The only feature chaining me to Arc is the automated PIP. That is when you switch between tabs or spaces PIP automatically activates. It is so useful.
I'd like to see it use something other than chromium.
The Firefox fork Zen is what I use now and it really covers everything I want, which is a nice looking browser with ublock origin, container tabs, and no extra features like AI or Pocket or whatever other annoying addon seems to be crammed into everything else.
Reliable and encrypted sync is important. One frustration I do have with Firefox is sync doesn't sync everything, certain settings don't sync, and about:config tweaks don't either.
Zero telemetry is also important, as is having a home tab that doesn't connect to anything externally by default. Ideally other than checking for updates on addons and sync, there should be 0 connections made until a website is visited.
Overall I just want a solid browser that lets me read web pages.
It's probably because I don't extensively customize my browser, but my only complaint about Firefox sync is that it doesn't sync the favicons of booksmarks. That seems like such a simple, obvious thing.
It is super nice that Mozilla/Firefox sync works across various forks of Firefox. I've used it successfully across Waterfox, Librewolf, and Zen.
The favicon thing always bugs me too, I imagine it would just increase storage usage on their end vs the minimal amount of text they store now.
> Reliable and encrypted sync is important.
Thank you, we will add this to our list!
Curious, what problem would the sync solve for you? Is it to sync between browser instances on different computers or do you want to use sync for cloud storage of history, bookmarks, etc?
Basically so I can use the browser on multiple devices with the same bookmarks, addons, and settings. P2P sync is nice but as long as it's client-side encrypted cloud storage is fine too.
From their readme
> You probably have 70+ tabs open right now. You're constantly fighting your browser instead of it helping you. Simple tasks like "order Tide Pods from my Amazon order history" should just work with AI agents.
What does automation have to do with the number of tabs? Also _should_ it _just_ work? Do we really need to automate consumption behavior now?
> We believe AI agents should automate your work locally and securely, not send your data to some search or ad company.
Ordering tide pods locally? It won’t change how much ad companies learn about you. It’s just someone else clicking for you.
What am I missing here, has anyone asked for this? Is anyone actually finding these AI browsers useful? Personally I want Arc, not Dia.
> What does automation have to do with the number of tabs? Also _should_ it _just_ work? Do we really need to automate consumption behavior now?
Our agent can also help with tab closing, grouping, etc. We want the entire browser experience to be improved by AI.
We will have semantic search over browser history, bookmarks as well soon.
> What am I missing here, has anyone asked for this? Is anyone actually finding these AI browsers useful? Personally I want Arc, not Dia.
I understand where you are coming from. The use cases that agent can independently handle are quite trivial right now. But we strongly believe as LLMs improve you will be able to have agents automate larger tasks. You might be able to show the agent some task you do and it learns to do that without your supervision.
What do you mean by you want Arc? What features? -- vertical tabs, spaces, shortcuts? We would love to understand!
Here's a real world example with Dia: I open https://www.cbc.ca/newsinteractives/features/wildfires-track... and ask Dia
> how were wildfires this year compared to last?
Looking at the page, it's immediately obvious that 2025 was worse than 2024 but much better than 2023. What did Dia come up with:
> Wildfires in Canada this year have been severe, but last year saw even more land burned.
> According to the CBC wildfire tracker, Canada experienced two of its worst wildfire seasons on record in recent years. In 2024, more than five million hectares burned, making it an extremely destructive year. However, 2023 was even worse, with wildfires scorching 16 million hectares—over three times as much land as in 2024.
> Alberta, including the tourist town of Jasper, was hit especially hard in 2024, with a third of the community burning in July. Early signs suggest that 2025 could also be an active wildfire season, particularly in Western Canada.
> So, while 2024 was devastating, 2023 remains the most destructive recent wildfire season in terms of area burned.
It didn't even know what year it was, it's just a gimmick. I get that you want to be part of the gold rush trying to capture market share by slapping AI on something people use every day. It's just not useful enough yet and hallucinations make it even worse.
