Already on ChatGPT and Claude — still need Gemini?

Two things recently got me using Gemini again.

One was the increasingly prominent Ask Gemini in the top-right corner of Chrome.

The other was a promotional email from Google, from which I learned Gemini is still going deeper into Gmail, Drive, Docs, Sheets, Slides and its other existing products.

But my first question was actually:

I already use ChatGPT and Claude a lot. Do I really need a third AI?

After using it for a while, my answer for now is:

I don’t need a third AI home base, but Gemini really has found a place where I’m willing to keep it.

Not because it’s suddenly much smarter than the other two.

But because it’s closer to what I’m doing.

What I feel most is Chrome

The place I’m most likely to use Gemini frequently now isn’t the Gemini website, but Ask Gemini in the top-right corner of Chrome.

For example, when I’m:

• looking at a JD and want to quickly grab the core requirements;
• reading an article and don’t understand one passage;
• looking at a document and just want to change one sentence;
• browsing an event page and want to quickly sort out the times and things to note.

Before, for these I’d usually:

copy the content → open ChatGPT / Claude → paste → add a bit of background → then start asking.

Now the current page itself is already the context.

Just ask.

Gemini in Chrome can also use several open tabs together as context. For example, open a few jobs, products or hotels at once, then compare them directly.

None of these abilities is astonishing in itself.

What you really feel is that lots of tiny actions disappear:

no copying, no screenshots, no pasting links, and no explaining “what I’m looking at right now.”

Each step saves only a little time.

But once the cost of calling on AI drops low enough, many small questions I used to be too lazy to open ChatGPT for start getting asked on the spot too.

I think this is an easily overlooked change:

how often you use AI doesn’t necessarily come from the model getting stronger; it may just be that it’s closer to where the work happens.

Gmail and Photos: not that others can’t do it, but the connection is smoother

I also turned on Gemini’s Personal Intelligence, connecting Gmail and Google Photos.

The feeling here wasn’t as strong as with Chrome.

Because “AI can read Gmail / Drive” stopped being unique to Gemini long ago. ChatGPT and Claude can connect to these services now too.

So if it’s just:

“help me find that email from before,”

Gemini being able to do it doesn’t particularly wow me.

The difference is still in the friction.

The email is already in Gmail.

The photos are already in Google Photos.

There’s no need for me to find the material first and then hand it to AI.

So Gemini’s advantage for me right now isn’t really:

it can do what others can’t.

It’s more like:

I used to have to carry context over to AI; now that step is starting to be skipped.

But “native AI” isn’t that magical either

I originally wanted to test one more scenario:

open a JD, have Gemini find my CV in Drive, and analyze the fit directly.

Then I immediately noticed a problem:

my Drive first has to contain the correct, latest CV.

If it holds five historical versions, Gemini can find the files quickly, but that doesn’t mean it knows which one is the source of truth.

This small problem actually struck me as well worth noting down.

People often say now:

“AI already has my context.”

But there are actually at least two different things here:

whether AI can access my information;

and

whether that information itself is accurate, complete and trustworthy.

A muddle of versions in Drive, an unmaintained Calendar, out-of-date information in emails — native AI won’t fix any of these automatically.

It just reads them more conveniently.

So one of my new conclusions is:

AI can save me carrying context around, but I still have to maintain that context.

If anything, the more AI can access your data, the more the source of truth matters.

Which situations do I think are worth trying Gemini for?

If you already use Chrome and Google products a lot, these are the three kinds I’d try first now.

1. Small tasks on the current page

This is what I recommend most right now.

Explaining a passage, changing a sentence, pulling out a page’s key points, looking at a JD, tidying up the current page.

The smaller the task, the more obvious the advantage of “not carrying context.”

2. Comparing several web pages

Open several jobs, hotels, products or research sources at once, and have Gemini compare them directly based on those tabs.

You could do this before too; preparing the material was just a hassle.

Now the browser is starting to swallow the preparation.

3. When the answer is already in your own Google data

Such as email, photos, Calendar, Drive.

In these cases, rather than first digging out the material and sending it to another AI, you can ask Gemini first.

Conversely, if something needs a lot of long-term background, repeated discussion and ongoing iteration, I don’t currently see a reason to move house just for Gemini.

So do you need to use ChatGPT, Claude and Gemini all together?

I now think there’s no need to force yourself to use all three for the sake of a “full AI kit.”

The three products overlap heavily in capability.

If the same project means maintaining three sets of background, chat history and files, adding a tool may actually add work.

I now prefer a simple split:

long-term context stays in the main AI.

Projects that iterate over weeks or months shouldn’t be carried around everywhere.

Immediate context goes to the AI closest to where you are.

If you’re in Chrome, use Gemini on the spot.

Important judgments get an independent check from a second model.

At that point, whether the second is ChatGPT, Claude or Gemini matters less.

I think this is more practical than fixed splits like “Claude for writing, Gemini for search, ChatGPT for analysis.”

Because model capabilities keep changing.

But where the context is can usually be judged right then.

So do I still need Gemini?

For me, right now:

I don’t need to add a new AI home base.

ChatGPT and Claude already hold a lot of long-built context and ways of working; there’s no need to move it all.

But I’m happy to keep using Gemini.

Especially Gemini in Chrome.

Because it occupies a position that didn’t exist before:

it isn’t me remembering AI and going to find it; it’s that while I’m doing something, it happens to be right there.

This has also made me start choosing AI tools a different way.

When I hit a task from now on, I may not first ask:

which model is best?

But first:

where is the information I need right now, and which AI is closest to it?

Sometimes carrying context one less time may be worth more than a slightly stronger model.