THE VINYL ACADEMYThe Gemini Course For Gemini
The Gemini Course · seven chapters · one sitting

Use Gemini like you mean it.

Everything a working professional needs to actually use Google Gemini — taught top to bottom, in one session.

Know which plan you're on, which model is answering, where your data goes — and the difference between a feature that's finished and one Google still calls experimental.

One training session, start to finishEvery claim checked against GoogleCheckpoints after every chapter
Start the course →
Hello
How can I help you today?
Explain how to create an avatar video
What kinds of questions can I ask?
Help me with my writing
×Chats are reviewed and used to improve Google AI. Gemini is AI and can make mistakes, including about people. Some content may not be suitable for everyone. You can manage your activity, including info about your location. Learn more about your choices
+ Flash
1

Where Gemini actually lives

By the end of this chapter: you can open Gemini on every surface you own, you know which account you're signed in with and why that changes everything, and you know the one place Google is currently moving the furniture.

Gemini isn't one app — it's the same assistant reaching you in about eight places. The web app, phone apps, a Mac app, Chrome, Google Messages, and inside Google Workspace. Start at the web app and treat everything else as the same brain showing up where you already are.

The main room

The web app

The full experience — every feature lands here first. Google supports it in Chrome, Safari, Firefox, Opera, and Edge. If a feature exists anywhere, it exists here.

In your pocket

Phone, tablet, Mac

Android, iPhone and iPad apps, plus a Mac app. Same account, same chats, same Gems. Google Messages carries it too.

Inside your work

The Workspace side panel

Docs, Sheets, Slides, Drive and Chat can each open Gemini beside the file you're already in — no copy-paste, no switching tabs.

The one that matters

Which account you used

A personal account and a work account get different features, different limits, and — as chapter seven shows — completely different data protection. Check before you type.

Getting the side panel open

Two clicks in each app. The button is always top-right, and the wording is nearly identical everywhere — which is exactly why people miss it.

Open Gemini beside your work
  1. In Docs — open a document, look top right, click Ask Gemini.
  2. In Slides — open a presentation, top right, click Ask Gemini.
  3. In Sheetstop right, click Gemini.
  4. Nothing there? On a work account an admin controls this per app. That's a request to send, not a bug to troubleshoot.

Admins control this service by service. Gemini and the side panel can be switched on or off independently across Gmail, Calendar, Drive, Docs, Sheets, Slides, Forms, Drawings, Vids, Meet and Chat. Missing it in one app doesn't mean you don't have Gemini.

Teach this one with a caveat — Gmail is changing. Google is removing the Gemini side panel from Gmail for Google AI Pro and Ultra subscribers in the US, replacing it with AI built directly into the page: AI Overviews and Help me write. In Google's words, those subscribers "may no longer see Ask Gemini in Gmail." If the panel isn't there, that's the change, not a fault. Google hasn't published a completion date, so expect this to be in motion for a while.
Checkpoint — before you move on
  • You've opened the Gemini web app and sent one real question.
  • You can say which account you're signed in with — personal or work — without guessing.
  • You've opened the side panel in Docs or Sheets at least once.
  • You know that a missing side panel usually means an admin setting, not a broken feature.
Next chapterThe PlansKeep going →
2

The plans, honestly

By the end of this chapter: you can name the four consumer tiers, explain how Google actually meters usage, and avoid the one purchase mistake that costs businesses real money.

This is the chapter most Gemini training gets wrong, because the lineup changed and kept changing. There are four consumer tiers — not three — and several of them come in more than one storage size at more than one price. Prices and tiers move often, so treat every number here as "check it before you quote it."

Tier 1

Free

No AI plan. Google calls this "standard limits." Real work is possible here — you'll just hit the ceiling sooner and drop to the fastest, lightest model.

Tier 2

Google AI Plus

The entry paid tier — 2× the standard limits. Google announced it in the US at $7.99/month, and a larger-storage variant is listed at $9.99.