What I personally like most about Arc is
- the bookmark system, i.e. persistent tabs
- vertical tabs
- spaces, it's nice to have work separated from other stuff
- nice to look at, it's all very pretty in my opinion
- the fact that it was a fresh new look at what a browser should be without adding much of the AI hype
> It didn't even know what year it was, it's just a gimmick.
I understand. But we are still in early stages of product-market-fit here, AI agent integration with browsers should get lot better.
Thank you! We are adding vertical tabs already, we will look into adding persistent tabs.
Regarding UI, we want to put a fresh coat on top of chromium. But building a new UI layer (like Arc) would involve throwing away a lot of chromium code and I think that comes with its own cost, like sacrificing security updates from Google (which I feel The Browser Company also realized and started to invest when building Dia).
One thing that I do think AI would genuinely be useful at, is a smarter “reader mode”.
I don’t want it to summarize the page. I don’t want it to do the reading for me.
But if it use visual analysis to remove ads, that can work even when they are served from the same domain as content.
A reader mode could identify the text of a page and remove all of the headers remove the footer and just give me the main article.
Various add-ons and browser modes do this to an extent, but they don’t always work reliably and this does seem like something we’re using AI could genuinely improve the experience.
I used to use Evernote reader mode before, which did something like this!
Totally see the value. If you get a minute, please add it to our github issue, we will prioritize.
Zen scratches most of my Arc itches, really, so take a look at its feature set. If AI agents must be a part of the product, it should be possible to turn them off entirely, not leaving the slightest trace anywhere in the UI. The only place I personally want AI in my browser might be for searching history and bookmarks, but a lightweight local model would be perfectly fine for that use case; no need to rope in third parties.
Got it! We already support local models.
Do you use chrome extensions? Were you able to import the extensions you use into Zen?
I don’t need any extensions that are specific to Chrome, and in fact Zen is preferable since its Firefox base allows uBlock Origin instead of just Lite. My main extensions are uBO and Stylus.
Got it.
For ad-blocking, we are already working on integrating Adblock-rust, should be available soon! https://github.com/brave/adblock-rust
I want a browser to let me do more things, not fewer. I barely trust myself to press the right buttons on a page, I certainly don't trust an AI to buy shit for me. I don't have a million tabs open because I close them when I'm done, and I don't pretend to be good at multitasking.
AI browser integration offers no value to me. I find GitHub Copilot and similar coding agents to be useful only because they can index the codebase and documentation faster than I can. There is no analogous problem AI can solve for me in the browser unless websites dramatically change how they interact with requests from AI.
I like Zen. I'm sticking with Zen.
> There is no analogous problem AI can solve for me in the browser unless websites dramatically change how they interact with requests from AI.
Curious: Does summarizing, Q&A with web pages a problem for you?
I am not the original poster, but I have a hard time understanding how this is useful to anyone.
I have tried these sort of tools before and never see the point – if it is something that I want to read in depth then I’m going to spend the time to read it.
If it’s something I’m not particularly interested in the details of, I’ll just skim it and read the parts that I /am/ interested in.
I am genuinely curious, can you help me understand why you would ever want to summarize a page?
This feels like skipping over your own life? Typically If I’m going to a webpage, it’s because I want to enjoy reading it.
I actually chat/Q&A with web pages quite often.
Examples:
- If I'm writing any email in gmail, I paste the enter email thread into chatGPT, claude and take its help with rephrasing or fixing grammatical mistakes
- I use chat to summarize HN threads too.
Definitely curious if the above use-cases are problems for you or not.
BTW when I want to do Q&A, I use LLM chat feature in BrowserOS -- which opens chatGPT, gemini, claude in split-view on any web page and I quickly copy paste the context using shortcuts.
Thanks! I’m glad to see that that’s something that works for you, and I think I understand what you’re using it for more even if it’s not something I would do.
Personally - No way. I don’t want to let generative AI anywhere near anything I write in email.
If I’m writing an email to somebody it’s because I have something to say, and I want to express my own feelings on it.
(I suppose theoretically I could see it I guess if I needed to talk to someone who only used another language?)
But when reading/composing email, I’m not just trying to optimize for raw information throughput- I’m trying to experience my life and the lives of the people who took time to write to me.
