I built a creator studio because I was tired of guessing at my own channel. Then I accidentally built the feature I now use most: my studio works inside Claude.

Not "AI-powered" in the way every tool claims now. I mean the literal thing: I open Claude — the same chat I use to think through scripts — and ask, "which comments am I behind on?" It answers from my actual channel. Then I say "reply to the second one — friendly, mention the follow-up video" and it drafts the reply and sends it. From the chat. (There's a real caveat on the write side, and I get to it below rather than burying it.)

The problem this actually solves

If you run a channel solo, you know the swivel chair: analytics in one tab, comments in another, an AI chat in a third where you paste screenshots and ask what they mean. The AI was supposed to reduce tabs. It became another one.

The copy-paste is the real tax. Every question about your own channel — is my CTR down? which video should I double down on? what's sitting unanswered? — requires you to go fetch the data first, carry it to the AI, and explain the context. Every time. The AI never remembers your channel because it never actually sees it.

What MCP changes (plain-English version)

MCP — Model Context Protocol — is a standard that lets an AI chat connect to outside tools and data. If you've heard of ChatGPT plugins or Claude connectors, it's that idea, but open and increasingly universal.

When a creator studio speaks MCP, the AI doesn't need your screenshots. It can look at your channel directly — with your permission, through your account — and act on it. The conversation stops being "let me describe my problem to you" and becomes "look at my channel and tell me."

What I actually do through it (real examples, my channel)

These are from my own workflow on Travel and Food Guy, running ChannelzIQ's MCP connector in Claude:

  • "What are my top videos in the last 30 days, and did any of my title changes move the numbers?" — It reads the analytics and the change log and answers in plain English. I'm not going to stage a screenshot for this, because the interesting half of the answer is the unglamorous half. The analytics half comes back genuinely useful: over the 90 days to 21 August 2026 it can tell me that Shorts were 50.6% of my views and 5.0% of my watch time, and that a search view is worth 13× a Shorts view in watch time (154 seconds against 12). The title-changes half comes back as a shrug — I've logged around fifty title changes since 8 August and nearly all of them are still running with no before-and-after click-through yet, so the honest answer is "too recent to read." A demo where the AI always has an answer is a demo, not a tool.
  • "List the comments I haven't replied to." — The unanswered pile lands in the chat instead of a browser tab. On my own channel that pile is currently very short: almost everything in my inbox is my own pinned first comment on each video, which — with 99.1% of my views coming from people who aren't subscribed — is its own uncomfortable finding, and not one I'd have noticed as quickly by scrolling Studio.
  • "Fix the title on the Lisbon video to X and push it live." — The edit goes to YouTube. Caveat, stated plainly rather than in a footnote: writing back to YouTube through any third-party connector depends on Google's OAuth verification, and where a channel sits on a Brand Account that write access can be limited or unavailable while verification is still pending. I'm not going to tell you it will work on your channel until you've checked the current status for your own account and setup. Read access and the studio-side actions are the parts I'd rely on today; treat chat-driven publishing to YouTube as something to verify before you plan a workflow around it.
  • "Take this idea and set it up as a project with a script draft." — The idea, project, and script land in the studio, ready when I sit down to work.

The pattern: questions get answered from real data, and small actions get done where I already am. The studio is still there for the deep work — the chat handles the in-between.

What it's not

Honesty section, because this space is drowning in demos that fall apart:

  • It's not autonomous. Nothing happens unless I ask. No AI is renaming my videos while I sleep.
  • It's not magic analysis. The answers are as good as the data and the question — same as the studio itself.
  • It's not a replacement for the studio. Threading a hundred comments or planning a month of content is still better on a real screen.

Why I'm telling you this before launch

ChannelzIQ isn't launched yet. I'm building it in the open, and the waitlist doubles as the roadmap vote — one question, "what are you trying to improve right now?", decides what ships first. The MCP connector exists because I needed it; whether it ships day one for everyone partly depends on what waitlist answers say people actually want.

Stop guessing

If you'd rather ask your channel questions than dig for answers, join the waitlist. It's free, there's one question on the form, and your answer genuinely sets the build order.

Join the waitlist →

Frequently asked questions

What is MCP, in one sentence?

Model Context Protocol — an open standard that lets an AI chat connect to outside tools and data, so it can look at your channel directly instead of waiting for you to paste screenshots into it.

Can it change things on my channel, or only read?

Reading is the reliable part. Writing back to YouTube depends on Google's OAuth verification, and for channels on a Brand Account that write access may be limited while verification is pending — check the current status for your own account before you rely on it. Nothing happens without you asking for it either way.

Is it doing anything on its own?

No. It's not autonomous, there's no schedule, and no AI is renaming videos while I sleep. Every action starts with me typing a request, and the destructive-ish ones get confirmed before they run.

Does it replace the studio?

No, and it isn't meant to. Threading a hundred comments, planning a month of content, or comparing packaging across a dozen videos is still better on a real screen. The chat handles the in-between questions — the ones that used to cost three tabs and a copy-paste.

Are the answers any better than pasting a screenshot into a chat?

They're better mostly because they're grounded and repeatable — it sees the real numbers, so it can say "too recent to read" instead of confidently inventing a trend. The analysis is still only as good as the data and the question, and small samples stay small samples no matter who's reading them.