It's crazy to me that there's no screenshots on the blog post. And when you click through to download it there's a bunch of aspirational pictures, but only one short clip showing what it actually looks like. Why? Is it too hard to show a screenshot?
It is just an 11.8 MB poor alternative to the browser, it is offering keyboard shortcuts and the ability to read your screen, you still need internet connection and manually authorizing access.
To me as an outsider, it seems that anything Google employees do is for internal promotion; as such, no screenshots are needed, the fact that something new had been delivered is enough.
Because app design can change, if you are showing it in the website, everytime app design changes you need to update website screenshots as well to stay consistent
question would be, is it necessary to introduce another point of failure?
of course it is possible, but is this the right time to engineer that solution and make automated screenshot uploads, or connect it to gemini to upload screenshots from time to time?
I still can't stand the fact that in order to use their mobile app you must allow sharing data for training. No thanks will close the app and it seems like they follow the same thing with their desktop app.
Outside of that, if I want to use Gemini through the browser they link allowing your data to be trained on with persistence of chats. If I don't want my data trained on I must allow them to train on my data.
For those reasons it's pretty much unusable for me.
About time, I've been having resource usage and stuttering issues on browser from time to time (both on Firefox and Chromium). It seems a little bit better on desktop.
I still don't understand why aren't they merging Antigravity - Gemini - Spark under one app though. Antigravity is still looks confusing to me.
Sorry, but nobody really cares. In the objective sense, indeed. There are so many players already in this area, and Google's products in this area do not stand out with anything particularly impressive (at this moment, perhaps if they released in 2023 we'd be having different conversation).
Saying this as s.o. calling their APIs on a daily basis for work, but would never let their models touch my everyday OS.
Interesting. I wonder if this will evolve into something closer to Codex or Claude Code, where it can actually work across files and tasks on your computer, not just act as a desktop chatbot.
They'll launch it as Google Gravity, because the anti- was always a negative that Marketing didn't like using, and alliteration tickles your brain better.
Also, that it tries to be the answer to the "everything app" and suck up resources and positions like a blackhole, as bright shining SVPs eventually implode into their golden parachutes.
The Spark feature's UI isn't as good as the others.
For example, I'm not sure if it should prominently show recent workload, and unless I use MCP to manipulate files on my computer, local file manipulation isn't possible. Personally, I don't think it's going to be that useful.
Not "that" bad, but I get frustrated as well since I've got used to GUI for Claude Desktop. In comparison all of the AI stuff for Google looks a little bit confusing and frustrating imo.
i was looking for this comment. raises question how managements work at big tech. why didn't they release in first hand along with androud and ios apps, only long-standing apps are the one that started before generative ai stepped. whats the vision here?
Well, yes? Did you expect a local model?
Or can gemini generate a script to automatically update the screenshots when updated?
of course it is possible, but is this the right time to engineer that solution and make automated screenshot uploads, or connect it to gemini to upload screenshots from time to time?
Outside of that, if I want to use Gemini through the browser they link allowing your data to be trained on with persistence of chats. If I don't want my data trained on I must allow them to train on my data.
For those reasons it's pretty much unusable for me.
It’s such a powerful model, truly useful for general purpose and marketing tasks… but the UX is basic beyond belief.
I literally find myself using AI Studio for general tasks as it’s a better workflow.
Saying this as s.o. calling their APIs on a daily basis for work, but would never let their models touch my everyday OS.
> In which topics
Graduate-level mathematics.
All three will receive no further updates.
Also, that it tries to be the answer to the "everything app" and suck up resources and positions like a blackhole, as bright shining SVPs eventually implode into their golden parachutes.
You enter prompts and images and get a respond back.
What issues are you experiencing?
For example, I'm not sure if it should prominently show recent workload, and unless I use MCP to manipulate files on my computer, local file manipulation isn't possible. Personally, I don't think it's going to be that useful.
https://gemini.google/overview/agent/spark/
Seems like it's still beta.