Question: Just how is OpenAI going to monetize this?
Just wanted to say thanks all for continuing to share feedback for the Atlas team! We've passed all of it along to the folks working on it.
How about a “click to dial” functionality for CRM usage
Hi Atlas team,
I love using the new Atlas browser and I have two suggestions that I think would significantly improve accessibility and usability:
1. Add support for macOS-style “Speak Selected Text.”
In Safari, Chrome, and other macOS applications, I can highlight any specific sentence and choose “Speech → Start Speaking.” Atlas doesn’t seem to support this yet. Being able to highlight a specific piece of text (on a webpage or in the side chat) and have it spoken aloud would be extremely helpful for accessibility and focus.
2. Add a “Speak Response” button inside the side chat.
I haven’t seen an option to have ChatGPT read its replies aloud in the Atlas side-chat. Adding a small “audio playback” button beside each message would make the side panel much more usable, especially for hands-free browsing or users who depend on auditory feedback.
Thanks for considering these improvements — they would make Atlas dramatically more accessible and user-friendly.
– Don Cormier
Hi Team,
A few more ideas:
- open access to AI from extensions (e.g., add some rate limiter like for codex, even shared one would do) so that multiple developers may build on Atlas capabilities
- (overall allowing Link with ChatGPT to 3rd party services for sharing compute credits would be a fun idea to play with, that could create more platform lock-in as now vendors may switch APIs at will not caring about user preferences, while if the user logins with ChatGPT providing access to their own credits, the user is in control and they are less likely to change provider;
- that would also boost some app developers as AI-related cost would be moved to user-side allowing them to lower their prices for OpenAI-enabled integrations)
- make Cmd+Return always work:
- I have Google Search as default, ChatGPT opens on Cmd+Return; but for long queries ChatGPT becomes default, which is fine-ish on its own, but now it only works when I press Return while Cmd+Return doesn’t do anything, which breaks muscle memory
- Cmd+Enter should always run fallback engine, even if it is now also considered primary
- // my personal love note: adding support for spatial web, like spatializing pages, once you launch on Vision Pro — but — I do understand that you won’t do that as this is a VERY niche demand that won’t matter for the next few years, and will likely be closed by upstream vendors
Hope that’s useful.
And thank you for all the work, I love the product, its speed of evolution, and openness of roadmap.
It’s been great seeing a lot of these feature requests being released so quickly 
Another small feature request: the ability to open local PDF files in a separate browser tab.
Use Case:
When you’re in a web meeting and sharing your browser window, it’s super useful to open local files in other tabs so you can easily toggle between them - without needing to share your entire screen.
Please add support for allowing extensions to run in Incognito mode
Right now there’s no way to enable them for private windows, which limits workflows that rely on separate sessions or privacy. A simple chrome “Allow in Incognito” toggle would solve this
Please add back in the Chrome feature ‘Create Shortcut’, where you can create an app icon link to a web page that you can add to the desktop or dock.
Hey folks,
I’d love to see auto-translate built into the Atlas browser. Right now I have to jump between tabs and external tools whenever I land on a non-English page, and it breaks the flow.
It would be awesome if there were a simple “translate this page” toggle or auto-detect function to make browsing smoother.
Since the launch of ChatGPT Atlas, I’ve encountered two significant issues that impact usability and strongly urge the development team to address them:
-
Severe Limitation on Extensions
Atlas currently blocks most browser extensions on chatgpt.com, including widely-used tools such as AdGuard and Tampermonkey. Strangely, these extensions function perfectly on Chrome, yet Atlas, positioned as a browser that enhances ChatGPT integration, restricts them. This seems counterintuitive—supporting extensions to increase user freedom and flexibility should actually be one of Atlas’s key selling points. -
Inability to Copy/Paste in Developer Tools
On macOS 15.1, I noticed that copy/paste functionality is entirely disabled within Atlas’s F12 developer tools, severely hampering routine debugging and reporting workflows. It’s baffling that such a fundamental feature is malfunctioning.
I sincerely hope the Atlas development team can swiftly address and rectify these issues to enhance the browser’s overall competitive edge.
Looking forward to the team’s response. Thank you!
Feature Request: Easier reuse and discovery of previous chat threads across tabs and URLs
Right now, reusing past chat threads in Atlas is a bit cumbersome. Navigating the chat history dropdown is tedious, and thread names aren’t always descriptive enough, which often forces me to open multiple threads just to find the one I need.
Example use cases:
-
I’m using a chat in one tab and want to continue that same conversation in another tab.
-
I’m responding to multiple email threads and want to quickly return to the specific Atlas chat I was using for each email chain.
Potential solutions:
-
Pinned threads: Allow users to pin specific chats so they’re easily accessible across tabs.
-
URL-associated threads: Optionally associate chat threads with URLs, making it easier to resume the right conversation when revisiting a page or workflow.
Both of these would significantly improve usability for people using Atlas across multiple tabs and ongoing tasks.
