Getting the Most Out of Gemini: Gems, NotebookLM & the Google Advantage

This Gemini setup guide covers what most users miss. Gemini isn’t the most famous AI — but its integration advantage is something the benchmarks completely miss.

Table of Contents

1. Why Gemini Gets Underrated
2. Step 1 — Connect Gemini to Your Google Life
3. Step 2 — Build Your First Gem
4. Step 3 — Set Up NotebookLM for Real Research
5. Step 4 — Configure Gemini Advanced
6. Key Features Deep Dive
7. Real Workflow Examples
8. Power User Tips & Shortcuts
9. 30-Day Gemini Challenge

Why Gemini Gets Underrated

In most AI comparisons, Gemini sits in the middle — not as dominant as ChatGPT in brand recognition, not as praised as Claude for writing quality. But this framing misses the point.

Gemini’s advantage isn’t about being the smartest AI in isolation. It’s about being the most connected AI in context. When your AI can actually read your Gmail, search your Drive, and check your Calendar without you copy-pasting anything, the gap between thinking and doing collapses in a way that matters more than benchmark scores.

If you work inside Google’s ecosystem — and most knowledge workers do, at least partially — you’re likely underusing one of the most practically powerful tools available to you. This Gemini setup guide changes that.

Not sure how Gemini compares to ChatGPT and Claude overall? I break down all three in this side-by-side comparison →

Step 1 — Connect Gemini to Your Google Life

The first and most important step in any Gemini setup guide isn’t about prompts — it’s about permissions. Gemini’s Google Workspace integration is only as useful as the access you give it.

How to connect:
Go to gemini.google.com → Settings → Extensions → Enable Google Workspace

This gives Gemini access to:

• Gmail (search, summarize, draft responses)
• Google Drive (find, read, and summarize documents)
• Google Docs (create and edit documents)
• Google Calendar (check availability, find upcoming events)
• Google Maps (directions, business information)

Start with a test:
Once connected, try: “Summarize the last 5 emails I received from [person’s name].” If Gemini pulls real email content and summarizes it accurately, the integration is working.

Privacy note: Gemini’s access is read-only by default for most functions. It can draft content on your behalf in Docs and Gmail, but it won’t send or delete anything without your explicit confirmation. Start with read access and expand as you become comfortable.

Step 2 — Build Your First Gem

Gems are Gemini’s version of custom AI personas — pre-configured assistants you build once and access repeatedly for specific tasks. Think of them as specialist colleagues you can summon instantly.

How to create one:
Gemini → Explore Gems → Create a Gem

The 3 Elements That Define a Good Gem

Name and description: Be specific. “Writing Assistant” is too broad. “Weekly Report Formatter” or “Meeting Prep Assistant” tells both you and the AI exactly what this Gem is for.

Instructions: This is the Gem’s system prompt. Cover its role, the context it should assume, how it should format responses, what it should never do, and any recurring information it should always have.

Knowledge files: Upload reference documents — your brand voice guide, your company’s product specs, a style guide, research papers. The Gem draws on these files when responding, which means its answers are grounded in your specific context rather than generic training data.

Example Gem: Weekly Report Builder

Instructions: You help me write weekly status reports for my team.
Format: 3 sections — Completed This Week, In Progress, Blockers.
Tone: Professional but clear, not corporate-speak.
Always: End with a one-sentence summary of overall progress.
Never: Add filler phrases or unnecessary context.
My role: [your role]
My team's focus: [what your team works on]

Once built, you open this Gem at the end of every week, give it your bullet points, and it produces a formatted report in your voice.

Step 3 — Set Up NotebookLM for Real Research

NotebookLM is the most powerful feature in the Gemini ecosystem that most people don’t know exists. It changes what research means in practice.

What makes it different from regular AI research:
Most AI tools answer questions from their training data — a static snapshot of the internet. NotebookLM answers questions from your documents. You upload your sources, and it becomes an AI that has genuinely read them.

How to set it up:
Go to notebooklm.google.com → Create new notebook → Upload sources

What to upload:

• Research papers, reports, or books you’re working with
• Meeting notes or project documentation
• Long articles or white papers you need to synthesize
• Your own previous work (for analysis and pattern identification)

What you can do with it:

• Ask specific questions and get answers with citations pointing to exact sections of your source documents
• Generate summaries, FAQs, or briefing documents from multiple sources simultaneously
• Create an Audio Overview — a generated podcast-style conversation about your topic (surprisingly useful for absorbing dense material while commuting)
• Identify contradictions or gaps across multiple sources

Where it’s most valuable:
Before any important meeting, presentation, or decision that involves multiple research inputs. Instead of re-reading six documents, upload them to NotebookLM, ask it to identify the three most important points from each, and note where they agree or disagree. What used to take two hours takes twenty minutes.

Step 4 — Configure Gemini Advanced

Gemini Advanced (available with Google One AI Premium) unlocks the full capabilities of the platform, including Gems, extended context, and deeper Workspace integration.

