Sidekickブラウザ: what it does and where it breaks down

Anyone searching for what Sidekick the browser is needs one fact before the feature list. The product is no longer distributed. Its official address returns a permanent redirect to a different company's product, and there is no supported way to install it. The design ideas behind it are still worth understanding, because the problem it was built to solve has not gone anywhere. Forty tabs, six logins, and no reliable way to find the one thing that matters right now is the same situation that sent most readers to this page.

The current status, stated plainly

The timeline is short. Perplexity acquired the team behind Sidekick in May 2025. About three months later, on August 3, 2025, Sidekick itself was sunset. As of September 2026, a request to meetsidekick.com returns HTTP 301 pointing at comet.perplexity.ai, which is an AI assistant browser rather than a continuation of the tab workflow.

There is an awkward preamble to that ending. Sidekick floated a message during 2024 saying it would sunset by the end of that year, then kept operating and kept selling Pro subscriptions into January 2025. Buyers who had picked up lifetime deals through AppSumo were openly unsure whether any shutdown was real. The actual stop came more than six months after the first notice.

That gap between announcement and stop is the part worth carrying forward. In this category, the useful question is not only what a product costs but how long it is likely to keep running and what happens to money already paid. Both of those live outside the feature comparison, and both are checkable in a few minutes.

What Sidekick was built to fix

The base was Chromium, the open source project that also sits underneath Chrome. Building on Chromium means page rendering, developer tools and Chrome extensions carry over unchanged, so the work goes into the shell around the page rather than the page itself.

Sidekick replaced the horizontal tab strip with a vertical rail on the left. Gmail, Slack, Notion and anything else with a URL became pinned entries on that rail instead of tabs competing for width. The reasoning is easy to state. Tabs were designed for pages that get opened, read and closed within minutes. Most work software is the opposite. It stays open from morning to night. When permanent residents and passing pages share one strip, the strip runs out of width, labels disappear, and what is left is a row of favicons.

The published feature set included pinned sidebar apps, sessions, workspaces marked as beta on the paid plan, split view, multiple accounts of the same service, a global search, ad and tracker blocking, and an automatic tab suspender that predicted which tabs would not be used and dropped them from memory. It ran on macOS, Windows and Linux, kept Chrome extension support, and offered alternative search engines including Brave and Qwant.

It is worth being precise about what separates this from features already in Chrome. Tab groups collect tabs under a label, and profiles separate cookies and history into distinct windows. Both help. Both also keep the separated things inside the same strip or the same window stack, which means the count still grows and the search still happens. Building a second surface, a rail that does not scroll away, is what changes the motion from finding to reaching.

The trade for that second surface is screen width and a new set of habits. A vertical rail takes horizontal space that a wide monitor can spare and a laptop cannot always. The habits take a few days, which is why several reviews mention a learning curve without calling it a flaw.

The tab suspender deserves a note, because it is not a decoration. Once a browser assumes ten web apps stay open all day, memory becomes a design problem rather than a user problem. Suspension was the balancing entry for the decision to keep everything resident.

Sessions and workspaces were not the same thing

These two words caused more confusion than anything else in the product, partly because one sat on the free plan and the other did not.

A session was a saved group of tabs. Fifteen tabs opened while researching something could be folded under one name and reopened the next day in the same arrangement. Switching sessions did not close the apps pinned to the rail, so audio kept playing and half written messages survived. That behavior is what made the phrase "close nothing, tidy everything" mean something concrete.

A workspace switched the resident apps themselves. A work Gmail and a personal Gmail could sit in separate containers, both signed in, neither aware of the other. So sessions handled research, workspaces handled identity, and the two solved unrelated problems despite sounding alike.

The distinction still matters when reading any product in this category. Counting tabs tells almost nothing. Counting identities that need to stay signed in at the same time tells almost everything. The mechanism that makes that possible is cookie isolation rather than window arrangement, and the Workspaces page sets out how that separation actually works underneath the feature names.

Where it broke down

Published reviews split cleanly. The positive side cited less tab chaos, lower memory use, and faster context switching for people running several projects at once. The critical side cited bugs, sites and extensions that did not behave, CPU spikes, a learning curve, and pricing that placed basics behind the paid tier.

