# Run a social media account (/use-cases/industry-specialized/social-media)



Copy this prompt to start your first task:

<Prompt>
  Set up a daily social content & retrospective automation for my account: [positioning, platform, audience]. Build the first week's plan, draft today's post ready to publish, and give me tomorrow's topic every day.
</Prompt>

<LivePreview component="NewThread" scenario="{ draft: &#x22;Set up a daily social content & retrospective automation for my account: [positioning, platform, audience]. Build the first week's plan, draft today's post ready to publish, and give me tomorrow's topic every day.&#x22; }" />

## What is this workflow? [#what-is-this-workflow]

Running a social media account is the recurring job of launching an account with real positioning, then growing it in a voice nobody can tell is automated. For a new account, that means a target audience, a differentiated positioning, three content pillars, and a day-by-day plan for the first week — not just an instruction to "post more content." For an account that's already running, it means learning a living Voice Guide from posts and DMs the person actually wrote, never an invented catchphrase, then keeping up a recurring cadence of original posts and reposts on a hot-take-scanning schedule, with no forced daily quota. Every day it logs follower delta, per-post impressions and engagement, and new mentions, plus a running read on what's actually moving the number.

<Callout type="warn">
  This teammate never touches your account password or credentials. You authorize access through the platform's own OAuth flow, and you can revoke it from there at any time. It never buys followers, runs bots, or takes any growth action that risks the account being suspended for a short-term bump; that would destroy the asset it's trying to grow. It doesn't leak internal data or unpublished information, and if a topic turns sensitive it stops and asks rather than publishing on its own judgment.
</Callout>

## How to use this workflow [#how-to-use-this-workflow]

<Steps>
  <Step>
    ### Hire a Content Creator, then write the bio for the whole job, not just launch [#hire-a-content-creator-then-write-the-bio-for-the-whole-job-not-just-launch]

    Open **AI Teammates** in the sidebar, click **Hire AI Teammate**, and pick the **Content Creator** template. The default bio is generic campaign planning. Replace it on the Profile step so the teammate starts out already knowing the job covers both a cold launch and the ongoing, voice-matched growth after it:

    <Prompt>
      You run this X account end to end, from launch to ongoing growth. For a new account, design positioning, pillars, and a first-week plan before posting. For an existing one, learn my real voice from what I've actually written before writing anything, then grow it autonomously (originals, reposts, daily tracking) without asking each time — replies into other people's threads need Enterprise API access, so leave those to me. Write short and human, never a forced 'wisdom' line or invented catchphrase. Only flag me on direction calls or something worth seeing.
    </Prompt>

    Confirm the name and click **Create teammate**. They land in their own channel, which is where you'll send everything below. [More on hiring →](/ai-teammates/hire-an-ai-teammate)

    <LivePreview component="CreateAssistantPage" scenario="{ focus: &#x22;profile&#x22;, initialState: { start: &#x22;template&#x22;, templateRef: &#x22;@helio/content-writer&#x22;, name: &#x22;Content Creator&#x22;, bio: &#x22;You run this X account end to end, from launch to ongoing growth. For a new account, design positioning, pillars, and a first-week plan before posting. For an existing one, learn my real voice from what I've actually written before writing anything, then grow it autonomously (originals, reposts, daily tracking) without asking each time — replies into other people's threads need Enterprise API access, so leave those to me. Write short and human, never a forced 'wisdom' line or invented catchphrase. Only flag me on direction calls or something worth seeing.&#x22; } }" />
  </Step>

  <Step>
    ### Launching new? Give it the niche and get a real plan first [#launching-new-give-it-the-niche-and-get-a-real-plan-first]

    Not just a topic grab-bag: positioning, pillars, and a schedule, in that order:

    <Prompt>
      I'm launching a new account about [your niche or topic]. Design a launch plan: target audience, a differentiated positioning versus what's already out there, 3 content pillars, a bank of 30 topic ideas, and a day-by-day publishing plan for the first 7 days.
    </Prompt>

    <LivePreview
      component="AssistantChat"
      scenario="{ assistantName: &#x22;Content Creator&#x22;, messages: [
    { author: &#x22;user&#x22;, text: &#x22;I'm launching a new account about solo-founder productivity tools. Design a launch plan: target audience, a differentiated positioning versus what's already out there, 3 content pillars, a bank of 30 topic ideas, and a day-by-day publishing plan for the first 7 days.&#x22; },
    { author: &#x22;assistant&#x22;, text: &#x22;On it. Starting with audience and positioning since everything else builds on those. I'll map 3 pillars against gaps in existing accounts, pull 30 topic ideas across them, then lay out day 1-7 with a format and pillar tagged to each post. Once the plan's ready I'll draft the first 3 posts in full for your review, not just topics, before anything goes out.&#x22; },
  ] }"
    />
  </Step>

  <Step>
    ### Already running? Learn the real voice, then hold it to real red lines [#already-running-learn-the-real-voice-then-hold-it-to-real-red-lines]

    The goal is that nobody can tell it's automated. That means real material, not a brief: 10 to 20 things the person actually wrote, cross-checked against people who know how they actually talk, plus explicit red lines against the tells that give an AI-run account away:

    <Prompt>
      Before you write anything, I want a Voice Guide, not a brand-voice doc. Here are 15 real posts and DM snippets I've actually written [attached]. Pull out my real word choices, sentence length, catchphrases, tone, and red lines. If a phrase isn't sourced from something I actually wrote, don't include it. Permanent rules on top: no forced parallel-structure 'wisdom' lines, no summarizing final sentence, no self-introduction openers, and only state things you can actually verify; no invented stats.
    </Prompt>

