I ran a faceless channel, and the videos were the easy part. It was the two hours after each one that ate my evenings. Now the transcript goes in and every block comes back ready to paste.
No card required to start. Built for weekly publishing — if you package one video a month, ChatGPT may be enough.
Two chat tabs open, because neither gives you something usable on the first try, and comparing two mediocre answers feels more productive than staring at one.
Then thumbnail text. Then the thumbnail concept itself. And unless you have set it up and keep it current yourself, every session starts by explaining your channel again — the niche, the format, the phrases you refuse to use.
Every block arrives ready to paste, with its own copy button and a line on why it is written that way. Built for faceless operators who publish on a schedule.
Two options, two jobs. One written for Browse — a curiosity or emotional hook. One for Search, with the keyword up front. They map onto YouTube's own title testing, which picks its winner on watch time rather than clicks.
The opening lines treated as their own job. The opening line is written as its own deliverable, because that is what a search result shows.
Real breaks only. Chapters mark actual topical shifts in your transcript, never padded to hit a count. Upload a caption file and they come with timestamps derived from it; paste plain text and you get the titles, honestly labelled as having no timings.
Ordered by search intent. Comma-separated, so one paste into Studio becomes separate tags instead of a single long string. A live counter tracks the character budget as you go and turns green in the range worth aiming for.
Four at most, and optional. A note beside them says when they help and when they are noise, so skipping them is a decision rather than a thing you forgot.
Matched to YouTube's own list. Inferred from what the transcript is actually about, with the upload toggles laid out beside it in the order Studio asks for them.
Direction, not a finished image. Three text options on different angles, a placement map that marks the zones to avoid, and a ready-to-paste prompt for whatever image tool you already use. packaged never generates the finished thumbnail — that stays yours.
The comment you pin under your own video. Written to start the conversation rather than beg for one, and ready before the video is live. A first comment often outconverts a link buried in the description, so it is treated as a deliverable instead of an afterthought.
A real generation, unedited. The transcript is synthetic so no creator's video is exposed; the output is not.
Contents manifest · ready to copy
Why this works: A withholds the overlooked lead; B front-loads the named disappearance and location for search.
One real timestamp, the rest blank. No caption file was supplied, so there is no timing data to derive them from — and packaged will not invent numbers that look measured.
Best fit for a narrated true-crime case examination.
THEY MISSED THIS
Bold white caps, red underline, deep shadow
4-MINUTE LAST CALL
Heavy condensed caps, pale text, red numeral
WHY MILE 47?
Large white caps, yellow question mark, stark contrast
Generated by packaged from a synthetic transcript we wrote, so no creator's video is exposed. The output is real and unedited. The first attempt failed our own validation and was repaired before it would have reached you — that check runs on every generation.
packaged isn't a chat window guessing at what might perform. Every generation runs the same fixed, YouTube-specific process — and you can watch it happen while you wait.
packaged reads the videos you package and keeps a short profile of your channel: your niche, your format, who you are talking to, and the phrases you actually use. Every later package is written with that in front of it, by default — you do not set it up, and you do not keep it current. That is the difference: not that a chat cannot hold your context, but that holding it stays your job.
Open your channel card and read back what it picked up, in your own words, quoted from your own scripts. Tell it what it got wrong, or wipe the whole thing, whenever you want.
What it learns from your videos improves your results only. Never pooled with other customers, never used to train a shared model.
Import your back catalogue in bulk so the profile starts from your real channel rather than from your next upload. It costs no credits, and it comes with any package.
Packages differ by volume, never by features. One pool, one rate, and a package is just how much of it you get.
One block of a new package.
One block back, on a new angle.
A retake costs the same as a first take. That is deliberate, so the price of improving a block never depends on which block it is.
A generation that fails is never charged for.
100 credits free, no card. That is 10 takes, spent however you like. The whole workflow, nothing held back.
If 40 takes per month is enough, don't pay for more.
Start freeWhen 40 takes per month stops being enough, you run a channel.
Start freeStill one channel. Move up when 80 takes per month stops covering you.
Start freeThe channel stopped being a project. It's an asset you own.
Start freeYou don't choose a package here. Start on the free credits, package a real video, and pick your volume from inside packaged once you've seen what it does.
And packaged remembers your channel, not everyone's. What it learns from your videos improves your own future results only — never pooled across customers, never used to train a shared model.
Billing cadence: Monthly.
Including the awkward ones. If an answer here is a hedge, that's a bug — tell us and we'll fix the answer or fix the product.
