I thought I had figured out how to use Gemini. I had a few prompts I liked, knew which tasks it was good at, and had gotten used to correcting it whenever the answer wasn’t what I wanted.
Then I made things easier for myself by tweaking some settings.
A few options let me control how Gemini responds and what it remembers about me. After I changed those, Gemini became a much more reliable productivity tool.
I customized Gemini’s responses
I stopped repeating the same instructions
I don’t completely trust Gemini to remember everything correctly.
I’ve noticed that during long conversations or when discussing the same topic across multiple chats, Gemini can confuse details or recall an old situation as if it were still relevant.
So I started using Instructions for Gemini for the details I want it to remember.
One of the first things I added was a list of the devices I own. If I’m asking about an Android feature, an app, or troubleshooting steps for a particular device, Gemini has some context about what I have instead of making me explain it every time.
I can do the same thing with other information that comes up repeatedly.
For example, I could tell Gemini:
These are the phones, tablets, computers, and other devices I currently use: [list].
When suggesting apps, features, or troubleshooting steps, consider these devices first.
I also have an instruction with details about the light conditions on my balcony and a list of plants I have in my indoor garden. And the useful part is that I can update those instructions when the situation changes.
When helping me plan something for my balcony, remember that it gets direct sunlight only in the morning.
The balcony is covered and receives diffused light for the rest of the day.
These are the plants I currently have: [List of plants].
To add your own instructions, visit the Gemini website and click the settings icon in the left sidebar. Then go to Personal Intelligence > Instructions for Gemini > Add.
Similar steps apply for the mobile app.
Gemini became more useful after I connected my other apps
I let Gemini connect the dots
A lot of the information I ask Gemini about exists somewhere on my phone. The problem is that it’s usually scattered across Gmail, Google Drive, Google Photos, Calendar, Keep, YouTube, and Search.
Before connecting those services to Gemini, I often did the tedious task myself: searching for information, copying it into Gemini, and asking it to interpret it.
Connecting my apps changed that.
With Google’s Connected Apps and Personal Intelligence, Gemini can use information from supported Google services to give me more personalized answers.
I can ask it to find something in my Gmail instead of searching through old emails myself, summarize information from my Drive, or find a particular photo or memory in Google Photos.
One example I like is combining information rather than asking about one app at a time.
If I’ve been discussing a trip in Gmail, have related information in Calendar, and have saved places or other relevant information elsewhere in my Google account, Gemini can use those pieces together instead of making me gather them manually.
Look through my recent Gmail, Calendar, and Google Drive information related to [project/trip/event].
Give me a summary of what I need to know, list anything I still need to do, and point out any deadlines or important details I might have missed.
To connect your apps, you can open the Gemini app or website and go to Settings > Personal Intelligence > Connected Apps. Select which available apps you want to connect.
I stopped repeating the same prompt
Gemini started feeling like a customized tool
I ask Gemini to do some things repeatedly, and I realized I was rebuilding the same prompt every time.
I’d explain what I wanted, give it the context it needed, explain what to focus on, and then start working. The next time I needed the same kind of help, I’d do it all over again.
That’s what made Gems click for me.
A Gem is a customized version of Gemini that has its own instructions. I can create one for a specific job, give it the context it needs, and come back to that same Gem whenever I need that kind of help.
For example, I created a Gem called My Meal Planner and gave it the following instructions:
Help me plan meals using ingredients I already have.
Prioritize high-protein and high-fiber meals that take less than 30 minutes on weekdays.
Avoid [ingredients].
When I give you a list of ingredients, suggest several meals I can make from them rather than asking me to go shopping for more ingredients.
After that, I don’t need to repeat all those details. I can ask:
I have eggs, spinach, cheese, tomatoes, and rice. What can I make tonight?
To create a Gem, open the sidebar on the Gemini website, select Gems > New Gem, name it, and write the instructions you want it to follow. You can optionally add files under Knowledge and preview the Gem with a prompt before saving it.
After I’ve created it, I can use the Gem from my phone. In the Gemini mobile app, I open Gems from the left sidebar and select the Gem I want to use. I can type my prompt like I would in a normal Gemini conversation.
The biggest Gemini upgrade wasn’t a new feature
After changing these settings, I can now tell Gemini what I want it to remember, let it use useful context from past chats, connect it to the apps where I already keep information, and set up Gems for things I do regularly.
The biggest difference is that I don’t feel like I’m starting from scratch every time I open Gemini.
