GOOGLE WORKSPACE
Gemini, built into the apps you already use
Use Gemini directly in the Google Workspace apps where your work already happens.
GOOGLE WORKSPACE
Gmail: AI cheatsheet
Summarize a long thread without leaving the inbox
Gemini can summarize an email or thread and surface the important points. Use it to catch up quickly, then open the original messages before relying on dates, commitments, or decisions.
Source: Google Workspace, Gemini in Gmail.
Core moves
- Summarize any thread: click "Summarize this email," or ask "What's this email about?" in the side panel — works mid-thread too ("What did I miss?").
- Help Me Write ✦: click the pencil in Compose, type a rough instruction ("Decline meeting, suggest Thursday"), then use Formalize / Shorten / Elaborate to iterate. Free for all Gmail users, not just paid tiers.
- Proofread: grammar and tone check on a full draft (AI Pro/Ultra feature).
- Suggested Replies: reads the whole thread, not just the last line — hover to preview the full draft before you tap.
- @ file reference: in Help Me Write or the side panel, type
@+ filename to pull a Drive file's content directly into your draft — the single most underused Gmail trick. - "Catch me up": ask the side panel any time for a live inbox digest, generated on demand.
Vendor / quote comparison
Returning from leave
Decline / reschedule
Pull a file into a draft
Catch-up digest
GOOGLE WORKSPACE
Google Docs: AI cheatsheet
Start, refine, and summarize documents with Gemini
Gemini in Docs can generate a first draft, work from files in Drive, refine existing writing, and summarize a document. Treat generated text as a working draft and confirm facts before publishing.
Source: Google Workspace, Gemini in Docs.
Core moves
- Help Me Write / Create: generate a full doc from a description — Gemini pulls the real referenced files, not placeholder text.
- Highlight-and-refine toolbar: select text → Rephrase / Shorten / Elaborate / Bulletize / Summarize / More Formal / More Casual. Prefer this over regenerating the whole doc — it preserves what already works.
- Help Me Understand: select dense text → right-click → get a plain-English explanation without leaving the doc.
- Multi-file @ references: reference several files with
@in one prompt.
Full doc from a description
Multi-file summary + draft
Power moves most people skip
- Match writing style: Side Panel → "Match writing style throughout" scans a multi-author doc and unifies tone.
- Auto-updating summary: type
@AI summaryinline — it regenerates as the doc changes; useful at the top of any long-lived report. - Audio version of a doc: generate a spoken read-out with a selectable persona and adjustable speed — good for reviewing a draft hands-free.
- Table-to-narrative: paste a data table, ask "Turn this into 3 paragraphs of executive commentary."
GOOGLE WORKSPACE
Google Sheets: AI cheatsheet
Build structured spreadsheets from a prompt
Gemini in Sheets helps structure spreadsheets, write formulas, summarize information, and surface patterns. AI-generated calculations and interpretations still need validation before they go into a report.
Source: Google Workspace, Gemini in Sheets.
Core moves
Build a full sheet from one prompt
Gemini proposes a plan, you approve it, then it builds the structure, formulas, and chart. Always sanity-check the formulas.
Natural-language cell formula
Consolidate multiple sources (high-demand use case)
- Natural-language formulas: click a cell → side panel → describe the logic in plain English ("look up the employee name from Sheet2 and return their department"). Works for SUMIFS, ARRAYFORMULA, IMPORTRANGE, QUERY.
- Fill with Gemini: start a column with a few examples → click Fill → it recognizes the pattern (categorize, extract, reformat, score) and completes the rest.
Power moves
- Chart from a description: select a range → side panel → describe the chart. Charts are fully editable and auto-regenerate as data changes.
- Explain this formula: click any complex formula cell → plain-English breakdown before you trust it.
- Error auditing: ask Gemini to audit the sheet for inconsistencies, blank cells, or data-type mismatches.
- Cross-sheet analysis: "Compare Sheet1 (Q1) and Sheet2 (Q2) and tell me which 3 items grew the most" — verify results on anything business-critical.
- Pull Gmail data into Sheets: "Create a sheet from all invoice emails I received this month."
