Everyday productivity: Gemini in Google Workspace, Claude Projects and automation under control

1Gemini in Google Workspace

For organisations using Google Workspace, Gemini is integrated, depending on the subscription, into everyday tools (features available in November 2026, subject to change):

• Gmail: summarising long threads, suggesting replies, searching the mailbox.
• Docs: drafting and rewriting, summarising a document, help with formatting.
• Sheets: help with formulas, analysing and organising data.
• Meet: meeting notes and summaries (informing participants).
• Drive: questions about stored documents, within the limits of the user's access rights.

Advantage: data stays within the organisation's environment, with the commitments of the professional plan. Point of attention: the AI sees what the user can see. If your folder sharing rights are too broad, the AI can surface sensitive information to people who should not read it. Clean up sharing before deploying.

2Claude day to day

With Claude (web, desktop and mobile apps), frequent professional uses are:

• Projects per file (a proposal, a country, a function) with reference documents and standing instructions; shareable with the team depending on the plan;
• Artifacts: documents, tables, small interactive tools or pages created in the conversation, which can be edited and reused;
• File analysis and file creation (documents, spreadsheets, presentations, depending on the version);
• Connectors to services such as Google Drive, Gmail or other tools, depending on the plan;
• for technical profiles, Claude Code, a tool that helps write and modify code, useful for analysis scripts or small internal tools, always with review.

With both assistants, productivity comes less from the tool than from the method: good briefs (week 2), well-organised reference documents, a request library and systematic checking.

3Saving time on routine tasks

A few proven uses, with checking time to plan for:

• Emails: reply from your key points; diplomatic rephrasing of a delicate message; summary of a long thread before a meeting.
• Terms of reference and job descriptions from an approved organisational template.
• Presentations: outline and key messages from a report.
• Planning: breaking a project down into tasks, a simple Gantt chart, a pre-mission checklist.
• Learning: explaining a concept ("explain the difference between resilience and early recovery with examples"), self-assessment quizzes, preparing for a new role.

Measuring the real gain. A 60-minute task done in 20 minutes with AI, plus 15 minutes of checking, saves 25 minutes, or about 42%. If checking is skipped and an error then has to be corrected with a donor, the "gain" becomes a loss.

4Agents and automation: promises and safeguards

Providers are developing agents: assistants able to chain actions (searching, reading files, filling in a form, sending a message, using a browser or a computer) with less human intervention. It is powerful, and risky.

Essential safeguards:

• Human approval before any binding action: external sending, payment, publication, deletion, modification of beneficiary data.
• Minimum access: the agent only has access to what its task requires.
• Log of actions taken, available for review.
• Testing on fictitious cases before any real use.
• A designated owner for each automation.
• Fallback plan: the task must be possible manually if the tool fails or changes.

A good first use case: an automation that prepares (draft weekly report from a table, sorting incoming emails by topic) and leaves the human to validate. A bad first case: an automation that sends messages to beneficiaries on its own.

5AI and your career

Humanitarian recruiters receive more and more CVs and letters generated by AI, often identical and without substance. To stand out:

• CV: AI can help structure, tailor to a job offer and phrase quantified achievements, based on your real experience. Never invented experience, degrees or skills: that is fraud, and reference checks uncover it.
• Cover letter: start from your reasons and your examples; then ask for improvements to structure and style. A generic letter is quickly spotted.
• Interview preparation: interview simulation ("ask me 10 competency-based interview questions for a MEAL coordinator position, one by one, then critique my answers using the STAR method"); this is one of the most useful uses.
• Written tests: if the organisation prohibits AI during a test, respect the rule.

This Mini MBA and its diploma demonstrate a real skill: knowing how to use AI productively, responsibly and in line with humanitarian principles.

6Staying up to date without burning out

Tools change every few months. A few tips:

• follow the providers' official announcements and the resources of sector organisations that publish on humanitarian AI (learning networks, inter-agency working groups);
• test a new feature on your real tasks before changing your practices;
• be wary of spectacular promises and "secret tricks";
• share your discoveries in the team's request library;
• keep time for work away from screens: relationships, field, reflection.