Gemini is weird cause it feels like one of the most inconsistent models. one day it's simply insane, next day it's a waste of time. this happens in other models ofc, but not that magnitude
personally I think it's best for images and html. use it all the time for that
The brainstorming observation is the one I keep coming back to - Claude works best when you bring the context, Gemini pulls from a wider net on its own. Different cognitive styles as tools.
I've started thinking of it as the difference between a great editor and a great researcher. One tightens what you give it. The other goes wide. Both useful. Rarely interchangeable.
And then of course you have ChatGPT, Grok, Perplexity... It's a full time job just to figure out which tools are the best at which tasks - and once you figure that out every single tool has had 2 to 3 updates.
Probably a naive question...How does one reconcile with the privacy concerns of connecting 'everything' with an LLM? Does that worry anyone? Like, what if it deleted important emails or documents? (Sorry if this is a noob question)
Hey there, great question! when you connect Gemini to your Google workspace (drive, gmail etc.) you aren't giving it a blank check to your computer, it's more like a super-powered research assistant with "read" access so it can scan your emails, summarize them etc. and "draft" access to write a reply. It doesn't have the autonomous power to do permanent actions like sending messages on your behalf or deleting your files.
A special note for anyone attracted to the ability to share the Gemini account with five other people. If ONE of those persons does something stupid that violates the ToS and Google shuts down your account, there is literally nothing you can do if they decide not to hear your appeal. It is extremely risky to share your account with anyone. I highly advise against it.
Another thing I do (I’m also learning Spanish haha) is I scheduled a task so every evening at 8 PM I get a new conversation topic in Spanish and can just chat with Gemini about it. It's more advanced, let's say, but I think you learn faster this way.
Related to the subscription, if you use VSCode, there's a Gemini extension which enables chatting, working with your data and coding, which is covered by the same subscription, at least if you are on the Google One AI Pro plan. That's pretty neat as it opens up a workflow similar to Claude Cowork or Claude Code, without a need for more subscriptions.
Love that addition! and Google has so many AI products included in the subscription anyway, it's hard to beat the value. The storage alone is a pretty big deal since it works across all Google products and you don't need to pay for it separately. Before I upgraded to Pro, I was paying for storage on its own 😅
Gemini has been my favorite from the beginning. I have been using the tools regularly. But Your guide is very comprehensive. Thank you for the same.
Glad it's useful even for someone like you already deep into Gemini!
Gemini stands out when you use its full stack, tools, research, and multimodal capabilities together
Yup, all inside the gigantic Google ecosystem
The real edge comes from knowing where each tool shines and combining them into a workflow that plays to their strengths
Couldn't agree more!
Google destroyed many agencies with this one upgrade.
But created a level ground for people like us.
Bookmarking this.
Very well structured guide. Gemini multimodal capability are the best! And the Gemini 3 thinking is just too good.
Glad to hear it’s useful!
Gemini is weird cause it feels like one of the most inconsistent models. one day it's simply insane, next day it's a waste of time. this happens in other models ofc, but not that magnitude
personally I think it's best for images and html. use it all the time for that
I can definitely agree with this :)) I never thought about it this way, but now that you mention it, it makes sense
The brainstorming observation is the one I keep coming back to - Claude works best when you bring the context, Gemini pulls from a wider net on its own. Different cognitive styles as tools.
I've started thinking of it as the difference between a great editor and a great researcher. One tightens what you give it. The other goes wide. Both useful. Rarely interchangeable.
And then of course you have ChatGPT, Grok, Perplexity... It's a full time job just to figure out which tools are the best at which tasks - and once you figure that out every single tool has had 2 to 3 updates.
oh yes, definitely! it’s harder to rely on just one tool, but practice does show you where each one shines and where it fits into yoru work
I would add - practice and reading articles like yours.
really appreciate that!
Comprehensive and well-structured, one of the better Gemini guides out there.
Thank you, Aniket!
Probably a naive question...How does one reconcile with the privacy concerns of connecting 'everything' with an LLM? Does that worry anyone? Like, what if it deleted important emails or documents? (Sorry if this is a noob question)
Hey there, great question! when you connect Gemini to your Google workspace (drive, gmail etc.) you aren't giving it a blank check to your computer, it's more like a super-powered research assistant with "read" access so it can scan your emails, summarize them etc. and "draft" access to write a reply. It doesn't have the autonomous power to do permanent actions like sending messages on your behalf or deleting your files.
Awww Daria!! This is Goldwert! I don‘t know if there is an equivalent for Goldwert in English 😅
Anyways, thank you so so much! The double fact check is genius and exactly what I need in my academic research 🤩
Oh I love that, glad it’s useful!
A special note for anyone attracted to the ability to share the Gemini account with five other people. If ONE of those persons does something stupid that violates the ToS and Google shuts down your account, there is literally nothing you can do if they decide not to hear your appeal. It is extremely risky to share your account with anyone. I highly advise against it.
that’s a very fair point to be careful about it
Google arbitrarily changes things and breaks stuff. Annoying inside a business workspace.
I'm not a big fan of Google, but it would be awesome if I could use Gemini to learn Spanish faster 😁
They actually have a product for this: https://labs.google/lll/en
Awesome 🤩 thanks!
Another thing I do (I’m also learning Spanish haha) is I scheduled a task so every evening at 8 PM I get a new conversation topic in Spanish and can just chat with Gemini about it. It's more advanced, let's say, but I think you learn faster this way.
Love it 🤩
Related to the subscription, if you use VSCode, there's a Gemini extension which enables chatting, working with your data and coding, which is covered by the same subscription, at least if you are on the Google One AI Pro plan. That's pretty neat as it opens up a workflow similar to Claude Cowork or Claude Code, without a need for more subscriptions.
Love that addition! and Google has so many AI products included in the subscription anyway, it's hard to beat the value. The storage alone is a pretty big deal since it works across all Google products and you don't need to pay for it separately. Before I upgraded to Pro, I was paying for storage on its own 😅
Do you know if the video analyzing taps out at a certain length?
There might be a limit in the Gemini app, but when using the Gemini API (and pay for spending and usage) there shouldn't be any limits.
NotebookLM is next level for knowledge synthesis. It helped me pass two biblically-difficult certification exams in the last 6 months.