← Back to library

Turn your X outliers into multi-platform drafts

Miles Deutscher's content-repurposer idea has an agent read your X posts, pick the ones that clearly beat your baseline, and redraft them for other channels. Our version stops at drafts with an approval hold so nothing publishes until you say so.

  • OutcomePull your own X posts, score which ones outperformed, and get LinkedIn, newsletter and short-video drafts you approve before anything goes out.
  • SetupMinutes
  • LevelStarter
  • CreditMiles Deutscher (@milesdeutscher)
View original
Works with
Grok BotChatGPT agentClaude

Published Oct 23, 2026Updated Oct 23, 2026AI dry run on sample input: passed Oct 23, 2026Source verified Oct 23, 2026

When should you use it?

  • You already post on X and want the winners reused on LinkedIn, email or short video
  • You waste time rewriting the same idea for each channel by hand
  • You want drafts ready, but you will still publish yourself

Not for: Auto-posting to any network; keep the human publish step; Inventing quotes, metrics or customer stories that were not in the source post; Accounts you do not control or competitor accounts

What do you need, and what do you get?

You need

  • Access to your own X posts (or an export / recent timeline the agent can read)
  • Your baseline for an outlier (for example 2x median likes or replies in the last 30 days)
  • Target channels and any length or tone limits
  • Facts you may claim (product names, numbers) that already appear in the source posts

You get

  • A short ranked list of outlier posts with the metric that made them outliers
  • One draft per chosen channel for each outlier, labeled Ready for your review
  • No posts, scheduling or DMs sent by the agent

How does it work, step by step?

  1. Set the baselineTell the agent how you define an outlier and which of your posts it may read.
  2. Score recent postsHave it list posts above the baseline with the raw metric and a one-line topic.
  3. Request channel draftsPick which outliers to reuse and ask for LinkedIn, newsletter or short-video versions.
  4. Approve then publishEdit the drafts yourself and post from your own accounts; the agent never publishes.

The prompt

Our own prompt, inspired by Miles Deutscher (@milesdeutscher) and written to work in Grok Bot, ChatGPT agent and Claude. Values in lime are placeholders.

Fill in the blanks (optional; the copied prompt uses your values)

Prompt3 variables
Read my own X posts from {lookback_days} days (or the export I provide).
Outlier rule: {outlier_rule}.
Channels to draft for: {channels}.

1) List posts that clear the outlier rule. For each: URL or post id, one-line topic, the raw metric, and why it cleared the rule.
2) For each post I select, write one draft per channel. Match that channel's length and tone. Use only facts already in the source post. Mark each draft "Ready for your review".
3) If a post's spike depended on private context that will not travel, say so and skip drafting it.

Never publish, schedule, reply or DM. I post every final version myself.

Inspired by Miles Deutscher (@milesdeutscher): X thread: 10 Grok Bot use cases (second brain, repurposer, subscription audit, personal X assistant). The prompt above is written by playbooks.bot in our own words; nothing from the original is quoted.

What can go wrong?

  • Do not let the agent schedule or post; an approval hold is the whole point.
  • Keep every claim limited to what the original post already said.
  • Skip posts that only spiked because of a private reply or a one-off news event if that context will not travel.
  • Platform APIs and rate limits change; confirm the agent can still read your posts before you rely on a weekly run.

Our take

Repurposing only works when you start from proof, not from a blank page. Deutscher's outlier idea is useful because it forces the agent to measure your own posts first, then rewrite the ones that already worked. The safety rule that matters is the hold: drafts only, human publish. We also keep the agent from inventing proof points that were never in the source post, which is how repurposed LinkedIn copy usually goes wrong. Distinct from a full social calendar engine, this is a narrow loop: score, pick, redraft, wait. In our dry run we fed a fictional creator timeline of twenty posts with likes and replies, a 2x-median baseline, and targets of LinkedIn plus a 120-word newsletter blurb. The agent flagged three outliers, skipped a reply-thread spike that had no standalone hook, produced channel drafts that reused only claims present in the originals, and labeled every piece Ready for your review with zero schedule or publish actions. Use it when you already have a posting habit on X; it will not invent a content strategy from nothing.

Where the time goes back: You stop guessing which posts to reuse and stop rewriting winners from scratch for every channel.

FAQ

Is this the same as a weekly social calendar skill?
No. social-content-engine-approval plans a calendar and waits for approval. This playbook starts from your existing X winners and redrafts them for other channels.
Will the agent post for me?
No. It stops at drafts marked for your review. You paste or schedule the final copy yourself.
Do I need Grok Bot?
The source thread shows Grok Bot examples. Any agent that can read your X posts and write drafts can follow the same pattern.

Source and credit

Inspired by Miles Deutscher (@milesdeutscher). We describe the idea in our own words, link to the original, and wrote the prompt ourselves; nothing is quoted.

Inspired by
X thread: 10 Grok Bot use cases (second brain, repurposer, subscription audit, personal X assistant)
Creator
Miles Deutscher (@milesdeutscher)
Published on
X
Original date
Oct 23, 2026
License
Not openly licensed (the poster keeps all rights), so we quote nothing and link to the original
Originally for
Grok Bot
AI dry run
Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. Outliers scored; channel drafts claim-safe; approval hold respected.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails.
Verified
Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.

Are you the creator? Claim or correct this page.

More in Marketing

All 97 Marketing playbooks →

Related playbooks

MarketingPrompt✓ AI dry run

Build an X brand assistant and hook bank

Research how you sound on X, keep a living bank of hooks and stories, and get post drafts you approve and publish yourself.

Credit: Miles Deutscher (@milesdeutscher)

Under an hour
MarketingPrompt✓ AI dry run

Find verified free backlink submission opportunities

Research up to thirty free, legitimate listing or resource pages for your site, each with a real submission URL, effort rating and a seven-day pursuit order you run yourself.

Credit: Grant O (@2whazzuup)

Under an hour
MarketingWorkflow✓ AI dry run

Refresh published articles from live search data

Run one agent that researches and drafts a daily article for a target keyword, and a second that revisits published pieces on a biweekly cycle, guided by how each page performs in search.

Credit: Cody Schneider (@codyschneider)

An afternoon+
MarketingWorkflow✓ AI dry run

Get on-brand email campaigns built as drafts for you

Have an agent assemble a complete, on-brand email campaign as an editable draft in your email platform, so your job is a short review and send.

Credit: Will Roman (@willroman)

An afternoon+
MarketingWorkflow✓ AI dry run

Let a browser agent edit and migrate your Substack posts

Hand three publishing chores to a browser agent, editing a live post, moving a site article to a new newsletter post, and adding a note to an old post, without emailing subscribers by mistake.

Credit: Prisca Onyx

Under an hour