Tier 3

Google AI Pro

4× standard limits, and the tier where the good stuff lives — Deep Research, the bigger Notebook allowances, and more. Listed at $19.99/month for the 5 TB variant.

Tier 4

Google AI Ultra

Two tiers, $100 and $200 a month. Google describes them as 5× and 20× the Pro limits. This is for people whose whole day runs through it.

The pricing trap, stated plainly: Plus, Pro and Ultra each come in multiple storage sizes at different prices, so "the price of AI Pro" isn't one number. Google's plan page is also built in a way that makes some variants hard to pin down. Never quote a Gemini price from memory or from someone else's course — open Google's own plans page and read the tier you're actually buying. That habit is the lesson.

How Google actually meters you

Here's the thing nobody tells you: Google does not publish how many prompts you get. Not for any tier. They publish multipliers instead — and if you ever see a course quoting "100 prompts a day," that number was invented.

What they publish

Multipliers, not counts

Google's own framing: no plan = "standard limits," Plus = , Pro = , Ultra = 5× or 20× higher than Pro. That's the whole disclosure.

The clock

A rolling five hours

In Google's words, your limit "refreshes every 5 hours until you reach your weekly limit." So there's a short cycle and a long one, and heavy afternoons borrow from the week.

What happens at the ceiling

You don't get cut off

Subscribers who hit the limit keep working on Flash-Lite, the fastest model. You're not locked out — you're downshifted. Knowing that removes the panic.

Why it's vague

Google reserves the right

They say outright they "may at times have to cap the number of prompts and conversations you can have." Capacity is shared and moves. Plan around the multiplier, not a number.

If you're a business, read this part twice

The business side changed in a way that saves money — and contains one expensive trap.

Good news

It's already included

Gemini AI features are now included in Business and Enterprise Workspace subscriptions. No add-on to buy. That covers the side panel in Gmail, Docs, Sheets, Slides, Drive and Chat, Help me write, and Take notes for me in Meet.

The old SKUs are gone

Stop shopping for them

The former paid add-ons — Gemini Business, Gemini Enterprise Legacy, AI Meetings and Messaging, AI Security — were discontinued when the features folded into the base subscriptions.

The expensive trap

"Gemini Enterprise" is not that

Despite the name, Gemini Enterprise is a separate paid Google Cloud product — not part of Workspace. Google states it plainly: it "isn't included with your Google Workspace subscription." Two different things, one confusing name.

Even on Starter

Every business tier gets it

Business Starter, Standard and Plus all include the Gemini app with data protections — that last part requires domain verification, and it's the whole reason chapter seven matters.

Checkpoint — before you move on
  • You can name all four consumer tiers in order.
  • You can explain why "how many prompts do I get?" has no published answer — and what Google gives you instead.
  • You know you don't get cut off at the ceiling, you get moved to Flash-Lite.
  • You can explain the difference between Gemini in Workspace (included) and Gemini Enterprise (bought separately).
Next chapterHow To AskKeep going →
3

How to ask — and which model answers

By the end of this chapter: you can pick the right model on purpose, write a prompt with all four parts, and stop confusing the names in the app with the names in the news.

Two skills, and the second one is worth more than people think. Pick the model on purpose instead of taking the default, and ask in a way that doesn't make it guess. Most bad answers are bad questions wearing a disguise.

The three models in the picker

Here's a trap worth calling out before you open it: the names in the app carry no version numbers. The picker just says Flash-Lite, Flash and Pro. The numbered names you read in the news — 3.1 Pro and so on — are the developer side. Same family, different signage. Teach what's on screen.

Flash-Lite

Fast and light

Google calls it "an efficient workhorse model designed for speed," ideal for summarizing and brainstorming. It's also where you land if you hit your limit.

Flash

The everyday default

"Balances speed and reasoning to solve a large variety of problems, from simple to complex." Start here for most real work.

