What Gemini is, the models, Personal Intelligence and connected Google apps.
The full script, word for word — 621 words, about 3 minutes to read. Current as at August 2026. AI tools change quickly; if something looks different when you try it, check the product's own help pages.
Hi — I'm Mel, and this is Gemini: Google's AI assistant. In the next few minutes you'll know what Gemini is and what it can do. Think of this as the map — most of what I touch on gets its own short video if you want to go deeper.
This is home: the chat, at gemini dot google dot com or in the app. If you're signed in to Gmail or any Google service, you're already signed in here. Just type in the box to ask a question, brainstorm, or draft something — and use the plus button to attach your own files.
Down here you choose the brain. Flash is the all-rounder. Flash-Lite is built for the fastest answers, and Pro for advanced maths and code. There's also extended thinking, which gives Gemini longer to work on complex problems. The default is right for almost everything — don't overthink it on day one.
See the gear in the bottom corner? That's your control panel — one click, and everything you can set up lives in one menu: your history, your usage, the look, your subscription. A few things here get their own moment later — Personal Intelligence, Gems, Scheduled actions. You won't need most of it on day one; just know it's all in one place when you do.
Gemini's superpower is that it lives inside Google. Personal Intelligence is where you switch that on: connect your apps — Gmail, Calendar, Photos — give Gemini standing instructions about how you like to work, and let memory build from your past chats. It's off by default: you choose what to connect, and you can turn it off any time.
Gems are custom versions of Gemini you set up yourself. Write the instructions once — your report format, your meal planner — and that Gem holds the setup every time you open it. There are pre-made ones to start from, and you can share a Gem with anyone.
Deep Research turns Gemini into a researcher. Give it a question and it works for minutes, not seconds — reading widely, then coming back with a full, cited report. And you can add your own files as sources too.
Canvas is where Gemini builds things rather than answers things: documents, working code, interactive quizzes — a shared workspace where you shape the result together.
Notebooks give a piece of work a home. Group related chats by topic and ground Gemini on your own sources — or set one up to study: lessons, practice quizzes, and progress tracking for any subject you're learning.
So far, everything happens while you're watching. Spark works when you're not. Hand it a job — pull every booking confirmation out of my inbox into a spreadsheet, or check for new enquiries each morning and draft the replies — and it works in the background, on a schedule or whenever the trigger arrives. It asks first before it spends money or sends anything on your behalf. You decide what it can reach, and availability varies by region and plan.
For creating, Gemini makes and edits images — just describe what you want. And video is possible too with Gemini Omni: combine text, photos and clips, put yourself in the scene, and shape it all just by asking.
Gemini doesn't only live in the chat. It's in Chrome as a browsing assistant, on the Mac, inside Gmail, Docs and Sheets, and in Gemini Notebook.
That's the map: chat, models, settings, connected apps, Gems, Deep Research, Canvas, notebooks, Spark, creation — and Gemini right across Google. It's free to start, so open a chat and give it something real from your day. Good luck — and see you in the next one.