GOOGLE WORKSPACE
Google Slides: AI cheatsheet
Create and refine presentation content inside Slides
Gemini in Slides generates slide content and imagery from prompts and Drive context. Use it to speed up a first version, then apply your organization's message, branding, and evidence.
Source: Google Workspace, Gemini in Slides.
The 5-minute deck workflow
- In-Slides generation: side panel → describe the deck → Gemini asks clarifying questions, proposes an outline for approval, then builds a fully native, editable deck.
- Match your brand automatically: attach an existing deck as a style reference: "Apply CURE's brand colors and use our default template."
- Canvas alternative route: gemini.google.com Canvas mode can also build a full deck from a topic or an uploaded doc, then export to Slides in one click.
Deck from a prompt
Power moves
- Refine slide-by-slide: "Make slide 3 simpler: bullet points only," "Add a chart from this data," "Remove the jargon on slide 2."
- Speaker notes on demand: "Write presenter notes for each slide as if I'm briefing a non-technical executive audience."
- Formatting tricks: "Make all slides consistent with slide 1's style," "Simplify each slide to max 3 bullets."
GOOGLE WORKSPACE
Google Meet + Calendar: AI cheatsheet
Meet
- Take Notes for Me: Activities panel → Take Notes for Me. Transcribes live, writes a structured summary, tags action items with owners, saves to Drive and attaches to the Calendar invite.
- Decisions log: notes include a dedicated Decisions section with status tags — Aligned, Needs Further Discussion, Disagreed, Shelved.
- Join-late catch-up: click "Summary so far" or ask the sidebar what was discussed before you joined.
- Works for in-person meetings too: open the Meet app → Take Notes for Me, without an active Meet call.
- Translated live captions: 60+ languages, each participant reads in their own preferred language.
Meeting recap for stakeholders
Calendar
- Describe, don't search: "Schedule a 45-min debrief with Ali and Marcus next week, avoid Monday mornings."
- Full event creation from a sentence: "Lunch with Nick downtown" generates date, time, and location suggestions.
- In-context conflict resolution: Gemini proactively suggests alternate times inside a conflicting event.
GOOGLE WORKSPACE
Google Drive + Chat: AI cheatsheet
Drive
- Ask Gemini in Drive: natural-language Q&A over your Drive content without opening files — "Find the vendor contract from March," "Show me documents Sarah shared about the rebrand."
- AI Overview in search results: search normally, and an AI Overview card summarizes the most relevant info at the top, with citations back to source files.
- Drive Projects: lock a fixed set of files/folders/emails as a persistent, shareable context — a team can have the same grounded conversation over the same file set without re-uploading, and files never leave their original location or permissions.
- Video & document analysis: double-click a video in Drive → Ask Gemini → "Summarize this video" or "List all action items from this recording."
- Never copies your files: Drive's Gemini respects existing permissions and DLP controls by design.
Find and cite the source
Chat
- Summarize a Space: "Summarize the last week of conversation in this channel" or "What decisions were made this week?"
- Ask Gemini across apps: "What did we decide about the vendor in our last meeting + the email thread?" — Chat is a selectable source inside the main Gemini app too.
- AI-drafted replies: click the star icon next to the reply box for longer, context-aware draft options.
GEMINI
Beyond Workspace: chat, research, and reusable assistants
gemini.google.com is where you go when a task outgrows a single app — deep multi-source research, a reusable persona, or turning an idea into a working prototype.
GEMINI
Gemini.com: power features cheatsheet
Deep Research plans before it produces the report
Use this when a question needs evidence from many sources, not just a quick answer. Gemini plans the research, works across sources, and returns a structured, citable report for you to review.
Official Google visual: Deep Research.

Deep Research
- Use it for questions that require evidence from multiple sources, comparison, synthesis, and citations.
- Best for: market research, regulatory research, vendor/technology comparisons, emerging trends, and literature reviews. Use normal chat for quick facts.
- When connected sources are available, you can ask it to combine approved internal context with public research. Always verify sensitive or decision-critical claims.
Strong research prompt — copy this structure
Canvas
- Interactive workspace inside Gemini chat: live-preview HTML/React apps, Python scripts, infographics, quizzes — and full Slides decks exportable in one click.
- Can turn a Deep Research report directly into an app, infographic, or interactive quiz.