It isn't a better model — it's a fixed process. Every generation runs the same YouTube-specific steps in the same order, so video 40 gets the same standard as video 1 rather than depending on how you phrased the request that day.
There's no prompt to write, the output arrives already formatted for the exact fields Studio asks for, and each block has its own copy button. And it remembers your channel — the tone you use, the phrases you avoid, your standing links — so you're not re-explaining yourself every session. That last part is the thing a blank chat window structurally cannot do.
It's the question this audience asks most, so here's the honest version. YouTube's inauthentic-content policy is aimed at content that is mass-produced with only superficial variation. Having a consistent channel format is explicitly fine — what isn't fine is publishing the same thing over and over with the details swapped.
packaged writes each package from your specific transcript, so the output describes that video rather than a template with the words changed. That's the reason it insists on a real transcript instead of a topic.
What packaged does not do: check your channel for policy problems, or tell you whether a video is safe to monetise. We built that once and removed it — a clean video getting a scary warning manufactures worry and helps nobody. You're responsible for what you publish, and we'd rather say so than sell you reassurance.
No, and packaged never asks for access to it. It cannot read your channel, publish to it, or change anything on it.
You paste a transcript in and copy the results out. Everything reaches YouTube through you, which also means nothing can be posted on your behalf by mistake.
Just the transcript. Paste it as text, or upload a .txt, .docx, .pdf or .md file. Never a video or audio file — packaged has no use for one.
If you also have a caption file (.srt, .vtt or .sbv), upload it and your chapters come back with timestamps derived from it.
It depends on how long your videos are, so we will not print a single number. The packages count takes, not videos. A take is a fixed amount of transcript, so a short video is one take and a longer one is more — the count is worked out from your transcript and shown before you start.
What we can tell you exactly is the ceiling on one run: packaging an accepted transcript uses up to 3 takes. Retakes and the other paid actions use additional takes on top of that.
What does not move is the price of a take. A take costs the same whether it is part of the first package or a retake afterwards, and retakes draw on the same allowance.
We are not going to tell you how many retakes a creator needs. We do not know yet, and a made-up number would be worth nothing to you.
Anything you do not spend rolls over rather than expiring. Top-up credits are not something you can buy yet.
No. Anything you don't spend rolls over and stacks on top of next month's credits, and we don't expire them, so taking a week off costs you nothing.
If your subscription lapses, your rolled-over credits aren't deleted — they're held. You just can't spend them until you're subscribed again, and then they're waiting on top of your new allocation.
Top-up credits are different again: those never expire at all.
Yes — one real click, no retention maze, no phone call. On billing cycles of six months or longer we email you seven days before the renewal, so a large charge is never a surprise. Monthly renews without an advance email — cancel any time and you keep everything through the period you paid for.
Credits you have not used are not destroyed at the end of the month either. They roll over and stack on top of your next allocation, and we don't expire them.
Regenerate any single block, and steer it — tap a direction chip or type your own note. You never have to redo the whole package to change one line.
Each regeneration is required to come back with a genuinely different angle rather than the same idea reworded — the engine keeps track of which angles it has already given you and won't repeat one.
And if a generation fails outright, you aren't charged for it.
No. What packaged learns from your videos improves your own future results only. It's never pooled with other customers and never used to train a shared model.
You can see what it has picked up about your channel, tell us when it gets something wrong, and wipe it entirely whenever you want.
A take costs 10 credits, and that price is the same whatever you spend it on. One flat rate, one pool, any mix you like.
Packaging a video uses as many takes as the transcript needs — a take is a fixed amount of transcript, and the count is shown before you start. Everything else is one take each: a full package from a transcript, a retake of any single block, restyling the description, baking your links into the description, a retake of a thumbnail text concept, a retake of a next-video idea.
Free, and staying free: the expanded brief behind any next-video idea, reading, copying, pinning and discarding anything already generated. A generation that fails is never charged for.
Every package includes every feature. The packages differ only in how much volume you need.
No, on purpose. You get three text options on different angles, a map of where text works and where it collides with YouTube's own overlays, and a ready-to-paste prompt for whichever image tool you already use.
The finished thumbnail stays yours. Text positions and sizes are starting points, not a template — move anything once you're in your own editor.
You get clean chapter titles with no timestamps, labelled honestly as having none. packaged will not invent timings it cannot derive — a made-up timestamp is worse than an absent one, because you'd have to check every single line.
Upload captions next time and the same chapters come back aligned to real times.
100 credits free. No card required. It starts learning your channel from the first one.
Start free