Content that gets better every week.
Marc writes the post, generates the reel, publishes it to your accounts and measures what happened — then rewrites its own prompt from the numbers. Week four outperforms week one because the system learned, not because you tried harder.
See how the loop worksIllustration of the pipeline — not a real account.
Most tools stop at “posted”. That is where the useful part starts.
A generator gives you more content. It has no idea whether any of it worked, so the thousandth post is a guess made exactly as blindly as the first. Marc keeps the post attached to the prompt that made it, so the numbers have somewhere to go.
It writes
You give it a topic. Claude writes the hook, the body, the close, the caption — and every image and video prompt behind them.
You publish
Approve the prompts, watch the reel build, then send it to your accounts or put it on the calendar.
It measures
Views, likes, saves, watch time and engagement come back hourly while the post is fresh, then daily.
It rewrites itself
Claude reads those numbers against the exact prompt that produced them, and writes the better version of it for next time.
…and back to the top, with a better prompt than it started with.
Your numbers are worth nothing spread across three dashboards.
TikTok knows the views. Instagram knows the saves. Neither of them knows what the post was trying to do, who wrote it, or which idea it came from — so neither can tell you anything useful about what to make next. Marc keeps the topic, the prompt, the reel and the result attached to one another. That linkage is the whole asset: it is the only condition under which a model can say something about your content instead of about content in general.
It finds the pattern you would not have
Across every account and platform at once — which hooks hold attention past three seconds, which formats get saved rather than liked, which topics only ever work on one platform.
It tells you when something has stopped working
Formats decay. The angle that carried you through March quietly stops earning its slot, and the numbers say so long before the feeling does.
It writes the next brief from the evidence
Not a generic best-practice list — the specific prompt, for your audience, with the reasoning shown so you can argue with it.
Ask Marc
A question box over your own post history. Not a chatbot with opinions about marketing — Claude, reading your actual reels, your actual prompts and your actual numbers, and answering with the posts it drew the answer from so you can check its work.
Scoped to one workspace at a time, so a client's numbers never inform an answer about another client's.
- Why did last week underperform?
- What should I post on Thursday?
- Which hook style works best for this client?
- Is this format still working, or is it done?
What's possible
Spin to the kind of content you want to be making and see what comes out the other end.
Work that is visual by nature and almost never gets filmed.
Drag, scroll or use the arrow keys. Example topics — you type your own.
Built to be run every day
Whether that is one brand posting daily or a studio running content for a dozen clients.
An approval gate on everything
Every prompt is editable before a cent is spent on generation, and nothing reaches an account until you have approved the post itself.
Batch runs
Up to twenty-five topics in one go. Each comes back as a complete post package — copy, captions, image prompts, video prompts — to review before anything executes.
A workspace per client
Content, connected accounts and billing belong to a workspace and never leak between them. Each teammate gets a role, from view-only up to owner.
More than one account per platform
Pick the accounts a post goes to, write the caption once, and send it now or put it on the calendar.
Measurement while it matters
Views, likes, saves, watch time and engagement pulled back hourly while a post is still fresh, then daily. Per post, per account, per platform.
Nothing loses its history
Every published post traces back to the reel, the prompt and the topic that produced it — which is the only reason the loop can improve anything.
Which one are you?
The pipeline is the same. What it saves you is not.
A dozen clients, one machine.
Every client gets a workspace of their own. Content, connected accounts and billing belong to that workspace and never leak into another one.
- Separate workspaces, with roles from view-only up to owner
- Per-client reporting that traces every post back to the prompt that made it
- Batch runs, so a week of content for a client is one sitting, not five
You keep control of the accounts
You connect an account you already own, and Marc does exactly two things with it: publish what you approve, and read back how it performed. Nothing else — and here is that in full, per platform.
| Platform | What it reads | What it writes |
|---|---|---|
| Your Pages, and video insights for videos Marc published (views, reach, watch time) | Publishes videos to a Page you select | |
| Your Business/Creator account, and insights for Reels Marc published (views, reach, likes, comments, shares, saves, watch time) | Publishes Reels to the account you select | |
| TikTok | Your basic profile information, and statistics for videos Marc published (views, likes, comments, shares) | Publishes videos to your account |
You can disconnect any account at any time, which revokes our access immediately. See the privacy policy for how long data is kept and how to delete it.
Questions people ask first
What is Marc?
What does "it learns from the numbers" actually mean?
Do I have to approve things, or does it just post?
Do you sell social accounts, or post from accounts you own?
Which platforms can it publish to?
Where does my data go, and can I delete it?
Stop making content into a void.
Publish it, measure it, and let the numbers pick the next one.