The most useful things Google Gemini does are not typed into the chat box, they are picked from a menu. AI Central's Google Gemini Best Practices sets out thirteen techniques, and the ones people miss tend to start the same way: open Canvas, click Create, choose Audio Overview, Web page, Quiz or Infographic. Transcription, image work and Deep Research run straight from the chat. Learn where those controls sit and the rest is prompting.
Reviewed August 2026. Google renames menu items often, so treat the labels as the ones AI Central recorded at the time of writing.
Why the menu matters more than the prompt
Most Gemini advice is prompt advice. AI Central's Google Gemini Best Practices is mostly navigation, and navigation turns out to be the more useful half. Of the thirteen techniques it lists, four involve opening Canvas, and four route through a Create menu holding Audio Overview, Web page, Video and Quiz.
That is a different skill from writing a good prompt, and it is the one nobody teaches. If you have only ever used Gemini as a chat window, you have been using a fraction of it, and no amount of prompt craft will surface a quiz builder that sits two clicks away.
Transcription is the fastest return
The first two techniques in AI Central's guide are the ones to try today, because for most people they replace a paid subscription. Video comes first.
Gemini can analyze video files and produce transcripts, captions, or concise summaries.
AI Central's steps are blunt: open Gemini, upload the video, paste the instruction to transcribe the video keeping everything intact. Audio works the same way, upload the file and ask for the text, and here the guide gives the sharper prompt of the two.
Transcribe this audio verbatim with timestamps and speaker labels
That wording is doing real work. Timestamps and speaker labels are what turn a wall of text into something you can search, cite and cut, and they are exactly what a model quietly drops when you ask loosely. AI Central points the technique at interviews, lectures and meetings, which is where the labels earn their keep.
Deep Research and Audio Overview are one workflow, not two
AI Central numbers these separately, eight and four, but the guide's own wording joins them. Deep Research breaks down a complex query, searches the web, and returns a sourced report you can take into Canvas. Audio Overview turns a doc, a slide deck or a deep research report into a podcast-style audio summary.
So the chain runs: enter the research question, click Deep Research, click Start research, open the report in Canvas, click Create, choose Audio Overview. A question becomes a sourced briefing you can listen to on a commute, and you never leave the tab.
That is the pattern worth taking from the whole guide. In Gemini, outputs are inputs. A research report feeds an audio overview. A structured answer feeds an infographic. The thirteen only look like thirteen separate tricks until you notice how many of them hand off to each other.
Gems are the part most people skip
Technique nine has the longest tail. Gems, as AI Central describes them, are reusable custom experts with saved instructions and reference files. You open Gems, select Create new, name it, add the instructions and the files, and save.
The difference between a Gem and a good prompt is persistence. A prompt gets retyped and drifts a little every time. A Gem carries its instructions and its source documents with it, so the context arrives before you type a word. If a task repeats weekly, a brand voice check, a contract read, a Monday report, that is a Gem and not a prompt.
The guide shows the other face of Gems too. Its children's storybook technique starts at Explore Gems and selects storybooks, so there is a gallery of ready-made ones sitting next to the ones you build. AI Central notes that book can be made in a browser or on a phone, and turned into a printable PDF.
The Google apps connection is the structural advantage
Technique ten is the one no rival can copy. Connect Gmail, Drive, Calendar and more in settings, and Gemini works inside them, drafting, summarising, planning and acting. In Gmail you tap Summarize this email. In Drive you open a file and use the Gemini panel. In Calendar you ask Gemini to create or edit an event.
Everything else in AI Central's guide is a capability. This one is a location. An assistant that already lives where your mail and your files live wins on friction rather than on intelligence, and friction is what decides which tool you are still using six months later.
The learning stack
Three techniques are about teaching, and together they make a small self-study kit. Guided Learning, selected from the menu at the top, has Gemini walk you through a problem step by step with visuals and practice, and AI Central's instruction is to ask for examples or quizzes as you follow the flow. Technique thirteen makes the quizzes formal: click Canvas, paste your prompt, click Create, choose Quiz. The guide frames those for self-learning, practice or testing your knowledge.
Use them in that order, and use them adversarially. Reading a clear explanation feels like learning. Failing a quiz you generated yourself tells you whether it was.
The creative tools, briefly
Three of the thirteen cover generation. Veo 3 video sits under Create, then Video, and takes a descriptive prompt.
Create short, realistic clips from a text prompt.
AI Central adds that you can include native audio such as ambience or effects, which is what separates a Veo 3 clip from silent stock footage. Images are simpler: start a new chat, paste an image prompt or upload a photo to edit, generate, then ask for refinements. That last step is not filler. Refinement is where image work is actually done, and the guide is right to build it into the process rather than treat the first generation as the answer.
The infographic technique is the odd one out. AI Central heads it as building infographics in Canva, while the steps themselves run through Canvas: paste your prompt, click Canvas, select Infographic. Follow the steps.
How this was counted
Google Gemini Best Practices is AI Central's own guide rather than a third-party carousel, and it numbers thirteen techniques. The two counts given earlier, four techniques that involve Canvas and four that route through the Create menu, come from reading each technique's step list as written, not from any tally the guide provides.
Two limits are worth stating plainly. The guide supplies a complete, copyable prompt for only two of the thirteen, the video and audio transcriptions, everywhere else it says to paste your prompt and leaves the wording to you. And it does not state which Gemini plan or region any feature requires, so check the Create menu in your own account before you build a workflow around Audio Overview or Veo 3.
What to do with it
Pick your starting point by how much Gemini you already use.
- New to Gemini: do the transcription pair first. Upload one recording, ask for a verbatim transcript with timestamps and speaker labels, and judge the result against whatever you pay for today.
- Daily chat user: give Canvas ten minutes. Open Create and read every option once, because that single menu is where audio overviews, web pages, quizzes and infographics live.
- Already researching in Gemini: chain Deep Research into Audio Overview once, then decide whether it works as a briefing rather than as a party trick.
- Any task you repeat weekly: rebuild it as a Gem with its reference files attached, and stop pasting the same instructions in.
- On Google Workspace: connect Gmail, Drive and Calendar in settings before anything else. The in-app actions are where the time genuinely comes back.
Do I need a paid Gemini plan for all thirteen techniques?
AI Central's guide does not list plan or region requirements for any of them. The only reliable check is your own account: open Gemini, click Create, and see which of Audio Overview, Web page, Video and Quiz actually appear.
What is Canvas in Gemini?
Canvas is the working surface, as opposed to the running chat thread. You open a file or a block of text in it, and the Create button inside it is what offers Audio Overview, Web page and Quiz. Four of the thirteen techniques in AI Central's guide begin there.
Can Gemini transcribe an interview and label who is speaking?
Yes, that is the second technique. Upload the audio file, then ask it to transcribe the audio verbatim with timestamps and speaker labels. The phrasing matters, because a vaguer request tends to come back as a summary.
What is a Gem, and how is it different from a saved prompt?
A Gem is a reusable custom expert with saved instructions and reference files. A saved prompt is text you paste back in. A Gem keeps the instructions and the documents attached to it, so the whole setup returns every time you open it.
Can Gemini turn my documents into a podcast?
That is Audio Overview. Open the file or the text in Canvas, click Create, choose Audio Overview, and you get a podcast-style audio summary. AI Central points it at docs, slide decks and deep research reports.