Frequently Asked Questions

A free suite of tools to help VTubers and content creators plan, write, and grow their content — built by one creator trying to make something that genuinely helps others.

Upload a short, tell it what the short is about (or let it read the visuals and audio), and you get:

Not at all. Everything runs right in your browser — just open a tool and start using it.

Yes, and that's hugely appreciated. The easiest way is to reach out on my socials and let me know, ideally with your channel links so I can find you. The more creators trying the tools, the better they become.

The parts you save to your account — projects, plans, settings — are stored so they're still there when you return. That data is kept encrypted in a private database (hosted on Supabase) with as much security applied to it as reasonably possible. It is not stored in the open or left publicly accessible.

Separately, the Optimizer processes your video entirely in your browser — see "Does my video get uploaded?" below.

No. The Optimizer runs entirely in your browser. The detection models load locally and process the video frames on your own computer — the video file is never uploaded to a server. The first analysis downloads the models (around 380MB total), after which they're cached in your browser so later runs are faster and fully offline.

No. Your data is never sold, rented, or traded — not to advertisers, not to anyone. This is a project built by one creator to help other creators, not a data business. If that ever changed, it would be communicated clearly and in advance, but it is not something this project does.

By default, only you. Your content is tied to your own account through database access rules. The only time anything is shared is when you deliberately choose to share it — for example, inviting an editor or manager to a project, or publishing a media kit to a public link. Nothing is shared unless you opt in.

Yes. You can delete individual projects and content from within the hub at any time. If you would like your entire account and all associated data removed, contact me through my socials and it will be taken care of. Your data belongs to you.

A formal Terms of Service is being written and will be linked here once it's ready. In the meantime, the plain-language summary near the bottom of this page ("How your data is handled") describes exactly what is stored, how it is protected, and what will never be done with it.

No. The hub is not connected to any LLM. It is not ChatGPT, it is not a chatbot, and it does not generate content from a language model.

While it would be technically possible to connect it to such a service, that is deliberately not how it is built. The goal is the opposite: results that are predictable, transparent, and grounded in real creator data rather than produced by a model.

It is machine learning and statistics — which is a different thing from generative AI, even though both get called "AI" in headlines.

The simplest way to picture it: four little parts working together. Hand-written rules (like a smart checklist), plain counting and statistics (what words and tags actually show up a lot), models that recognise — they read your speech, see your frames, and measure how close two ideas are in meaning — and, increasingly, a bit of math that learns what works for your channel from your own numbers. Not one of those four invents content. They recognise, measure, and rank — that's it.

Machine learning here means models and math that recognize and measure: turning speech into text, reading what's in an image, measuring how similar two pieces of text are in meaning, and finding patterns in real performance data. Generative AI means a language model inventing new content — and the hub deliberately contains none of that. Suggestions come from a large database (drawing on VidIQ data and other creator resources), hand-written templates, and patterns measured in what actually performs — never from a model making sentences up.

The practical difference: everything here is predictable, inspectable, and grounded. The same input gives the same answer, every score can show its receipts, and nothing hallucinates.

It's machine learning, running entirely on your device — and being precise about the difference matters here:

• Whisper turns speech into text — high-quality automatic captions. • Vision models (Florence-2, CLIP) describe what's in your video frames and measure image↔text similarity. • The optional smarter search in Ask My Planner uses a small embedding model that measures how close two phrases are in meaning, so "that spooky collab thing" can find your Phasmophobia card.

All of these recognize — none of them generate. They don't make decisions, form opinions, or write content for you. They run in your browser, your files never leave your machine, and nothing you make is used to train anything.

Yes, it's learning to do exactly that — and it stays true to everything above. As you use the hub, it can quietly notice which of your titles, tags, thumbnails and posting times actually did well, and gently lean its suggestions that way. Things like "your titles with a question in them tend to do better" or "your Friday-evening posts get more views."

That's plain math on your own numbers — the same kind of statistics a spreadsheet does, just automatic. It runs in your browser, your data never leaves your device, and clearing your browser resets it. It learns to predict and rank what's worked for you — it still never writes anything for you.

And to be crystal clear: your numbers train only your own copy, on your own machine. They are never sent anywhere, never pooled with other creators, and never used to train any company's model.

It combines several local models, then weighs what you tell it above what the video shows:

Every short gets 5 hashtags, mixed by reach size, so your video has a fighting chance at every stage of the algorithm:

No. The growth rubric is based on publicly known short-form best practices plus outlier research into hashtags that actually perform. The tag pools are refreshed regularly using VidIQ data, but the hub is an independent project and is not affiliated with or endorsed by any of those platforms.

It is entirely free. Every tool and resource on the hub runs on a "support it if you'd like to" basis — there is no paywall, no locked features, and no trial timer. If something here helped you and you wish to support the project, that is genuinely appreciated, but it is never required.

Primarily out of pocket, because it's a project worth building. Optional tips (there's a Ko-fi link) help cover running costs such as the database and data sources, but the hub does not depend on them and will remain free regardless.

Please do. The more creators using it, the more feedback comes back and the better the tools get for everyone.

Hi, I'm Eggie — a VTuber and content creator who got way too into the systems behind growing a channel: the hashtags, the posting cadence, the analytics, the spreadsheets. This hub is everything I built for myself, cleaned up and shared so other creators don't have to start from scratch.

Everything here is free. I build it between streams because I love this stuff and want the VTuber community to have tools that actually fit how we work.

It's tuned for VTubers, streamers, and short-form creators, though the tools work for anyone making content. These are the niches kept most in mind:

If you're wondering whether your niche is covered, here's the full database. Each entry has its own tag pool (small / medium / large mix) and trigger words the detector listens for. Don't see yours? Request it below — new pools get added all the time.

If your community has its own language, hashtags, or characters and it's not in the list above, send it over and I'll fold it in. Same goes for questions, bug reports, or tool suggestions — Twitter DMs are fastest, but anything works.