. Google's developer documentation for Workspace AI emphasizes data isolation, but enterprise security teams have heard similar promises before.

The trajectory is clear even if the timeline is not. Storage is becoming substrate; the interface layer is becoming intelligence. Whether Google executes that transition before users conclude that a smarter filing cabinet is still a filing cabinet will determine if this pivot sticks or slides into the graveyard of announced-and-forgotten Workspace features.