bmux

The browser is quietly becoming the most important tool in a social media operator’s stack — and bmux is a bet on that shift

If you run social accounts for a living, you already know the dirty secret of the job: most of your “strategy” happens inside a browser with fifteen tabs open. One tab is the TikTok Creative Center, another is your scheduling tool, another is a competitor’s Reels page you’re reverse-engineering, another is a ChatGPT or Claude session drafting captions, and somewhere in there is a half-finished Canva export. The work isn’t glamorous; it’s tab management with a content calendar stapled to it. So when a tool shows up that rethinks the browser itself as a multi-session workspace with agent automation baked in, I pay attention — even when it wasn’t built for creators at all.

That’s the frame for bmux, a free, open-source browser from maker Tõnis Tiganik that just hit Product Hunt in early preview. On the surface it looks like a tmux clone for the web — split panes, named sessions, profile isolation. Underneath, it’s a quiet argument that the browser should be programmable, session-aware, and agent-friendly. For social media managers juggling multiple client logins, that argument is more relevant than the maker probably realizes.

What bmux actually is (and what it isn’t)

Let me be precise, because the Product Hunt copy is written for developers and it undersells the creator use case. According to the maker’s own description, bmux is a browser that lets you:

• Keep docs and apps side by side in split layouts
• Maintain separate project workspaces as named sessions
• Run different logins inside the same desktop window using profiles
• Let “agents” browse via a CLI — navigating, inspecting pages, clicking, typing, and taking screenshots in the background
• Attach to an agent’s session when you want to check its work
• Use floating reference panes, keyboard click hints, and configurable Vim-style navigation
• Save layouts and detach windows while pages keep running

The maker also notes it’s been his daily browser for a few weeks, it’s free and open source, and it’s still an early preview. Extension support is experimental — there’s a Bitwarden installer, plus built-in ad blocking and Dark Reader.

What it is not: a social media scheduler, an analytics dashboard, a content repurposing engine, or anything with a “publish to Instagram” button. There’s no API integration with Buffer, Hootsuite, Later, or Metricool mentioned anywhere in the source. If you came here hoping for another scheduling SaaS, close the tab. This is infrastructure, not a workflow product — and that distinction matters enormously for whether it’s worth your time.

The real problem bmux is solving — and why it maps onto creator work

Here’s the operational reality I keep running into. Last month I scheduled roughly thirty posts across five platforms for a single brand, and the browser overhead alone was absurd. I had:

1. The scheduler’s composer open in one window
2. The client’s Instagram logged in via a separate Chrome profile
3. A second client’s LinkedIn in yet another profile because the first profile’s cookies kept bleeding
4. A research tab pulling hooks from competitors
5. An AI chat window drafting variations
6. A screenshot folder for assets

Chrome’s profile switcher technically handles multi-login, but it’s clunky — you either open a whole new window per profile or you accept that switching accounts means logging out and back in. Firefox’s Multi-Account Containers is closer to what I want, but it’s an extension bolted onto a browser that wasn’t designed around it. bmux’s pitch — profiles mixed inside the same split layout, sessions named and saved — is the first browser-native answer I’ve seen to a problem that agency folks and freelance social managers hit every single day.

Why this matters more for agencies than for solo creators

If you manage one brand, you probably don’t feel this pain. One Instagram login, one LinkedIn, done. But if you’re a freelancer with four retainer clients, or an in-house operator running a portfolio of sub-brands, the login-segregation problem is real and constant. The bmux model — named sessions, isolated profiles, mixable in one split view — is essentially Multi-Account Containers plus tmux plus a CLI. That’s a niche combination, but it’s exactly the niche that social agency work lives in.

My take: the maker built this for developers managing coding agents, but the second-most-obvious audience is anyone who logs into five client Instagram accounts a day and has been quietly doing it wrong with Chrome profiles.

How bmux differs from the incumbents you’re already using

Let me place it against tools creators actually pay for, because the comparison clarifies what bmux is not competing with.

Against schedulers like Buffer, Later, and Metricool: These are publishing and analytics layers. They talk to platform APIs, queue posts, surface engagement metrics, and generate reports. bmux does none of that. If your job is “get posts out the door and measure them,” a scheduler is still your core tool. bmux is the workspace around the scheduler — the place where you draft, research, and review before anything hits a queue.

Against browser extensions like Multi-Account Containers or SessionBox: These solve the multi-login problem inside a browser you already use. bmux asks you to switch browsers entirely. That’s a much bigger ask, and it’s the single biggest adoption hurdle I see. The tradeoff is that bmux gets to design session management natively instead of fighting a host browser’s tab model.

Against AI browsing agents like Comet, Dia, or the agentic modes shipping in Arc and Edge: This is the most interesting comparison. The big browsers are racing to add AI agents that browse on your behalf — summarize this page, book this flight, fill this form. bmux’s angle is different: agents work in the background, in a session you can attach to and inspect, without stealing focus from what you’re doing. For a creator running a research task — “pull the top ten hooks from this competitor’s Reels page” — the ability to let an agent grind in a side session while you keep editing your own content is genuinely useful. Whether the CLI-driven agent model is friendlier than a chat sidebar is an open question, and I’d bet most non-technical creators will find the CLI intimidating at first.