Pro

The heavy lifter

Google's "most advanced model," built for complex math and coding and hard reasoning. Slower, and it spends your allowance faster — use it when the problem earns it.

The habit to build: match the model to the job, not to your mood. A meeting summary does not need the most advanced model — and running everything on Pro just means you meet the ceiling by mid-afternoon. Flash for the day, Pro for the problem.

The four-part prompt

Same discipline that works everywhere else, because it's not a trick — it's just telling someone enough to do the job. Miss a part and Gemini fills the gap with an assumption.

Goal

What you actually want

Not "help with the report." A finished thing: a one-paragraph summary, a five-row table, a first-draft reply.

Context

Why, and for whom

Who reads this, what they already know, what decision it feeds. This is the part people skip and then wonder why the tone is wrong.

Expectations

Shape, length, tone

Formal or plain. Bullets or prose. Under 200 words. Say it, or you'll get the model's taste instead of yours.

Source

What to use — and only that

This document, that thread, the file I attached. Naming the source is what turns a generic answer into your answer.

Weak: "summarize this quarter." → It has to guess the audience, the length, and which numbers matter. Strong: "One paragraph on Q3 for our ownership group, plain language, from the attached figures only — lead with the single biggest change and say why it moved." Same effort to type. Completely different answer.
Checkpoint — before you move on
  • You've switched models on purpose at least once and noticed the difference.
  • You can explain why the app says "Pro" while the news says "3.1 Pro."
  • You've rewritten one weak prompt into a four-part one and compared the answers.
  • You can say what happens when you hit your limit — and why that's not a lockout.
Next chapterGemsKeep going →
4

Gems — stop re-explaining yourself

By the end of this chapter: you've built a Gem for a job you actually repeat, you know what you can load into it, and you know the one thing it can't do yet.

A Gem is a saved version of Gemini that already knows who you are and how you want the job done. If you're typing the same three sentences of background before every request, that's not a prompt any more — that's a Gem you haven't built yet.

What it is

Instructions that stick

A saved assistant with standing instructions: your role, your standards, the format you always want, the things it must never do. Written once.

Say"You write supplier update emails for me. Always plain language, under 150 words, and never promise a delivery date."
What it can hold

A Knowledge section

Beyond instructions, a Gem carries files — from your device, from Google Drive, and from Gemini Notebook notebooks. That's how it stops being generic and starts knowing your business.

Where it follows you

Built once, everywhere

Google's wording: create a Gem in the web app and "it will appear in the Gemini mobile app and the Gemini side panel in Google Workspace." One build, every surface.

The limit to know

Not in Gemini Live

Straight from Google: "For now, Gems can't be used with Gemini Live." So your Gem doesn't come along in a live voice conversation. Worth knowing before you rely on it.

Build your first Gem — four steps
  1. Open the sidebar in the Gemini web app and click Gems.
  2. Click New Gem and give it a name you'll recognize in six months — "Supplier emails," not "Test 2."
  3. Write the instructions. Who it is, who it's writing for, the format, and the hard rules. Plain English — no special syntax.
  4. Add Knowledge, then Save. Attach the documents it should always work from, and it's live everywhere you use Gemini.

Don't start from blank. Google ships premade Gems — Career guide, Writing editor, Brainstormer — and you can copy one and edit it. Reading a well-written Gem is the fastest way to learn to write your own.

A second thing also called Gems — don't mix them up

Google reused the word, so here's the distinction in one line each.

1
Gems → a saved assistant

Instructions plus knowledge. What we just built. Personal account, age 13 and up.

2
Gems from Google Labs → mini-apps

Custom workflows bundled into small AI apps. A different feature with different rules: 18 and up, and you can only create or edit them on a computer.

3
The practical difference

A Gem answers you. A Labs mini-app runs a workflow — and you can share it publicly, by link, or keep it private.