Hello,
I would like to suggest adding a quick search panel to the Atlas browser, similar to the search box available when using the Agent. This panel could appear as a floating element, making it both convenient and visually clean, and allowing users to perform searches without opening a new tab.
I’ve attached a sample screenshot for reference.
Thank you!
It would be extremely useful if Atlas allowed users on macOS to save images directly into the Photos app — just like other native apps can. Currently, images downloaded from the web or generated via tools like ChatGPT remain in the Downloads folder or require manual import steps into Photos. Providing a “Save to Photos” option (alongside “Download” and “Copy”) would:
-
streamline the workflow for users who manage personal photo libraries in macOS Photos
-
make Atlas feel more integrated and native on macOS
-
be an appealing feature for Apple-centric users considering a full switch to Atlas
Use Cases:
-
Saving screenshots, inspirational images, and AI-generated visuals directly to the Photos library
-
Avoiding extra file organization steps on disk
-
Keeping personal collections in one place with metadata, albums, and iCloud sync
Expected Benefit:
Improved Mac integration and a smoother visual content workflow. Could increase adoption among macOS users who expect deep system compatibility.
Cheers, Karl
Hello, I would also like to request a feature for the ChatGPT Atlas browser: please enable the use of extensions in incognito mode, as currently extensions only work in the main profile. This limitation is quite inconvenient.
@vb Will this be seen by the Atlas team? Not sure how to reach them. Would be great if you could ping them. Thank you!
Hi and welcome to the community!
This topic has been forwarded to the Atlas team:
I can’t tell you how they are prioritizing new feature requests though. I do know that several suggestions made in this topic have already been implemented.
Hope this helps!
Hi team — adding a complementary idea that I didn’t see covered yet in this thread.
Across mainstream desktop browsers, there’s a clear expectation around Save / Export behavior in the File menu:
-
Chrome and Comet expose “Save Page As…”
-
Safari goes further, allowing users to save an entire page (full scroll) as PDF or image, not just the visible viewport
Proposed feature for Atlas:
Add Save / Save As / Export options to the File menu.
Ideal capabilities:
-
Save the current page or chat as PDF, image, or text
-
Support full-page capture (entire scrollable content)
-
Preserve visual + textual context, not just raw text
Why this is especially powerful for Atlas:
One of the most valuable AI workflows is:
capture what’s on screen → attach it to an AI → reason, summarize, or give instructions based on that exact context.
Safari already enables this visually. Bringing a similar capability into Atlas would make it much easier to:
-
Export context intentionally (instead of copy/paste or print-to-PDF)
-
Attach full-page snapshots to chats for deeper reasoning
-
Treat Atlas as an AI-native productivity workspace, not just a browser
Nice-to-have (optional):
-
Standard shortcut support (⌘S / Ctrl+S)
-
An option like “Send page snapshot to chat”
-
Capture scope selection (visible area vs full page)
Thanks for all the great work on Atlas — happy to elaborate further if useful.
Fab 
Can you please have our agent request requests placed into a folder instead of them being dumped in with all our regular prompts for ChatGPT
Hi Atlas team, and everyone else,
I’ve skimmed the thread and while a lot of the features i’m hoping will show up (Tab handling like Arc, with folders, pins, etc.) has already been mentioned, i don’t think anyone has cooked up the reason why this is worth doing.
The Project system on ChatGPTcom could be leveraged here, or something like it.
- Tabs underneath “Add Tab" button in Atlas means those tabs are expendable, so treat them similar to tabs in Arc that are below that button, and automatically close them after 12, 24, 36, whatever hours (Configurable in the settings, should also be able to be turned off )
- Tabs above “Add Tab” button means those tabs are pinned and do not close on their own, they can now be used as actual context - because people usually don’t wanna bookmark things they don’t wanna look at
- Folders among those Tabs above the “Add Tab” button gives even more context, because now you can treat those folders as a project (or merge project and “folders” into one and the asme) - now you can suddenly chain a ton of context together, you have the tabs and what they contain on the pages, you can have files added to the project/folder, conversations, images, files, you name it - now the AI suddenly knows A LOT more.
So yes, while i really really really want the tab management features from Arc and Zen in Atlas, i ended up thinking how it would actually be useful in an AI world, and this seemed like the most obvious way to make it even more intelligent with what everyone uses the internet for.
Not super sure how to handle it with profiles, but something other than Projects could be the solution which has it’s own segmentation inside to show that in MYLIFE profile you have all of this, and in WORKSUCKS you have all this.
Would also allow users to segment work and personal in the same browser, while giving the same AI access to what they do.
Granted work needs to be onboard i guess for that one, but could work for any kind of profile i guess 
For the web sidebar chat, have the option to choose the project folder/custom gpt using the command / or @ would be great.
It would help in the following:
- Chats to automatically go into the selected folder or project, reducing clutter and improving organisation.
- Easier reuse of custom GPTs tied to specific projects, without mixing them into general-purpose chats.