- Can turn a hand-drawn wireframe or whiteboard photo into a functional prototype UI — useful for early-stage mockups and presentation support.
- Available on every plan tier — usage limits scale with tier, not feature access.
Gems
- What it is: a saved persona + instructions + optional reference files, reusable forever instead of re-explaining context every time.
- Build approach: copy one of Google's premade Gems (Brainstormer, Career guide, Coding partner, Learning coach, Writing editor) and rewrite the instructions for your recurring task.
- Ground it with knowledge files: upload a style guide, FAQ, or product spec so outputs stay on-brand and on-policy.
- Sharing: Gems can be shared within CURE via link.
For technical and IT teams
Coding, debugging, and troubleshooting were among the most-requested AI skills at CURE. The Coding partner Gem, plus general chat, can explain an error, review a snippet, or draft boilerplate — always with a human confirming the change before it ships.
Debugging with a guardrail
Extensions / Connected Apps
- Invoke with
@AppName: @Gmail, @Drive, @Calendar, @YouTube, @Maps, and community-built extensions. - Must be turned on first: Settings → Extensions/Connected Apps.
GEMINI
Gemini Notebook (formerly NotebookLM)
Organize a project once, work from the same trusted sources
Add the material you want the AI to rely on, then ask questions, create summaries, and build learning outputs from that defined source set. Every answer comes with a citation.
Official Google visual: Gemini Notebook.

- What it's for: grounded research over a fixed, private set of your own sources (PDFs, Docs, audio, YouTube links, Drive files) — every answer cites its source, sharply reducing hallucination risk versus open chat.
- Audio Overviews: generates a 5–30 minute two-host discussion of your sources; Interactive Mode lets you pause and ask a question mid-way.
- Mind Maps: auto-generated visual maps of how concepts across your sources relate.
- Study tools: auto-generated study guides, glossaries, practice questions, flashcards.
CURE use cases
- Onboarding: upload all onboarding docs once — new hires query it directly, with citations.
- Competitive intel: upload industry reports, ask "what are the 3 biggest threats competitors highlight?"
- Briefing guides: generate a briefing guide from a stack of source docs for a client presentation.
CROSS-CUTTING
Decision framework: which tool, when
The single most useful table in this training. Check it before defaulting to plain chat out of habit.
| Situation | Use this |
|---|---|
| One-off quick question, brainstorm, short draft | Plain Gemini chat — no setup cost |
| Same task, same persona/instructions, 2+ times a week | A Gem |
| Grounded understanding of a fixed set of your own documents, citations matter | Gemini Notebook |
| Moving from understanding to a deliverable (app, deck, infographic, quiz) | Canvas |
| Synthesis across many external/web sources, 5–20 minutes is fine | Deep Research |
| Team needs a persistent, shared, permission-safe file set | Drive Project |
| Repeatable multi-step cross-app workflow that should run with little manual triggering | Workspace Studio agent |
| Need a Google-quality answer across CURE's connected systems (Workspace, Jira, Salesforce, etc.) | Gemini Enterprise search |
| Jira ticket, requirement, or change record needs review | Rovo |
CROSS-CUTTING
Prompt mastery for business users
The formula: Context + Task + Constraints + Format
- Context: who you are, what this is for, who reads it.
- Task: the exact action, with a verb — "draft," "summarize," "compare," not "help with."
- Constraints: tone, length, audience, what to avoid — make these objective ("3 sentences or fewer") not vague ("keep it brief"). Vague constraints raise hallucination risk because the model fills the gap with invented specifics.
- Format: bullet list, table, email, word count.
Weak vs. strong
Advanced moves
- Role-prompting: "You are a skeptical CFO reviewing this budget..." changes tone and rigor dramatically — this is literally how a Gem's persona is built.
- Few-shot examples: paste 1–3 example outputs as a target format for Gemini to imitate.
- Fixing a bad output — don't restart: point at the specific wrong part ("the tone in paragraph 3 is too casual") and use "Double-check response" before trusting any number or citation.
- Iterate out loud: "Good — now make it shorter and lead with the recommendation," rather than a fresh, unrelated prompt.
- Ask for options: "Give me 3 versions: one formal, one casual, one very brief" when you're unsure what tone lands best.