Checkpoint — before you move on
  • You've built one Gem for a job you genuinely repeat.
  • You've put at least one real document into its Knowledge section.
  • You've opened that Gem on a second device or in a Workspace side panel and seen it there.
  • You can explain the difference between a Gem and a Gem from Google Labs.
Next chapterResearch & CanvasKeep going →
5

Deep Research & Canvas — the two that do real work

By the end of this chapter: you've run one real research report and built one thing in Canvas — and you know where both buttons moved to.

These two are the difference between an assistant and a colleague. Deep Research goes away and comes back with a cited report. Canvas builds the actual document, deck or app beside you. And a heads-up that will save you five minutes of hunting: both now live under the same control.

Where the buttons went. Deep Research and Canvas used to sit on the composer bar. They're now reached through Add Files, below the text box. If you learned Gemini a while ago — or from a course written a while ago — that's why you can't find them.

Deep Research — it writes you a report

Google's description: "in-depth and real-time research on almost any subject." You give it a question, it goes and reads, and it comes back with a cited report. The part people miss is that you get to approve the plan first.

The plan step

Edit before it runs

Gemini writes a research plan and shows it to you. Click Edit plan, fix what it misunderstood, then Start research. Thirty seconds here saves a ten-minute report that answers the wrong question.

Where it looks

The web — or only your files

It defaults to Google Search, but you can add your own Gmail and Drive, uploaded files, and Notebook notebooks. Deselect Google Search and it researches only your sources — which is the trick worth learning.

How long

Five to ten minutes

Google's own figure, "since Gemini analyzes many sources," and longer for complex reports. Start it and go do something else — this isn't a chat you sit and watch.

Getting it out

Straight to Docs

Under Share & export in the Canvas panel: Share Canvas, Export to Docs, or Copy Contents. The report lands in Google Docs formatted and ready to edit.

Run one now — four steps
  1. Click Add Files below the text box, then choose Deep Research.
  2. Ask a real question — one you'd actually assign to someone. Vague in, vague out.
  3. Read the plan and fix it before clicking Start research.
  4. Check the citations. Every claim you plan to repeat, open the source. That's the job.

There are daily limits, and Google won't tell you the number. They confirm limits exist on daily requests and on how many can run at once, and that Pro and Ultra get higher ones — but no figure is published. Also note: Deep Research requires you to be 18 or over and signed in.

Canvas — it builds the thing

Google's line: "collaborate with Gemini in Canvas to create or edit a doc, app, slides, or code." Instead of a wall of chat you get a working document beside you, and you edit it directly.

What it makes

Docs, slides, code, formulas

Write a document, build a slide deck, generate a small working app, or lay out LaTeX formulas. Real artifacts, not descriptions of artifacts.

Say"Turn these notes into a six-slide deck for a supplier meeting — one message per slide."
How you fix it

Edit in the panel

You don't have to re-prompt for every change. Edit the text directly in the Canvas panel, and it saves as you go. Ask Gemini for the big moves, type the small ones yourself.

Where it goes

Out into your real tools

Export to Google Docs, Google Slides (with a PDF option), the clipboard, Google Colab for Python, or PDF for LaTeX. It leaves Gemini as a normal file.

The age gate

Two different bars

Basic Canvas just needs you signed in. But adding AI features to an app, or creating audio overviews, requires 18 or over. Same feature, different floor depending on what you do with it.

Checkpoint — before you move on
  • You've run one Deep Research report on a question that matters to you.
  • You edited the plan before letting it run.
  • You've opened at least two of its citations and confirmed they say what the report claims.
  • You've built one thing in Canvas and exported it to Docs or Slides.
Next chapterNotebookKeep going →
6

Gemini Notebook — answers from your sources only

By the end of this chapter: you've built a notebook on a real body of work, you know your source limits, and you've made an audio overview of it.

If you only take one thing from this course, take this chapter. A notebook is a workspace where Gemini answers only from the documents you put in it — with inline citations pointing at the exact source. It's the difference between "the internet says" and "your files say."

