The problem
Forty posts, eight sites, one person, every week
It is Tuesday. Forty posts have to go out this week across eight sites. The calendar lives in a spreadsheet, the copy lives in a second spreadsheet, three Pages are in Business Suite and the rest are not, and Instagram gets posted by hand at the right minute because there is no other way to schedule it. You rewrite the same caption six times so Meta does not read it as duplicate content, and you swap the image too, for the same reason. The Pinterest bulk file comes back rejected because one column is in the wrong format and it does not say which. On Friday someone asks which post brought traffic, and the honest answer is that nobody knows.
None of that is a content problem. You already have the content — it is sitting on the sites. It is a queue problem, and it gets worse with every site you add.
What it does
One queue, three networks, every site you own
Vaustore takes a site, a batch of posts from it, and a set of destinations, and turns them into scheduled rows — real scheduled posts with text and images attached, not reminders to go and do it later. The same job can fan out across every Page and account you run without you opening another tab.
- Facebook Pages, Instagram accounts and Pinterest, in one queue and one calendar.
- AI writes the post text — not from one prompt, but from twelve styles, each a bank of angles.
- AI generates the image, with the right aspect ratio per platform and that page's own logo on it.
- Every caption, generated or written by you, is measured against a validator before it can publish.
- Every outbound link is rewritten as a short /go/ link, so clicks are counted with bot traffic filtered out.
- Every post lands in a ledger that is reconciled against what Meta says actually went live.
- Accounts contain projects; each project keeps its own sites, pages, credentials, styles and budget.
Building a job
A job is five decisions, then it is scheduled
The wizard is deliberately short. You are not composing posts one at a time — you are describing a batch and letting the app write it out.
- Which site the posts come from.
- Which posts or recipes go into the batch.
- Which Pages and accounts they go to.
- Which style writes the copy, per platform.
- Which time slots, spread across days or weeks.
When you confirm, every row is written, imaged and placed on the calendar: which post, to which destination, at which minute, all of it visible before it is scheduled. Forty posts across six Pages is one pass through this wizard, not forty passes through Business Suite.
Three platforms
Facebook, Instagram and Pinterest do not work the same way, so the app does not pretend they do
Treating the three as one "social" checkbox is where most tools quietly hide a compromise. Here the differences are handled explicitly, because they change what you can rely on.
the post is handed to Meta with a scheduled publish time. Meta holds it and publishes it, so it goes out whether or not anything of ours is awake; the ledger checks afterwards that it really appeared.
there is no scheduling API. Anyone who claims otherwise is holding the post themselves, and so do we: the app keeps the row and publishes at the minute. More usefully, it writes the caption and generates the image 24 hours ahead. A caption you have never read is a caption you cannot fix, so the day before the slot it is sitting there, readable and editable, and you can change it or throw it out while that is still free.
the app writes the pin title, description and keywords, matches the pin to a board by keyword, and exports Pinterest's official bulk-upload CSV: the real template, with the exact columns and date format the importer accepts. You upload the file. That is honestly what Pinterest allows, and the page is not going to call it something else.
Instagram has no scheduling API. So the caption and the image are written 24 hours ahead — the day before the slot they are sitting there, readable and editable, and you can change them or throw them out while that is still free.
The copy
Twelve styles, not one prompt
The text does not come out of a single instruction repeated forever. It comes from styles — twelve of them, per platform, each a bank of angles: number-first, the moment, the step that decides it, what you already have, texture, regional, one swap, why it usually goes wrong, what goes with it, season, honest timing. You choose the style when you build the job, which is also what makes the measurement later worth anything: twelve styles are an experiment you can settle rather than a novelty.
And if you would rather write the copy yourself, or have a writer do it: export the job, write the text anywhere you like, import it back. The imported text goes through exactly the same checks as generated text.
Quality gates
Every caption is measured before it can publish
This is the part that matters at scale. Getting an AI to write a hundred captions takes an afternoon; the trouble starts on day thirty, when one of them drifts into the wrong language, another opens with a hashtag wall, and a third gives away the entire method so nobody clicks. At one account you catch that by reading everything. Across a dozen sites and sixty Pages you do not.
So generated text does not go straight to a Page. It goes to a validator that checks it against explicit, mechanical rules:
- No hashtags.
- No URLs in the caption body.
- No stiff formal address — it should read like a person talking, not a notice.
- One language, from the first word to the last.
- No engagement bait.
- The first line fits inside Instagram's visible budget, before the "more" cut.
- The dish or subject is named first, not buried in line four.
- A pointer to the bio, where the link actually is.
- It must not give away the whole method — the caption earns the click, the site keeps the payoff.
Failing the check triggers a rewrite, not a shrug
A caption that fails gets a corrective retry against the specific rule it broke. A caption that fails hard refuses to publish: the row stops and is flagged. That is the correct trade — an empty slot costs you one post, a bad one costs you a Page. Nothing is published because a queue slot came up, and there is no route into publishing that skips the check, including the one your own imported text takes.
Images get their own guard. Each is generated for the aspect the platform wants and carries the destination page's logo, and an originality rule stops the same picture fanning out across sibling pages inside a window — Meta reads that as duplicate content and quietly throttles it, and it is precisely the mistake a batch tool makes at scale before anyone notices. Each Page keeps looking like its own Page, even when one person is running all of them.
