A decent blog post takes hours to write. Then it gets one link dump on LinkedIn, maybe a post on X, and dies with forty views while its author starts the next one. The writing was the expensive part, and the distribution got ninety seconds.
Everyone already knows the advice: repurpose it. Turn the post into a thread, a LinkedIn essay, three short takes. I wrote a whole guide on doing exactly that by hand, and the method in it works. What the method ignores is Tuesday. Nobody reopens a two-week-old blog post on a random Tuesday to write the fourth derivative post. Repurposing fails not as an idea but as a habit, so the fix is not more discipline. It is handing the boring middle to something that does not get bored.
For me that something is an AI agent. Planpost is a social scheduler built so an agent can run it: you connect Claude, ChatGPT, Cursor, Gemini CLI or another agent over MCP or the API, and it can draft, schedule and publish to Instagram, Facebook, LinkedIn, X, Threads, TikTok and YouTube. Turning an article into posts is reading, compressing and reshaping, which is language work, so the agent does that part. The scheduler does what a scheduler should: hold the posts, check them against each network's rules, and publish them on time. This post is the workflow I use on my own writing, including the parts the agent is bad at.
The workflow, start to finish
Step one is less ceremony than it sounds. Most agents can fetch a public page themselves, so "here is my new post" and the URL is enough. If yours cannot, or the article is not public yet, paste the text into the conversation. Either way the agent works from the full article, not from a summary of it.
Then the brief. Mine is usually one paragraph:
"Turn this article into a week of posts. One LinkedIn post, a thread on X, two posts on Threads, one on Facebook. Each one should make sense without clicking through. Spread them Monday to Friday at good times, and nothing on the weekend."
The agent checks which accounts I have connected, writes a separate version for each network, asks Planpost for the times my audience has responded to before, and creates the posts. Each network gets its own text, not one caption copied five times.
Each network has its own limits
This is where most cross-posting goes wrong, and where it helps that the scheduler is strict. The limits are not suggestions:
- X: 280 characters a post, so the argument becomes a thread. Planpost sends nothing to X with a link in it, because the X API charges far more for a post that carries one.
- Threads: 500 characters.
- LinkedIn: 3,000 characters, enough for a real short essay.
- Instagram: a 2,200-character caption, links in it are not clickable, and it needs an image or a video.
- TikTok and YouTube: video only.
Planpost checks every post against these rules when the agent creates it. A caption that runs long, or an Instagram post with nothing attached, comes back with the reason and what to change, and the agent fixes it and tries again. You do not find out on Thursday morning that a post failed.
The practical consequence is that a text article turns most naturally into LinkedIn, X, Threads and Facebook posts. Instagram needs an image you supply, and TikTok and YouTube need a video, so I only add those when I have something visual that belongs to the article.
The editing step is not optional
Honest assessment of what the drafts are, from someone with an obvious bias toward telling you they are perfect: they are good scaffolding. The agent reliably finds the argument of the article, splits it into thread-sized pieces, and does not bury the point in the sixth paragraph.
What it cannot know is which sentence you would fight for. The hook it picks is a reasonable choice from the text; the best hook is often the thing you almost said, the opinion behind the article, the number that surprised you while writing it. I rewrite the first line of most drafts. One trick that helps: ask the agent for three openings per post, pick the one closest to how you talk, and rewrite that. Plan for two or three minutes per post. That is editing, not writing.
The drafts also inherit whatever the article gave them. A vague blog post produces vague social posts. The agent moves quality around; it does not create it.
Turn on approval before you look away
When you connect an agent to Planpost, the consent screen has a switch called "Ask me before anything goes live". It starts off, so by default the agent's posts go out at the time it sets. Turn it on there, or any time later on the Agent page, and every post the agent schedules or publishes waits in Posts, under In review, until you approve it. Home tells you how many are waiting.
That is where I do the editing pass. I open each post, change what needs changing, and approve it, which keeps the time the agent picked. If a slot passes before I get to a post, it does not go out; it comes back as a draft for me to reschedule. For anything with your name on it, turn the switch on. The gap between "drafted" and "live" is where mistakes get caught.
Which articles earn a campaign
Not every post deserves the full treatment, and pretending otherwise is how a good habit turns into noise. My rough sorting, from my own writing:
Opinionated posts work best. An article that argues something produces a thread with tension and a LinkedIn post people disagree with in the comments, which the algorithms read as a good day. Step-by-step guides come second; each step can stand alone as a tip, so one guide can feed a month of small posts. News commentary ages fastest and should go out the same week or not at all.
The announcement post about your own product sits at the bottom, because nobody shares an announcement twice. One week of posts, then let it go.
An older article that still gets search traffic is the sleeper pick. It already proved people want the topic, and almost nobody in your current social audience has seen it. Hand the agent the URL as if the article were new.
Spacing the campaign
The lazy failure mode is drafting six posts and publishing them all on Thursday at noon. What you want is one article stretched into a week or two of presence: the thread on day one, the LinkedIn post two days later, a short contrarian take from the article's most arguable claim at the end of the week, a quote or statistic coming back the week after.
Each piece should stand alone without the article; someone who never clicks through should still get value from the post itself. Making each network's version feel native instead of syndicated is the whole subject of the cross-posting playbook.
The posts land in the same calendar as everything else you publish, so moving one is a sentence to the agent ("move the Thursday post to next Monday") or a drag in Calendar.
Watch what the campaign does, too. Five posts from one article are a small experiment for free: same idea, five framings. The agent can read the analytics for those posts and tell you which opening won, and that lesson feeds straight into the next article.
Making it a habit
The workflow only matters if it survives a normal week, so I tied it to the one moment I never skip: hitting publish on the blog. Right after that, I give my agent the URL and the same paragraph of brief. When I come back, the week is drafted and waiting in review. If your agent can run tasks on a schedule, it can also check your blog every Monday and draft a week for anything new; the posts still wait for you. The agentic social media scheduler page shows how to connect one.
The point of all of it is embarrassingly simple. The blog post already contains the week of social content; it always did. What was missing was a path from the published article to scheduled posts that does not depend on remembering. Give the article to your agent, keep the final word for yourself, and the path is there for the next thing you write.