First, the name. This was called NotebookLM until July 16, 2026, when Google renamed it Gemini Notebook. Nearly every article, video and course still says NotebookLM — so if you're searching for help, you'll want to know both names. It's the same product; Google's words: "It remains a standalone product focused on being your premier research tool."
Why it's different

Grounded, with receipts

Google describes "grounded information based on your sources with clear in-line citations." Every claim points back to the document it came from — so you can check it in one click instead of trusting it.

What goes in

More than documents

You can upload PDFs, websites, YouTube videos, audio files, Google Docs and Google Slides. A recorded meeting and a contract can sit in the same notebook and be asked about together.

Where it lives now

Inside Gemini too

Notebooks now appear in the Gemini app and sync both directions — Google: "Your notebook content automatically syncs." Open the sidebar and click New notebook.

The one restriction

No notebooks inside notebooks

Straight from Google: "Notebooks can't use an existing notebook as a source." Each one stands alone — so scope them by project, not by year.

Your actual limits — real numbers, for once

This is the one place in the whole Gemini product where Google publishes hard numbers instead of multipliers. Worth knowing before you plan a big notebook.

1
Standard — 100 notebooks, 50 sources each

Plenty for a single project or a small department.

2
Plus — 200 notebooks, 100 sources each

The step up when notebooks become how you work, not something you tried once.

3
Pro — 500 notebooks, 300 sources each

Where most serious business use lands.

4
Ultra — 500 notebooks, 500 or 600 sources

The ceiling depends on which Ultra storage tier you're on.

5
Business and Enterprise

Gemini Notebook, audio overviews included, comes with Workspace Business and Enterprise subscriptions.

6
How to use the limits

Sources-per-notebook is the number that bites first. One notebook per project beats one giant notebook per year.

Audio and video overviews — the part that surprises people

A notebook can turn your documents into a conversation between two AI hosts discussing them, or into a narrated video. It sounds like a gimmick until you listen to one about your own contract on the drive to work.

Audio Overviews

Four different shows

Generated from the Studio panel in four formats: Deep Dive (the default), The Brief, The Critique, and The Debate. Over 80 languages; length control is English only.

Which to pick

They're not interchangeable

The Brief when you want the gist on a commute. The Critique when you want the holes in your own document found. The Debate when a decision has two defensible sides.

Video Overviews

Three formats, many styles

Cinematic, Explainer, and Short (about 60 seconds). Explainer covers 70+ languages; Cinematic and Short are English only and 18+.

Set expectations

It is not instant

Google warns generation "may take a while" — sometimes more than 30 minutes. Don't start one five minutes before you need it, and don't demo it live without making it first.

Checkpoint — before you move on
  • You've built one notebook on a real project with at least three sources.
  • You've asked it a question and clicked a citation to confirm the answer came from your document.
  • You know your source-per-notebook limit on the plan you're actually on.
  • You've generated one audio overview and listened to enough of it to judge it.
  • You can explain to a colleague that NotebookLM and Gemini Notebook are the same product.
Next chapterAgents & RulesKeep going →
7

Agents, connected apps & the rules

By the end of this chapter: you know what Gemini can do on its own, which of it Google still calls experimental, and — the important one — exactly who can read what you type.

Two subjects that belong together, because they're the same question from two directions: what will Gemini do without you watching, and what happens to your words after you send them. The second one decides whether any of the first one is safe.

What it will do on its own

Be careful here — this is where courses age worst. Almost everything genuinely agentic, Google itself labels experimental. Teach it as promising, not as finished, and check the label before you build a process on it.

The solid one

Scheduled actions

The closest thing to finished. Ask Gemini to do something on a schedule — daily, weekly, monthly — and it runs without you. Managed under Settings & help → Scheduled actions. Hard cap: 10 active at a time.

Say"Every Monday at 7am, summarize my unread mail from the weekend into five bullets."
Experimental

Gemini Spark

Google's personal AI agent for multi-step workflows — and in their own words, "an experimental feature in early development." Personal accounts only, 18+, paid plan required, and not available in the EEA, Switzerland, the UK or Nigeria.