Worth it if:

• You regularly work with long documents
• You use Google Workspace as your primary work environment
• You want Gems functionality for custom AI personas

Personalization settings:
Gemini → Settings → Gemini Apps Activity → On (this allows Gemini to reference your activity for better responses)

Set your preferred response style in your first conversation with Gemini Advanced. Unlike ChatGPT’s Custom Instructions, Gemini’s preferences are set conversationally:

For all our conversations, please:
- Lead with the most important point, not background context
- Use numbered lists only when sequence matters
- Be direct about uncertainty — say "I'm not sure" rather than guessing
- Keep responses under 300 words unless I ask for detail

Key Features Deep Dive

Deep Research Mode
Gemini can run extended research tasks — spending several minutes searching, synthesizing, and structuring information from across the web. Unlike a quick search, Deep Research produces a structured report with sources. Useful for competitive research, market analysis, or any topic where you need synthesis rather than a single answer.

Google Flow (Creative AI)
Flow is Google’s generative media platform — music, video, and image creation integrated with Gemini’s capabilities. For content creators working in Google’s ecosystem, this brings creative generation into the same environment as your other work. Still maturing, but worth watching.

Gemini in Google Docs and Gmail
If you’re writing a document in Google Docs, Gemini can draft sections, rewrite paragraphs, and summarize content — without leaving the document. In Gmail, it can draft replies based on the email thread context. These embedded features reduce switching costs significantly.

Real-Time Search Integration
Unlike Claude’s knowledge cutoff, Gemini searches the web in real time. For any question where current information matters — news, prices, recent research — this is a meaningful practical advantage.

Real Workflow Examples

Workflow 1: Research Synthesis
I upload all my notes, articles, and documents on a topic to NotebookLM. I ask it to identify the three most recurring themes, find any significant contradictions across sources, and generate a one-page briefing document. Total time: 15 minutes for input I’d otherwise spend hours organizing manually.

Workflow 2: Email Catch-Up
After a day of meetings, I open Gemini and ask: “Summarize any emails from my team that require a response today.” It reads my Gmail, identifies relevant threads, and gives me a prioritized list with suggested next steps. I review, decide, and respond — without opening 30 individual emails.

Workflow 3: Meeting Prep with Drive Integration
Before a client meeting, I ask: “Find the last proposal I sent to [company name] in my Drive and summarize the main commitments we made.” Gemini finds the document, extracts the relevant sections, and gives me a briefing I can review in two minutes.

Workflow 4: Gem-Based Weekly Processing
I have a Gem configured for weekly reviews. Every Friday, I open it and give it bullet points from my week. It formats a structured report in my style, asks three follow-up questions to deepen the reflection, and saves the output to my Drive. The whole process takes 12 minutes.

Power User Tips & Shortcuts

Tip 1: Use “@” to reference specific Google content
In Gemini’s chat, type @ followed by a file name or email subject to reference specific content from your Drive or Gmail without switching apps.

Tip 2: Ask Gemini to compare documents
Upload two versions of a document and ask “What changed between these two versions and what are the implications?” Useful for tracking how strategies, proposals, or plans have evolved.

Tip 3: Use NotebookLM’s Audio Overview for dense material
If you have a 60-page report to absorb, generate a NotebookLM Audio Overview and listen during a commute. It’s not a replacement for careful reading, but it gives you a structural map of the content before you go deeper.

Tip 4: Build Gems for your recurring meeting types
A Gem for one-on-ones, a Gem for client presentations, a Gem for strategy documents. Each one knows the context, format, and tone for that specific meeting type.

Tip 5: Use Gemini for real-time fact-checking
Because Gemini searches the web in real time, it’s useful during live conversations for quickly verifying claims, checking figures, or finding a reference you half-remember. ChatGPT requires web search mode; Gemini does it by default.

30-Day Gemini Challenge

Week 1 — Integration

• Day 1: Enable all Google Workspace extensions
• Day 2: Run your first integration test (email summary, Drive search)
• Day 3–4: Set up NotebookLM with a current research project
• Day 5–7: Use Gemini for one Google-connected task per day

Week 2 — Gems

• Build your first Gem for your most repetitive task
• Use it every day for one week
• Note: what did it get right? What needs adjustment?

Week 3 — Research

• Upload 5+ documents to NotebookLM for an active project
• Use it for three distinct research tasks
• Try the Audio Overview feature

Week 4 — Evaluate

• Where did Google integration actually save time?
• Where was Gemini’s research quality better than alternatives?
• Where did you reach for a different tool instead — and why?

The Honest Comparison

Gemini is the right choice when your work is already inside Google and you want an AI that can access it, rather than requiring you to manually bridge the gap between your files and the AI’s chat window.

Its research capability — especially with real-time search and NotebookLM — is genuinely strong. Its integration advantage is real and daily. Its Gems system is as capable as ChatGPT’s Custom GPTs for most use cases.

Where it’s weaker: sustained reasoning over very long documents, and prose quality for writing-intensive work. If your primary work is long-form writing or document-heavy analysis, Claude’s deep work approach is likely the better fit. If you need breadth, creative tools, and the world’s largest custom AI ecosystem, ChatGPT has the edge →

But if your computer is open to Gmail, Drive, and Docs most of the day — and you want an AI that’s already there when you need it — Gemini’s integration makes it the most practical choice for everyday use.

Still deciding which AI fits your work best? This comparison guide breaks down all 3 side by side →

And if you want prompts that work across Gemini, ChatGPT, and Claude for real work situations, the Work Prompt Pack has 45 tested prompts across four categories. $19.

Which Gemini feature surprised you the most? Drop it in the comments.