Tools & Products

Google Developing "Superprojects": Cross-Product Workspace Architecture Unifying Gemini, Drive, and GCP

Google is engineering Superprojects, a unified workspace system that links Gemini chat, Google Drive folders, Google Docs, and Google Cloud repositories under a shared persistent context graph. The architecture replaces isolated Gemini Projects with an ecosystem-wide collaborative canvas.

By Julian Thorne · 2026-10-06 · 10 min read

Internal architectural leaks from Google confirm the company is developing Superprojects, a cross-product orchestration layer designed to unify Gemini across the entire Google ecosystem.

Current Gemini Projects exist as isolated conversational sandboxes inside the web chat application. Users upload PDFs or reference text snippets, but those assets remain detached from live corporate documents in Google Drive or database tables in BigQuery. Superprojects eliminates these barriers.


The Persistent Context Graph: Unifying Workspace and Cloud

Superprojects centers on a cross-product context graph maintained on Google Cloud infrastructure. When an engineering team creates a Superproject, the environment links directly to target cloud resources:

┌────────────────────────────────────────────────────────────────────────┐
│                      Google Superproject Context Mesh                  │
├────────────────────────────────────────────────────────────────────────┤
│                           [ Superproject Hub ]                         │
│                    Unified Vector & Entity Index Engine                │
│                                     │                                  │
│         ┌───────────────────────────┼───────────────────────────┐      │
│         ▼                           ▼                           ▼      │
│ ┌────────────────┐         ┌────────────────┐         ┌──────────────┐ │
│ │  Google Drive  │         │   GCP / Code   │         │ Gemini Chat  │ │
│ │  Live Docs,    │         │ Cloud Run,     │         │ Collaborative│ │
│ │  Sheets, Slides│         │ BigQuery SQL,  │         │ Multi-user   │ │
│ │  Sync via OT   │         │ Artifact Reg   │         │ Prompt Mesh  │ │
│ └────────────────┘         └────────────────┘         └──────────────┘ │
└────────────────────────────────────────────────────────────────────────┘

Instead of requiring manual file re-uploads, changes committed by colleagues in Google Docs or pull requests merged in connected Git repositories are indexed into the Superproject graph automatically.


Feature Matrix: Gemini Projects vs. Superprojects

The technical evolution transforms a personal assistant sandbox into an enterprise development control plane:

Capability Current Gemini Projects Upcoming Superprojects
Document Ingestion Static file upload (PDF/TXT) Live bidirectional Drive/Docs sync
User Collaboration Single-user only Real-time multi-seat team workspace
Code & Database Access Manual copy-paste Native GCP Cloud Run & BigQuery links
Context Window Size Up to 1M tokens (Chat only) 2M tokens continuous rolling memory
Access Control None (Local chat state) Google Cloud IAM role-based bindings

Security Boundaries and Zero-Data-Leak Protocols

Enterprise customers frequently express concern that unified AI workspaces could expose confidential executive spreadsheets to junior engineering staff.

Superprojects enforces Identity-Aware Proxy (IAP) context filtering:

  1. Even if a financial model spreadsheet is linked to a corporate Superproject, Gemini strips out that data from reasoning responses if the specific team member querying the model lacks view permissions in Google Drive.
  2. In-flight token embeddings are encrypted with customer-managed encryption keys (CMEK), ensuring Google cannot use internal Superproject code for general model training.