Experimental

Auto browse in Chrome

Gemini completing multi-step tasks on the web for you. Google: "an experimental feature." US only, 18+, paid plan, English device language. You review its plan before clicking Start Task.

Ultra only

Deep Think

Google's "highest level of advanced reasoning," for genuinely hard problems. Reached by selecting the Pro model, then Thinking Level → Deep Think. Ultra subscribers only.

Connected Apps — how Gemini reaches your stuff

Formerly called Extensions. This is what lets Gemini read your mail, your calendar and your Drive — and it's off until you say otherwise.

Turn one on, and understand the trade
  1. Go to Settings & help → Connected Apps (it may appear as Personal Intelligence → Connected Apps).
  2. Toggle on only what you need — Gmail, Calendar, Docs, Drive, Keep, Tasks, Photos, and third parties including WhatsApp, Dropbox, GitHub and Spotify.
  3. Know the permission rule. Google: Gemini "won't access your personal information or content in these services without your permission."
  4. Know the dependency. Connected Apps require Keep Activity to be on — which is the setting we're about to talk about.

Search, Maps, Flights, Hotels and YouTube work differently. They're automatic public-information services rather than toggles you switch — but they still require Keep Activity to be on.

Who can read what you type

The most important section in this course, and the one where the answer depends entirely on which account you're signed in with. Get this backwards in front of your leadership and you won't get a second meeting.

Personal account

Humans do review some chats

Google states it plainly: "A subset of chats are reviewed by human reviewers" to improve their services, and that this use extends to their AI models. Their own advice: "don't enter confidential information that you wouldn't want a reviewer to see."

The myth to kill

Paying does not buy privacy

This surprises people, so say it clearly: Google draws no free-versus-paid distinction in how consumer chats are handled for training or review. Upgrading to Pro or Ultra buys you capacity, not confidentiality.

Work account

Genuinely different

Workspace is a different contract: "Your content is not human reviewed or otherwise used for Generative AI model training outside your domain without permission," and "your interactions with Gemini stay within your organization." That protection even holds between users inside one company.

The one-line rule

Right account, right work

Company information belongs in the work account. If you're signed in with your personal Gmail, treat it like talking in a public place — useful, but not private.

Keep Activity — the switch worth understanding

One setting controls retention, training, and half the features in this course. Here's the honest version, including the part people get wrong.

What it does

Stores your history

When it's on, Gemini keeps your activity in your Google Account. Default retention is 18 months, and you can change it to 3 months, 36 months, or keep it indefinitely.

Turning it off

Off doesn't mean gone

The detail everyone misses: even with Keep Activity off, Google says conversations are still "saved with your account for up to 72 hours" to run the service and process feedback. Off means not used for training — it does not mean nothing is stored.

The cost of off

You lose Connected Apps

Turn it off and Gemini can no longer reach Gmail, Calendar, Docs, Drive, Keep and Tasks. That's the real trade — privacy for capability. Make it deliberately, not by accident.

Where the switch is

Two clicks

Settings & help → Activity, click On, then choose Turn off or Turn off and delete activity. Worth visiting once just to see what's in there.

Google's own warning, worth reading out loud: "Take care when using these features since Gemini can make mistakes, like making purchases or sharing your data with third parties unexpectedly." That's the vendor talking about their own product. Speed from the machine, judgment from you — and on this one, Google agrees.
Final checkpoint — the first-week five
  • Open the web app and set up one Gem for a job you repeat (ch. 4).
  • Build one notebook on a real project and check a citation (ch. 6).
  • Run one Deep Research report — and edit the plan before it runs (ch. 5).
  • Open Settings & help → Activity, look at what's stored, and decide on purpose (ch. 7).
  • Know which account you're typing into, every single time — and verify every fact, figure and name before it leaves your hands.