Measurement
Which style, which hour, which angle — measured, not guessed
Publishing is the easy half. Because every link is rewritten as a /go/ link and every click is checked for bot signatures, what you are reading is human clicks, not scraper noise. Because every post is logged and reconciled back against Meta, you can see what actually published, what quietly vanished afterwards, and for Reels the real reach and retention rather than an impression count.
The comparison and evidence views then answer the questions a spreadsheet cannot. You stop arguing about whether the number-first angle beats the texture angle and go look.
- Which of the twelve styles produced clicks, not just impressions.
- Which time slot works on which Page — they are not the same.
- Which angle inside a style keeps a Reel watched to the end.
- Which posts disappeared after publishing without anyone noticing.
Operations
When something breaks, it stops and says so
The dangerous failure mode of an unattended publisher is not silence — it is confidence: posting something wrong, on schedule, to every page, and you find out a week later. The default here is to halt and raise it, never to guess and publish anyway.
- Incidents recorded when a job or a page fails, with the job paused rather than pushed through.
- A heartbeat and a health endpoint, so a dead worker is visible in minutes instead of on Friday.
- A daily spend cap on AI generation — a runaway batch costs you a cap, not a bill.
- Per-page throttle backoff when a platform starts pushing back.
- Credential and token checks before a batch runs.
- A kill switch that stops all publishing at once.
Who it is for
It starts paying at the third site and the twentieth page
This is built for operators, not personal brands: people running more than one content website and more than a handful of social accounts, currently holding it together with spreadsheets, Meta Business Suite and a great deal of copy-paste. Recipe blogs, niche content networks, small publishers, and people who run sites on behalf of other people.
The multi-tenant structure follows that shape. A client, a brand or a language gets its own project, and nothing crosses between them unless you put it there — no shared prompt bank, no shared credentials, no shared bill.
If you run one Instagram account, this is far more machinery than you need, and you should not buy it.
Honest limits
What Vaustore is not
Worth saying plainly, so nobody signs up for the wrong thing — and it is easier to trust a tool that draws its own boundaries.
There are no customer logos and no testimonials on this page, because Vaustore is opening to its first users now.
- Not a generic AI content generator — the writing is the input to a quality gate, not the product.
- Not a link-in-bio tool.
- Not an ads platform. Nothing here buys media or manages ad spend.
- Not for a single Instagram account, or a single site.
- Pinterest is an export, not a direct publish. The app produces Pinterest's own bulk CSV; you upload it.
- No pricing yet and no roadmap promises here. Early access means early.
Questions
Answers before you ask
Does it publish automatically, or do I get to review things?
Both, depending on the platform. Facebook posts are handed to Meta with a scheduled publish time. Instagram rows are held by the app and published at the minute — and because those go out unattended, the caption and image are written 24 hours ahead so you can read and change them the day before. Every caption on every platform has to pass the validator before it is allowed out.
Why is Instagram handled differently from Facebook?
Because Instagram has no scheduling API. Meta will hold a Facebook post and publish it at the time you set; Instagram will not. So the app keeps those rows itself and posts at the minute. The 24-hour pre-write exists specifically so an unattended Instagram post is not the first time a human reads the text.
Does it post to Pinterest directly?
No, and it says so. It writes the pin title, description and keywords, matches each pin to a board by keyword, and exports Pinterest's official bulk-upload CSV in the format Pinterest actually accepts — including the details that silently reject a file. You upload it.
What happens when the AI writes something bad?
The validator catches it. Every caption is checked against explicit rules — one language throughout, no hashtags or URLs, no engagement bait, first line inside Instagram's visible budget, subject name first, a pointer to the bio, and it must not give away the whole method. A failure triggers a corrective retry against the rule it broke. A hard failure refuses to publish and flags the row. Nothing bad goes out quietly.
Can I write the copy myself instead of using the AI?
Yes. Export the job, write the text wherever you like — your own writer, your own tools — and import it back. The imported text goes through the same validator as generated text. There is no route into publishing that skips the check.
How does it stop Meta reading my Pages as duplicate content?
An originality guard tracks generated images and blocks the same one from fanning out across sibling Pages within a window. Per-page logo overlay and per-platform aspect handling keep each Page visually its own, even when one operator runs all of them.
What happens when something fails at three in the morning?
It stops. A failed AI call or a failing page pauses the job and raises an incident instead of continuing. There is a heartbeat and a health endpoint so a dead worker is visible quickly, a daily spend cap on AI generation, per-page throttle backoff when a platform pushes back, and a kill switch for everything at once.
How do you know a click is a real person?
Outbound links are rewritten as short /go/ links and every click is recorded with bot detection applied, so crawler and scraper traffic is separated out. What you compare styles and posting times against is human clicks, not raw hits — and the ledger reconciles every post against Meta for real reach and Reels retention.
I manage sites for several different clients. Does that work?
That is the structure. An account holds projects, and each project keeps its own sites, Pages, credentials, writing styles and budget. Nothing is shared between projects unless you put it there.
What do I need before I can start?
The websites you publish from, and access to the Facebook Pages, Instagram accounts and Pinterest boards you want to post to. Credentials are stored per project and checked before a batch runs.
What does it cost?
Pricing is not decided yet — that is the honest answer rather than a placeholder. Early access participants get in before pricing is set and have a say in it.