- Chain prompts across apps: Deep Research report → Canvas turns it into a deck outline → export to Slides → Gemini in Slides adds theme and notes → Gemini in Gmail drafts the distribution email referencing the deck via @-mention.
GEMINI ENTERPRISE
The agentic platform above Workspace, M365, and Jira
Enterprise search, pre-built agents, a no-code Agent Designer, and the governance layer that makes it safe to run at scale — connected across Workspace, Microsoft 365, Salesforce, ServiceNow, Jira and Confluence.
GEMINI ENTERPRISE
Enterprise search & AI assistant
- Google-quality search over CURE's internal data with Gemini's reasoning layered on top — it retrieves facts and synthesizes answers from across the organization rather than just listing links.
- Pre-built connectors for Google Workspace, Microsoft 365, Salesforce, ServiceNow, Jira, and Confluence let people query emails, contracts, and technical documentation from one interface.
- Ask a question and the assistant summarizes what it finds; ask follow-ups to refine the answer.
- Provide context by uploading files or @-mentioning specific people and documents.
- User-directed saved memories: explicitly tell the AI to save or discard specific facts, overriding inferred preferences.
Cross-system client prep
GEMINI ENTERPRISE
Security, data governance & policy controls
| Control | What it does |
|---|---|
| Centralized dashboard | IT admins view, audit, and manage every agent across the org. |
| Model Armor | Screens for malicious inputs like prompt injection and filters outputs to prevent leaks of sensitive data. |
| Data sovereignty | Define where data is stored and processed to meet regional requirements. |
| Observability | Logs, traces, and metrics give full transparency into agent invocations and data handling. |
| Agent & Tool Registry | A single pane of glass to discover, manage, and audit every agent and tool operating across CURE. |
| Agent Gateway | Central enforcement point for policies governing agent-to-agent and agent-to-tool interactions. |
TRANSLATED INTO DAY-TO-DAY WORK
Example use cases by role
Scenario
An Account Executive is meeting a high-stakes client with little time to prepare.
The pain
Hours spent hunting client info across CRMs, folders, and public databases.
The solution
A unified, permissions-aware view of corporate data and the web via Salesforce, SharePoint, and Drive.
Example prompt
"My customer is Cymbal Insurance, meeting their Head of Procurement Thursday. Find their latest earnings report and give me talking points."
Scenario
A finance lead spots an expense anomaly but can't pinpoint the cause without a custom query.
The pain
Extracting a narrative from raw data is tedious and usually needs an already-overbooked analyst.
The solution
Data Insights lets employees "talk" to raw CSVs, Sheets, or Excel files directly.
Example prompt
Upload a raw expense ledger and ask for the top 5 expense categories quarter-over-quarter, with a visualization.
Scenario
Production monitoring turns up failed jobs and no one has time to triage them all manually.
The pain
Sorting failures by cause is slow, repetitive, and easy to deprioritize until something breaks loudly.
The solution
Enterprise search + Data Insights group failures and surface the pattern in minutes, not hours.
Example prompt
"Review today's failed jobs, group them by probable cause, and recommend what should be investigated first."
JIRA + ROVO
Rovo for Jira: AI embedded in the team's change workflow
Jira is CURE's system of record for requirements, development, testing, approvals, and change management. Rovo makes that record faster to build and more complete — without becoming the control owner.
Find. Understand. Improve. Act — with the human still accountable.
Rovo is built around finding organizational knowledge, learning from context, and acting through Search, Chat, Agents and Studio. For this team, those capabilities matter most when they improve requirements, ticket quality, test readiness, and release communication.

Daily Jira digest
What the team should be able to do after this training
Find work faster
Use natural-language Rovo Search to locate relevant tickets, documentation, decisions, and prior incidents.
Improve requirements
Challenge vague requirements, surface missing acceptance criteria, and identify questions before development begins.
Strengthen change evidence
Check whether implementation, testing, validation, rollback, risk, and approval information is actually present.
Spot automation opportunities
Move repetitive review and summarization toward agents and automation without automating accountability.