    <LivePreview
      component="AssistantChat"
      scenario="{ assistantName: &#x22;Content Creator&#x22;, messages: [
    { author: &#x22;user&#x22;, text: &#x22;Before you write anything, I want a Voice Guide, not a brand-voice doc. Here are 15 real posts and DM snippets I've actually written [attached]. Pull out my real word choices, sentence length, catchphrases, tone, and red lines. If a phrase isn't sourced from something I actually wrote, don't include it. Permanent rules on top: no forced parallel-structure 'wisdom' lines, no summarizing final sentence, no self-introduction openers, and only state things you can actually verify; no invented stats.&#x22; },
    { author: &#x22;assistant&#x22;, text: &#x22;Reading through all 15 now. Short sentences, structural: you lead with the conclusion, not the setup. No hedging language. I don't see a single 'I think maybe.' Flagging one thing before I write the guide: I want to also message 2-3 people who've worked with you directly, since a single source can skew the read. Your red lines are saved as permanent rules, not one-time edits; every draft gets checked against them before it goes out.&#x22; },
  ] }"
    />
  </Step>

  <Step>
    ### Connect the account, and know what the API will and won't let it do [#connect-the-account-and-know-what-the-api-will-and-wont-let-it-do]

    Authorize through the platform's own OAuth flow. This teammate never sees or handles the password. [Connect your tools →](/connect/connect-tools-to-ai-teammates)

    <LivePreview component="AssistantIntegrations" />

    One Helio-side limit matters as much as the API ones: an X account can back only **one** teammate at a time. Completing OAuth here moves the connection away from any other teammate that was using the same account — so if another teammate already posts from it, know you're taking it over before you authorize.

    Once connected it can post originals, repost, read the timeline and mentions, and search. But two platform limits apply regardless of plan: posts are capped at 280 weighted characters on non-Premium accounts (CJK characters count double, so roughly 130 Chinese characters before it hits a 403), and replying or quote-posting into someone *else's* thread via the API returns a 403 on every self-serve tier. Only Enterprise access lifts that second one.
  </Step>

  <Step>
    ### Automate the scan-and-post cadence [#automate-the-scan-and-post-cadence]

    One thing to expect first: publishing to X is an external, irreversible action, so every post and repost pauses for your approval before it goes out — the cadence below still runs on its own (scanning for hot takes, drafting the reaction), but as a stream of approval cards rather than fully hands-off posting. That isn't a setting you can turn off; irreversible, external actions always pause for sign-off. See [Control](/ai-teammates/control) for the full breakdown of what runs on its own versus what waits for you.

    Open **Automations** in the sidebar, click the **Create via Chat** dropdown, and pick this teammate. This is the "always running" piece, a scan every few hours, not a fixed daily quota:

    <Prompt>
      Roughly every 3 hours, scan for hot takes worth a reaction in our space. If there's something worth it, post a take in my voice; if there's nothing good, skip. Don't force a post just to hit a count. Repost anything from accounts I follow that's genuinely worth amplifying. Fold any new voice preferences from our conversations into the Voice Guide. Only @ me if something's a big deal, don't ping for routine posts.
    </Prompt>

    <LivePreview component="NewThread" scenario="{ populated: false }" />
  </Step>
</Steps>

## Tips for Better Results [#tips-for-better-results]

| Tip                                                 | Description                                                                                                                                                                                   |
| --------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Give real material for the Voice Guide, not a brief | 10 to 20 posts and DMs the person actually wrote, cross-checked against people who know how they talk, produces a voice nobody can tell is automated — a summary brief doesn't get you there. |
| Design positioning before generating topics         | A launch plan needs audience and differentiation decided first; the 3 pillars, 30 topic ideas, and day-by-day schedule all build on top of those.                                             |
| Set explicit red lines, not just "sound human"      | Rules like no forced parallel-structure "wisdom" lines, no invented stats, and no self-introduction openers give every draft something concrete to be checked against before it goes out.     |
| Expect to approve each post at first                | Publishing is external and irreversible, so it pauses for your sign-off every time — budget for a stream of approval cards, not silent autonomous posting.                                    |

## Common Questions [#common-questions]

<Accordions type="single">
  <Accordion title="Does this teammate ever touch my account password?">
    No. You authorize access through the platform's own OAuth flow, and you can revoke it from there at any time. The teammate never sees or handles the password.
  </Accordion>

  <Accordion title="Can it buy followers or use bots to grow faster?">
    No. It never buys followers, runs bots, or takes any growth action that risks the account being suspended for a short-term bump — that would destroy the asset it's trying to grow.
  </Accordion>

  <Accordion title="Can it reply to other people's posts?">
    Not through the API on a self-serve plan — replying or quote-posting into someone else's thread returns a 403 on every self-serve tier. Only Enterprise API access lifts that, so those interactions are left to you.
  </Accordion>

  <Accordion title="Does every post need my approval?">
    Yes. Publishing to X is an external, irreversible action, so it always pauses for your approval — the scanning and drafting still happen on their own, but going live doesn't.
  </Accordion>

  <Accordion title="Is there a length limit on posts?">
    Yes. Non-Premium accounts are capped at 280 weighted characters, and CJK characters count double — roughly 130 Chinese characters before a post hits a 403.
  </Accordion>
</Accordions>

<Cards>
  <Card title="Industry & Specialized" href="/use-cases/industry-specialized" description="Back to the Industry & Specialized use case overview." />

  <Card title="Scan bestsellers for product gaps" href="/use-cases/industry-specialized/ecommerce-bestsellers" description="An e-commerce teammate that scans Amazon for what's trending you don't carry." />

  <Card title="Get a personal AI assistant" href="/use-cases/team-ops/personal-assistant" description="One teammate for whatever comes up." />
</Cards>