If my friend wrote me a letter, I want to enjoy every word of it. Even for something more perfunctory, the nuance of the particulars in the way it is expressed carries a lot of information.
Likewise, I suppose with reading sites like HN comments- I want to see the perspective of the people who are writing, that’s the whole point of reading comments.
It helps me to understand them as people, their use cases, and how they interact with an particular article.
If I didn’t care about any of that, if I didn’t want their perspective, I wouldn’t bother reading the comments…
Not OP but this kind of use case just isn't appealing to me personally. If anything, I'm delighted when I land on a website that's an old school wall of text with high information density and would rarely reach for a lossy LLM distillation built-in to the browser.
Not really. I could imagine AI-integrated docs pages being useful just from the huge volume of information available and how overwhelming it can be to learn something new.
I've used plenty of AI tools and models, local and paid, agents and chatbots, and the hallucination problem is a huge roadblock every time. I treat it like an autocomplete and I babysit its output because I have to. It just can't be trusted to give a correct answer.
I think LLMs being good at producing language is a trap. The chatbot interface is the obvious product to build around it, but it just sucks right now.
To be honest, I don't want a browser. Browsers fundamentally suck, because they never fit the bill perfectly. I want a collection of loosely coupled programs that functionally compose into a user agent application.
I'd wish someone with sufficient resources would someday look at the current behemoths, decompose them into independent subsystems (as small as physically feasible and rationally possible), clearly define the interfaces between those, and release it all with an opinionated glue that makes the whole thing "just work" out of the box. So if one wants to tweak something, they're not in for forking the whole unholy mess, but just the relevant piece.
Basically, I want an user agent built out of Lego-like blocks, in spirit of the original UNIX way (I don't mean pipes as a method of communication though, of course). If I don't like some piece, I can pull in an alternative, or splice in a filter/router/adapter in front of the existing implementation to tweak the behavior. This is fundamentally different to current approach to extensions that merely sit aside and are consulted by the monolith. E.g. if I want an alternative cookie storage with own ideas on site isolation and cross-site interactions, I can realistically have it. Or e.g. if I want a reader-mode decluttering solution - it can be a filter spliced in somewhere appropriate - heck, a smart enough filter can replace the whole website with a different frontend. Consider that we had alternative IM clients piggybacking on the official backend APIs back in '00s and that was really good for the users.
Then automation (AI or not) becomes a matter of having pieces small enough.
(Making it tolerably performant is most likely going to be a giant issue bordering on impossible, but - hey - a man can dream, right?)
This is a bold vision!
Framework-like laptop but for browser :)
I was a huge user of Arc until it became obvious it was no longer being supported. For me, it was all about the tab "model". The key things that made Arc work for me:
- Vertical tabs
- Drag & Drop between "sticky" and "ephemeral" sections
- Hiearchically nested folders
- Command palette with fuzzy match to open/recent tabs
- Meeting integration
There were some other nice features ("Smart" folders with GH PRs come to mind), but really 95% of my usage was due to the above. The tab model was so good it actually allowed me to replace "to do list" type tools with just folders of tabs.
For me, I think they had their target market all wrong. They built a bunch of design features nobody used and cared about (custom theming?), when the core pitch was very simple: People LIVE in their web browsers, but everyone just seems to accept that having 100 impossible-to-read chrome tabs open at the same time is somehow acceptable. There are so many browser "power users" out there in white collar roles that don't even realize a better world is possible. These people are not going to be downloading (and likely can't download) a sketchy sounding "firefox fork" named "Zen", but they will absolutely download a professional-looking Chrome fork with security guarantees that allows them to organize their work better.
Thank you, this is a very well put vision/feature list!
We will prioritize this! I miss the above too -- vertical tabs, folders, command palette!
Multiple profiles with different sets of cookies in the same window. Hope you mean that under "workspaces".
Using a consistent design language. Don't use 4 sets of UI fonts and 7 color schemes in the same browser window.
> Multiple profiles with different sets of cookies in the same window.
This would be cool, will look into this.