Where Rovo is most useful in the Jira lifecycle
| Workflow | What Rovo can assist with | Human responsibility remains |
|---|---|---|
| Requirement intake | Summarize request, structure requirements, expose gaps, draft acceptance criteria | Confirm business intent and approve the requirement |
| Development readiness | Identify dependencies, open questions, impacted components | Technical design and implementation decisions |
| Testing | Draft positive, negative, regression and exception scenarios | QA determines coverage, executes testing, records evidence |
| Change management | Check completeness of scope, risk, implementation, validation, rollback | Change owner and approver make the decision |
| Release communication | Summarize completed tickets into business-readable release notes | Owner validates accuracy before publishing |
| Historical research | Find similar tickets, past defects, prior resolutions | Engineer determines whether the historical case applies |
Product visuals in this section are from Atlassian's public Rovo materials. Product behavior and availability can change — verify against your tenant before relying on a feature live.
JIRA + ROVO
Rovo Search, Chat, Agents and Studio
You don't need to memorize product terminology. What matters is choosing the capability that matches the work: finding something, understanding something, performing a defined task, or automating a repeatable workflow.
Rovo Search
Best when you know roughly what you need and want a list of relevant Jira, Confluence, or connected results to inspect.
Rovo Chat
Best for summarizing, comparing, transforming, and asking follow-up questions conversationally.
Rovo Agents
Specialized AI teammates configured with instructions, knowledge, and actions for a particular kind of work.
Rovo Studio
Atlassian's no-code environment for building agents, automations, and apps — mainly an advanced/IT-track tool.



JIRA + ROVO
When to use Search vs. Chat
| Situation | Use | Example |
|---|---|---|
| You need the actual prior tickets | Search | "Find Jira issues about address verification updated in the last 6 months." |
| You found 8 tickets and need the common pattern | Chat | "Compare these issues and summarize recurring symptoms, root causes, and fixes." |
| You need everything assigned to a project/component | Search | "Show open production-support issues for [component] assigned to [team]." |
| You need a management update | Chat | "Using these tickets, summarize completed work, blockers, decisions needed and next steps." |
Search → understand → create
JIRA + ROVO
Writing better requirements
Rovo's biggest value here isn't writing requirements for the team — it's challenging and improving them before development starts.
Standard requirement-review prompt
A thin ticket becoming a good one
Requirement: Add MFA for agency users.
Have Rovo surface the questions a developer would ask anyway, before the ticket ever reaches a sprint:
JIRA + ROVO
Ticket quality, testing & development readiness
Development-readiness prompt
Test-scenario prompt
JIRA + ROVO
Change-management review
Jira holds CURE's change-management evidence, so Rovo's second major job is reviewing a change record for completeness — as a quality-control layer, not the control itself.
Change-completeness review prompt
JIRA + ROVO
The governance boundary
RESPONSIBLE AI
Privacy, governance & rules of the road
Use approved AI tools, protect sensitive information, verify important outputs, and keep human accountability.
AI doesn't replace how you work — it removes the work that got in the way.
RESPONSIBLE AI
Frequently asked questions
PROMPT LIBRARY
Ready-to-use prompts for real work
Start with a strong prompt, replace the bracketed details, and refine from there. The best prompts give AI context, a clear task, boundaries, and the exact output you want.
The prompt formula to remember
Use this when you are building a prompt from scratch.
TASK — What exactly should AI do?
CONSTRAINTS — What must it use, avoid, verify, or not assume?
FORMAT — What should the final answer look like?
Make every important prompt safer
Add: “Use only the information provided or cited sources. Do not invent missing facts. Mark anything unsupported as ‘Not found’ or ‘Needs verification.’”
For decisions, numbers, compliance, customer-facing content, or production changes, keep human review in the loop.
Deep Research — decision-ready
Compare options without losing the evidence
Catch up on a long thread
Professional reply from rough notes
Turn notes into an executive brief
Improve requirements without inventing them
Analyze a dataset — evidence first
Build the right formula
Create a QA check
Build an executive deck
Generate speaker notes
Decision-focused meeting recap
Find the authoritative document
Turn a vague task into a strong prompt
Cross-system project status
Requirement quality review
Generate traceable test scenarios
Change completeness review
TRAINING FEEDBACK
Help us improve the training
Tell us what worked, what helped most, and what you would like to cover next.