Google Workspace: Gmail, Calendar, Drive, and Docs as AI source

What are the Google Workspace APIs?

Google Workspace (formerly G Suite) is Googles office platform, with over 9 million paying customers worldwide as of May 2026. The Workspace APIs are a collection of separate REST interfaces, each responsible for one service: Gmail API, Google Calendar API, Google Drive API, Google Docs API, Google Sheets API, Admin SDK, Apps Script API. In contrast to Microsoft Graph, there is no unified endpoint but a dedicated API per service.

All APIs use OAuth 2.0 with Google as the identity provider. Authentication runs either via user consent (delegated access, the user clicks through the consent screen) or via service accounts with domain-wide delegation (background access on behalf of users, approved by the Workspace admin). The service-account variant is the right choice for AI pipelines.

Rate limits are specified per API: Gmail allows 250 quota units per second per user (a mail send operation costs 100 units, a mail read 5 units). Drive allows 20,000 queries per 100 seconds per project. Calendar sits at around 600 queries per minute per user.

May 2026 brings two GA news. First: AppSheet integration via Workspace APIs is stable and available, which enables low-code AI apps with Workspace data. Second: Vertex AI Connect is a new mechanism that can forward Workspace data directly to Vertex AI models without the data leaving the Google Cloud perimeter. That is an important step for privacy-sensitive workflows.

Why it matters for Swiss fiduciary

Google Workspace is less dominant in Switzerland than Microsoft 365 but very strong in two customer profiles. First: young SMEs, startups, and creative-driven industries (agencies, software houses, architecture firms) prefer Workspace for easier collaboration. Second: subsidiary mandates of US corporations often use Workspace because the parent prescribes it.

For fiduciary offices that serve such mandates the Workspace connection is mandatory. The AI use-cases overlap with Microsoft 365, but there are accent shifts. Example Gmail auto-label: instead of only doing triage, the AI can set labels on mails directly via Workspace APIs (Gmail-specific concept that does not exist in Outlook the same way). Client correspondence is automatically sorted by topic: "Tax 2026", "Payroll Q2", "Client onboarding".

Example document search: Drive offers its own full-text search, but it is not semantic. A RAG layer over Drive content (PDF contracts, sheets with master data, docs with minutes) gives fiduciaries semantic search: "Which client had a similar problem with withholding-tax correction in 2024?" The LLM answers with source citation and Drive link.

How it works

Entry happens in five steps: create a Google Cloud project, enable APIs, create a service account, have the Workspace admin grant domain-wide delegation, query endpoints.

The Google Cloud project is created in the Google Cloud Console. There you enable the needed APIs (Gmail API, Drive API, Calendar API). The service account receives a JSON key file that you use as credentials. In the Workspace admin centre the service account is approved for domain-wide delegation, with the OAuth scopes it may use (e.g. https://www.googleapis.com/auth/gmail.readonly).

An example call to read the latest mails:

```bash # Get token from service account (Google Auth Library helps a lot here) ACCESS_TOKEN=$(gcloud auth print-access-token --impersonate-service-account=ki-bot@project.iam.gserviceaccount.com)

# Read latest mails from [email protected] curl -X GET "https://gmail.googleapis.com/gmail/v1/users/[email protected]/messages?maxResults=10" \ -H "Authorization: Bearer $ACCESS_TOKEN" ```

The response delivers message IDs; with a second call per message you fetch subject, from, snippet, and body. For AI workflows the batch endpoint is useful, which bundles multiple requests into one HTTP call.

For Drive you use the files API with a search query:

```bash curl -X GET "https://www.googleapis.com/drive/v3/files?q=mimeType=\"application/pdf\"&pageSize=100" \ -H "Authorization: Bearer $ACCESS_TOKEN" ```

Files are downloaded via /files/{id}/export or /files/{id}?alt=media as bytes. For Google Docs you must specify the format via export (e.g. text/plain or application/pdf).

For change notifications Google uses the Drive Push API (watch channels) with semantics similar to Microsoft Change Notifications. Channels are valid 24 hours and must be renewed.

Google Workspace integration in 5 steps

1. 01Create the Google Cloud project, enable the required APIs (Gmail, Drive, Calendar), check quotas.
2. 02Create the service account, store the JSON key file securely in a secrets manager.
3. 03Ask the Workspace admin for domain-wide delegation, keep OAuth scopes minimal (principle of least privilege).
4. 04Set up the read pipeline: fetch Gmail messages, Drive files, Calendar events via the service account, pagination via pageToken.
5. 05Set up watch channels for push notifications, renew channels every 23 hours, set up a validated receiving endpoint.

When to use

The Workspace integration is worthwhile when (a) Workspace is the primary office platform of the firm or an important mandate and (b) at least two AI use-cases run in parallel. Typical use-cases: Gmail auto-label, Drive RAG, Calendar briefing.

For fiduciary offices with mixed mandates (50 percent Microsoft, 50 percent Google) you integrate both APIs through an abstracted data-access layer. The LLM answers uniformly, regardless of the source platform.

Google Workspace also offers a special strength: Apps Script. A lightweight scripting framework within the Workspace docs that can launch AI calls directly from a Google Sheet or Doc. For quick prototypes and client demos Apps Script is unbeatable.

When not to use

If the firm and all mandates use exclusively Microsoft 365, the Workspace integration is superfluous.

For highly sensitive data (FINMA-supervised asset management, professional secrecy under Art. 321 SCC in serious cases) the Workspace standard integration is not sufficient. Google processes some Workspace data in US data centres; pure EU/CH processing requires Workspace Premium with data residency and additionally Vertex AI with EU region. Carefully review the FADP implications before the integration.

For one-time data exports (such as archiving an old Drive library) the API is usable, but Google Takeout (Workspaces own export tool) is simpler and faster here.

Trade-offs

FAQ

Related topics

Sources

1. Google Workspace Developers: API overview · 2026-05
2. Google Workspace: Vertex AI Connect (GA, May 2026) · 2026-05
3. Gmail API usage limits · 2026-04
4. Google Cloud Service Account documentation · 2026-05