> Using a consistent design language. Don't use 4 sets of UI fonts and 7 color schemes in the same browser window.
Agree, we want to improve UI/UX much more. If you are designer/frontend engineer (or anyone reading is), please hit me up -- nithin[at]browserOS.com
Great point - Firefox container tabs and the “temporary containers” are essential to me and my workflow.
It makes it so trivial to be logged into multiple AWS accounts or isolate shared cookies by environment.
I would love to have a chromium based browser be able to do this- the best I have found is juggling profiles, but even that doesn’t work nearly as well as FF containers.
Arc has this containers-like feature built in. I work as a contractor on 3-4 projects in parallel, and it helps my workflows a lot.
Arc's fluid transition between tabs, bookmarks, and favorites was excellent UX. It finally helped me tame 100s of open tabs by helping me treat "apps" as distinct from "pages".
Every new browser should be doing something to recognize the fact that browsers are an application platform, and they should be absorbing more desktop OS features in the direction of native-feeling applications.
One of the things I want from a browser is to fade to the background and to expose those applications as natively as possible at the OS level (e.g. I want to be able to put browser-based applications into my dock without having to download special electron wrappers for them.)
> One of the things I want from a browser is to fade to the background and to expose those applications as natively as possible at the OS level (e.g. I want to be able to put browser-based applications into my dock without having to download special electron wrappers for them.)
This would be awesome, thank you for sharing your feature request!
I loved the older version of Android which used to expose very single chrome tab as an app, that was a great UX as well.
Will add this to our list!
Honestly, Zen has done a good job of replicating most of the best parts of Arc; the big missing piece, IMO, is Little Arc, which was pretty workflow-changing. I'd love to see Little Arc / temporary browser windows become more common.
Zen surprised me in a very pleasant way. They made major improvements in the last few months.
Only problem I've had so far is not being able to share a single tab in a Google Meet, being forced to share the whole window instead. Tolerable but annoying.
Agree - plus it’s a FF fork so it allows UBlock.
Ublock lite works just fine
Until you want to block a custom selector.
Honestly, I'd go a different direction entirely. The web has become too frenetic. I want a browser that helps me FOCUS.
If I were to go build something—and maybe I should—it would be something like this:
1. Named workspace tab groups that you can focus on and hide everything except what you're doing right now. Arc did something interesting here. I don't know that it has to be the way Arc did it, but the idea is good.
2. Make the world's best bookmark manager. If AI can help organize it for you, that's neat. Bookmarks are a great idea and they've been neglected for too long.
3. Builtin RSS reader.
And uBlock Origin, of course.
One thing your comment it made me think of that I would find really useful as a future, if you did want to use a LLM as part of a browser-
I’d love for it to be able to read a given webpage and extract the different articles/pages into my RSS feed, even if that page doesn’t support RSS/Atom
Full tree style tabs or even a tree history. Not just one level, infinite depth
I want the vertical tabs and custom webpage scheme (I created a dark scheme for HN). After learning that they discontinued Arc for Dia, I even started my own https://github.com/shi-yan/shrome.
nvm, just saw this https://chromium-review.googlesource.com/c/chromium/src/+/67...
Yep, we are enabling this as well. I think it'll take a while for google chrome to roll it out.
Recent and related:
Atlassian is acquiring The Browser Company - https://news.ycombinator.com/item?id=45126358 - Sept 2025 (484 comments)
What is Arc? Dia is a diagramming tool... what does it have to do with ai?
The main main feature that had me itching to move away from basic chromium shell (I use brave though) was split view, command palette and tab search keyboard driven start to finish. Now chromium has both. So I'm fairly happy. Next thing I want is a command palette and I'll be completely satisfied.
We are adding command palette, that should be available soon!
What commands do you want in command palette, I can think of:
- tab, bookmark, history search
- would you want command palette with AI features? -- like "group my tabs", "close all tabs but this one", "summarize this page"
I would love for it to feel as polished and performant as Jira, please.
If it support plantuml and mermaid that would be awesome
Hmm, wdym?
Native viewing of those diagram formats in the browser, I guess. Seems niche, but I wouldn't turn it down. :)