Those complaints sort into three kinds, and only one of them is a defect.

The first is compatibility. Rebuilding the shell on top of Chromium means tracking upstream changes forever, and some sites behaved differently than they did in Chrome. The second is resource use, which is the awkward flip side of keeping everything resident. The third is not a fault at all but a scoping decision: multiple accounts of the same service, which is the exact reason most people go looking for this category, sat on the paid plan.

The first two are worth separating from opinion, because they have a structural cause. A product that rewrites the browser shell inherits an upstream release cadence it does not control. Chromium ships regularly, and every release can move something the shell depends on. A large vendor absorbs that with a team. A small one absorbs it with delays, and delays show up to users as broken sites and stale extension support. The same cause explains why this category has seen several products change hands or narrow their free tiers over the past few years.

Resource use has a similar structural root. Keeping ten web apps resident is not free, whatever the browser does about it. Suspension recovers memory but costs a reload when the app is touched again, which some reviewers experienced as sluggishness rather than as savings. Tuning that balance is genuinely hard, and no product in this category has made the trade invisible.

That third point has a practical consequence. Someone whose whole motivation is a second Gmail could not evaluate the product on the free tier, because the thing being evaluated was not in it. Free tiers in this category are worth reading for what they exclude rather than what they include.

What the acquisition carried over, and what it did not

The team moved. The product did not. Comet, built at Perplexity with that team on board, rolled out free worldwide in October 2025, and it is organized around issuing instructions to an assistant that then works across several sites. That is a different daily motion from pinning ten apps to a rail and switching containers.

For a former user, two things did not survive. The first is the configuration itself. The second is the hours spent building it. Bookmarks and passwords export cleanly from any Chromium based browser, and moving those between browsers is a standard, documented procedure. The container layout is not standard, has no export format, and gets rebuilt by hand in whatever comes next.

That asymmetry is the real switching cost in this category, and it is invisible on pricing pages. A monthly fee stops the day a subscription is cancelled. The design work does not come back.

Where the configuration lives, and why it decides the exit cost

Products in this category answer one question in three different ways, and the answer sets how painful leaving will be.

Some hold the configuration in the vendor's own cloud, tied to an account created during installation. Sync across machines works well, support can see what is happening, and the copy of record sits on someone else's servers. Some sync through storage the user already controls, so the file is reachable even if the product stops. Some keep everything local, which means backups are a manual habit rather than a feature.

None of these is wrong, and the difference only becomes visible at the moment it is too late to change. Sidekick's ending made that concrete for a lot of people at once. A vendor account is fine until the vendor is acquired, and then the useful question becomes whether anything can be exported before access ends.

There is a simple test that takes about two minutes on any candidate. Open the settings, look for an export option, and see what format comes out. Bookmarks as HTML and passwords as CSV are common and portable. A layout of containers and pinned apps, exported as a file that another machine can read, is rare. When that file does not exist, the layout lives only in memory and in whatever notes were kept outside the product.

That is why the recommendation in this category tends to be the same regardless of which product wins. Keep the requirement written down somewhere the product cannot take with it.

Reading the category by the rows that matter

When comparing what is still available, three columns carry more weight than the feature list: whether the product is currently shipping, which platforms it targets, and what happens to a local copy if the vendor changes direction.

Product	Status, September 2026	Platforms
Sidekick	Discontinued, official address redirects elsewhere	macOS, Windows, Linux before shutdown
Wavebox	Shipping	macOS, Windows, Linux
Shift	Shipping	macOS, Windows
Rambox	Shipping	macOS, Windows, Linux
Ferdium	Shipping, Apache 2.0 open source	macOS, Windows, Linux, FreeBSD

The bottom row is the one that answers the question this article started with. When source is published, a vendor decision does not stop an installed copy from running. That is a different kind of guarantee from a good roadmap, and it costs something in polish. The comparison with Ferdium lays out that trade in rows rather than adjectives, and the comparison with Sidekick covers what carried forward from the vertical rail idea into products that are still shipping.

What to change first

Write down every identity that has to stay signed in at the same time, using email addresses rather than service names. That list is the requirement, it survives every product change, and it eliminates most candidates before anything gets installed. Then check each remaining option, including SpaceDeck, for how many of those identities the free tier actually covers.