Taking work private
Buzz or Berd settles where a piece of work should start. This article is about the move that comes up later: work already lives in Buzz, and part of it wants to leave the room — a draft to rework, thinking to do without witnesses, material someone wants off shared infrastructure. Sometimes the move is right. More often the answer is to stay, and the third option people reach for — wait, since Block says the two products are growing together — is the one to be most careful with.
What the move actually is
There are two shapes, and one hard fact underneath both.
The narrow, shipped route is the Buzz Handoff skill (Berd v0.6.3 and later): paste a buzz://message link or a channel id into a Berd chat, and the agent reads the thread through the Buzz command-line client, continues with you privately, and can send back at most one reply — after you approve its exact text, channel and target. The mechanics are on Buzz or Berd, and the step-by-step is on the Coach: Hand work between Berd and Buzz.
The broad route is files: take copies of the material, work locally, and bring back one finished artefact.
The hard fact: nothing gets unpublished. Every message already in the channel is signed and kept. Taking work private changes what happens from now on; it does not edit the record.
The reverse trip — what survives when Berd work goes multiplayer, object by object — is on the Berd Coach: What survives the trip to Buzz.
Why move, why not
Reasons that hold up: a draft becomes shared when it is ready, not while it is taking shape. Tooling you build around your own way of working goes with you from job to job and was never the team's. A long solo stretch in a channel pays relay round-trips for collaboration that is not happening. And material can need a smaller circle than the relay offers — though that is a hosting decision to make in the open, not a quiet relocation, and neither tool is automatically the safer one.
Reasons to stay: the moment work leaves the room, the room stops learning — colleagues and their agents keep answering from a picture that is quietly going stale, and nothing signals it. The record thins exactly where someone will later need to point back at a decision. That third test — will anyone point back at this later? — is the one that catches people, because work can be entirely your own and still fail it.
Or just wait?
Outlook — direction, not commitments
In August 2026, Bradley Axen told VentureBeat in emailed answers that Block's focus going forward is Buzz, with plans to bring the best parts of Berd's single-player experience into it. Two weeks later, on Block's own On The Block, Berd was presented as a product with its own roadmap — a durable home for your work, where the harness fades into the background, not the app. Both statements are first-party; treat the tension as unresolved rather than quoting whichever suits the argument. The seams are visibly being built on the Buzz side — stronger ACP handling, inviting owned agents from standalone forums — but no shipped release moves a session between the products in either direction. Last checked: Buzz v0.5.23, Berd v0.6.4, 2026-09-19.
Waiting is only a strategy for decisions that can be deferred. Where this week's draft gets written is not one of them. Plan on the two-products reality, and revisit when a documented import ships — not when a roadmap sentence appears.
Why it matters to you
The asymmetry is the point: moving out is easy and silent, and the costs land later, mostly on other people. The product will not warn anyone that the real work now happens off the record. A private move done well is invisible — which is exactly why it needs a habit around it rather than a feeling.
How to apply it
Decide the material first, then the tool — the rule on Buzz or Berd — and never reason backwards from convenience.
Announce the checkout. One line in the channel: what you are taking away to work on, and when something comes back. The record stays coherent, and the channel knows it is waiting rather than abandoned.
Bring back one artefact, not a diary. The channel needs your conclusion and its reasoning in presentable form, not your working notes.
If the worry is who can read the channel, fix the membership or the hosting, not the tool. Moving one document off a relay whose operator you do not trust leaves every other document behind.
Set the handoff identity deliberately: the reply carries whatever key is configured in the agent's environment, marked 🤖 — under your name unless you set up a distinct agent identity first.
If you ignore this
Positively: drafts leave the room rough and come back better, and nobody had to watch the middle part.
Negatively: a project drains into private laptops one convenience at a time. The channel still looks active, but the decisions are being made elsewhere — and the handover, or the client's question six months on, is when everyone finds out.
Examples
Checking a file out of a shared archive, with a sign-out slip left in the drawer. Anyone who looks knows where the file is and when it returns. The alternative — photocopying pages and keeping them at home — feels the same to you and completely different to the archive.
A specialist's question in the engagement channel deserves a considered answer. You paste the thread link into Berd, work it through privately with a critic agent, and approve one reply. The channel shows the question and a single 🤖-marked answer; the working notes never leave your laptop.
Moving a whole project into Berd "because the apps are merging anyway". Nothing shipped moves a channel into Berd or a session into Buzz, the merge has no date, and the signed record you stop producing has no substitute.
Last checked: Buzz v0.5.23, Berd v0.6.4, 2026-09-19 — compiled from this guide's pages verified 2026-09-10 and 2026-09-11; no separate release check.
Verified against Buzz v0.5.23 · Berd v0.6.4 · Updated 19 Sep 2026