Gemini. Turn everyday Google work into a clear process.

Three product routes, three implementation questions

Workspace features extend existing work interfaces. Google documents Gemini Enterprise as separately licensed, with edition-specific contractual frameworks. The Gemini API serves custom applications. We therefore start with a product record: required interface, authorised users, sources, contractual basis and operation. That turns a brand name into an explainable decision.

Review feature availability and data access separately

Enabling a Gemini feature does not by itself define which information a workflow should use. Google distinguishes feature access from Workspace Intelligence source settings. Disabling active search does not automatically exclude explicitly referenced content. For a pilot, name permitted sources and test access with actual user roles.

Assess privacy against the service actually used

Google’s commitments for qualifying Workspace services exclude model training on customer data without permission. They do not establish uniform retention across every Gemini interface. The Privacy Hub distinguishes history and retention by service; third-party apps have their own terms. We therefore document the service, history settings and onward data path for each use case.

Example: a project update from approved sources

A project team selects an approved Drive folder, a status template and an accountable owner. The pilot creates a short overview of decisions, open issues and next steps. The owner checks references and supplies missing context. Include conflicting versions and a document the test role should not access. The output remains a draft until accepted.

Start with one working process

The AI System Check establishes whether the bottleneck is search, preparation or hand-off. It leads to a bounded starting point: a Workspace workflow, a Studio flow or, where appropriate, a custom application. Further expansion makes sense once sources, acceptance and ongoing effort are manageable. Other providers remain part of the evaluation.

Current Gemini models and new ways of working

As of September 23, 2026, current releases include Gemini 3.8 Flash, new Live models and Flash TTS, alongside the Windows app and additional Workspace MCP connectors. Model, surface and access remain separate decisions: app access does not establish API access, and an API model does not establish an identical Workspace feature. The initial Workspace rollout of notebooks in the Gemini app excludes the EEA; Spark currently excludes work accounts and Germany. We therefore select an actually available route that fits the working objective.

Decision matrix