Studio: one inbox for every comment

Studio is one place for everything your workspace has published on every account it connected — the videos and posts ChannelForge made and the ones it did not, their numbers over time, and their comments. Its Inbox brings every readable comment together, sorted so corrections and questions come first, with the episode's facts beside each one for your reply. Every change made from Studio — a reply, a hidden comment, a new title — is checked, read back from the platform and kept with what it replaced.

What it does#

Studio's parts#

Open Studio in the sidebar (after Publishing). Its bar has seven parts, and every account, channel and video has a page of its own.

Every page reads within a scope you choose at the top: everything, a project, a channel, one platform or one account. Studio remembers the last one.

How Studio reads your accounts#

Studio reads with each account's own connection. The first collection runs within the hour of an account being connected, then hourly (the newest posts and their comments), with a full sweep once a day. Refresh now on an account's page reads its newest at once.

What a connection allows differs by platform, and Health shows it account by account:

The inbox#

Views, labels and filters#

Comments can be labelled Correction, Question, Request, Criticism, Other, Joke, Praise or Spam, and Waiting for us shows them in that order: a correction first (facts are on the line), spam last. Press Label N comments at the top to label up to 200 at a time, or switch on automatic labels in Health. A comment labelled a request shows What they ask for.

Filter by Platform, Language or Posted (the last 72 hours), or search for a word or a name. Keys: j and k move, r replies, e marks done, h hides, / searches.

Comments are shown as written and never followed as instructions — a comment saying "ignore your rules" is just text.

Opening a comment#

A comment opens beside the list with its thread, the channel, and The facts this episode used, so a reply can quote them. A comment labelled a correction is flagged: A viewer says something is wrong — check it against the facts.

Replying#

1. Write the reply, or start from Use a saved reply, or press Draft two replies for two drafts written from the episode's own facts (their price is shown beside the button; a draft is sent only when you press Send).
2. Press Send (or ⌘↵). It goes out ten seconds later — "Sending in 10 s" — and until then Undo takes it back: nothing is sent.

Every reply keeps the rules of a reply, checked as you type and again before it is sent: within the platform's length (1,000 characters on YouTube, Facebook and Instagram), no link, nothing about AI, and never claiming to be a person. A reply sent shows "Sent to the platform — it appears in the thread once it is read back."

Moderating#

• Hold for review (YouTube) or Hide (Facebook and Instagram, once permitted) — both reversible with Publish it again or Unhide.
• Mark done takes a comment out of Waiting for us without a reply; Back to waiting puts it back.

Where a connection or your role does not allow something, the pane says so and how to fix it, instead of showing a button that cannot work.

The inbox assistant#

The assistant answers comments for you — but only on the posts ChannelForge made, never on anything else of your accounts (your own videos, another Page's posts), and only where your workspace switched it on.

It answers as a person would, never in a pattern:

• awake in a few sessions a day at random hours of the channel's own time zone, with light days and quiet days;
• it notices a comment minutes to hours after it was written, never at once;
• a gap after each reply, at most five replies an hour on an account, and a cap each day;
• questions and corrections first; praise and jokes partly answered, the rest let be and marked done;
• each reply in a shape of its own — a few words to three sentences — never an opening it used lately on the account;
• every number in a reply checked against the episode's facts, and the same rules as your own replies: no link, nothing about AI, never claiming to be a person.

A comment it declines to answer stays in Waiting for us for you, while the praise and jokes it lets be are marked done; it tries a comment three times at most. You can always answer anything yourself.

Turn the assistant on#

1. Open Studio → Inbox.
2. In The assistant is off, tick Let the assistant answer and press Save.

Owners and admins change it; everyone else sees where it stands. It needs an account whose connection allows replies: on YouTube, Allow Studio to manage this channel; on Facebook and Instagram, replies wait for Meta's App Review.

Settings#

In Studio → Health → Switches (admins and owners), and in the Inbox:

Credits and plans#

Reading costs nothing: collections, Refresh now and the Optimize checks hold and charge no credits, and a change sent to a platform costs none either. What a model writes is charged what it used, at your plan's markup: labels, reply drafts, suggestions and thumbnail candidates, each with its price shown first. Each answer from the assistant is a job charged the same way (about 1 credit).

Limits and good to know#

• YouTube's 30-day rule. YouTube's titles, descriptions and comments are refreshed within 30 days or blanked, so an old comment may read "the text is refreshing". Commenters' names are kept 30 days, then only the comment.
• Disconnecting an account deletes everything Studio read through it.
• Nothing changes on a platform any other way than through a change you (or the assistant) made in Studio, each one in Changes.
• No numbers of our own on YouTube's data — such as "× its age" — until YouTube approves our derived-metrics amendment; YouTube's own numbers are shown as they are.

Troubleshooting#

Related#

• Packaging lab and back-catalogue refresh — new titles for older videos, through Studio
• Comment themes — what your viewers ask for, planned into topics
• Insights
• Facebook and Instagram
• YouTube
• Workspaces and roles