Where the math breaks

Here’s the honest limitation. The agent workflow in bmux is CLI-based. That means typing commands, not clicking buttons. For developers, that’s a feature. For a social media manager whose entire job is clicking things in a visual interface, it’s a wall. The maker explicitly frames the product as “for people who spend their day in the browser or work with coding agents” — and reading between the lines, that’s a developer-first audience, not a creator-first one. Any creator adoption will be a secondary effect, not the design target.

What creators and social teams can borrow from bmux’s design

Even if you never install bmux, the design choices are worth stealing as mental models.

Named sessions as a content workflow primitive

Most creators I know run their work in one chaotic browser window. bmux’s “named sessions” idea — one session for research, one for personal browsing, one for agent work, one per client — is a genuinely better mental model than “I’ll just open another tab.” You can approximate it today with Chrome profiles, Firefox containers, or a tool like Shift or Rambox. The bmux framing just makes the abstraction explicit: a session is a workspace, not a tab.

Background agents as a research assistant

If you’re already using ChatGPT, Claude, or Perplexity for content research, the bmux agent model points at something more ambitious: an agent that actually navigates pages, clicks, and screenshots rather than just summarizing text you paste in. Tools like Browse AI and Zapier’s browser automation already do pieces of this. bmux’s twist is running it inside your daily browser with a session you can inspect afterward — which is a meaningfully different trust model than a black-box cloud scraper.

Keyboard-first navigation as an operator skill

Vim-style navigation and keyboard click hints sound like developer affectations, but the underlying principle is real: the faster you can move between panes and tabs, the more content you can review in an hour. If you’ve ever spent a day doing nothing but QA’ing fifty scheduled posts across five platforms, you know that shaving two seconds off each tab switch is a real productivity gain. This is the same reason power users of Raycast or Keyboard Maestro swear by them.

Why TikTok creators should care more than LinkedIn ones

If your content workflow is mostly text — LinkedIn posts, X threads, newsletter drafts — you can approximate most of bmux’s value with browser profiles and a good note-taking app. The pain is manageable. But if your workflow is video-first — scraping hooks from TikTok, pulling trending sounds, screenshotting competitor Reels, cross-posting to YouTube Shorts and Instagram Reels — you’re doing far more browser-based research and asset collection. That’s where a session-aware browser with background agents actually compounds. TikTok’s Creative Center alone is a research rabbit hole that eats hours; anything that lets you run that research in a side session while you edit is worth a look.

Where my judgment says bmux falls short

I want to be transparent here, because the Product Hunt page is a maker’s pitch and it’s easy to get swept up.

It’s an early preview, and the maker says so. No user counts, no funding, no roadmap dates — all not disclosed. That’s fine for an open-source project, but it means you’re adopting something that could change shape or stall. Don’t build your agency’s entire workflow on it yet.

Extension support is experimental. The source mentions a Bitwarden installer and built-in ad blocking and Dark Reader, but nothing about the broader Chrome Web Store ecosystem. If your workflow depends on extensions — Grammarly, Loom, a social scheduler’s browser extension, CapCut’s web tools — you need to verify those work before committing. Not disclosed in the source, so test it yourself.

The agent CLI is a real barrier. As I said above, this is developer-shaped. A creator who’s never opened a terminal will bounce off it. The maker’s own framing (“people who work with coding agents”) confirms this isn’t aimed at the creator market — it’s aimed at engineers who happen to also browse. Creators adopting it would be early, patient, and technically comfortable.

No platform integrations, no publishing, no analytics. To state the obvious: bmux will never replace your scheduler. It’s an additional tool in the stack, not a consolidation of it. If you’re already drowning in SaaS subscriptions, adding another browser to your workflow is a real cost — in attention, if not dollars, since it’s free.

The “attach to agent session” trust model is unproven at scale. Letting an agent click and type inside a session you can inspect is a nice idea, but I’d want to see how it handles logins, 2FA prompts, and platform rate limits before trusting it with anything that touches a real client account. My take: keep agents away from authenticated social accounts until the project matures.

What I’d watch / test next

If you’re a social media operator curious about bmux, here’s what I’d actually do this week — no hype, just tests:

1. Install it and recreate your current Chrome setup. Log into two client accounts in separate profiles, open your scheduler in one pane and a research tab in another. Time how long it takes you to switch contexts versus your current setup. If it’s not faster, don’t switch.
2. Try one background research task. Point the agent at a public page — a competitor’s TikTok profile, a trending hashtags list, a blog post — and see if the CLI workflow is tolerable for you. If you bounce off the terminal, that’s your answer.
3. Audit your extension dependencies. List every browser extension you use for social work and check whether bmux supports it. The source only confirms Bitwarden, ad blocking, and Dark Reader, so assume nothing else works until proven.
4. Watch the repo, not the launch. It’s open source and early. The signal to watch is commit frequency and whether extension support broadens. If it stalls in three months, you’ve lost nothing.
5. Steal the mental model regardless. Even if you never install bmux, adopt named sessions in whatever browser you use. The workflow insight is free; the tool is optional.

The bigger story here isn’t bmux specifically — it’s that the browser is being rebuilt around agents and multi-session work, and creators are going to feel that shift whether they adopt it deliberately or get dragged into it. The operators who figure out how to run research, drafting, and QA in parallel sessions will out-produce the ones still doing it one tab at a time. bmux is one early, rough, developer-flavored answer to that problem. Whether it becomes your answer depends entirely on how comfortable you are with a terminal — and how much you hate Chrome’s profile switcher.