The auto-publish loop — what runs in the background every day

    The daily rhythm: scans, gap detection, drafting, scheduling, and publishing — all without you touching a thing.

    Ask ChatGPTAsk Claude

    The auto-publish loop — what runs in the background every day

    TL;DR — Every day: scans run, gaps get detected, drafts are generated, the calendar is populated, and (on Autopilot) posts publish on schedule. You can watch it happen or ignore it entirely — the loop keeps going either way.

    The daily rhythm

    At the same time every day, for every active project:

    1. Scans run across all seven engines for every active prompt
    2. Scores update using the mentions/responses formula
    3. Gaps are re-ranked — new opportunities, disappearing ones
    4. Drafts are generated to close the top gaps (respecting your remaining credits)
    5. The calendar is populated — new drafts get slotted into open publish times
    6. On Autopilot, previously-approved-style drafts publish to connected channels
    7. A daily brief is delivered — email + optional WhatsApp — summarising what moved

    Every step is idempotent. If the loop is interrupted, the next day picks up where it left off.

    When each step actually happens

    • Scans — roughly 03:00 UTC per project, spread over ~2 hours to respect engine rate limits
    • Drafting — 04:00–06:00 UTC
    • Publishing — Autopilot posts go out at platform-appropriate times per your timezone (LinkedIn morning, Instagram midday, etc.)
    • Daily brief — 07:00 and 19:00 UTC, prioritising CET; you can pick which

    What you see the next morning

    • Fresh visibility % on the dashboard
    • The delta vs yesterday, with the top three prompts that moved
    • Any new opportunities queued for approval
    • The next 3–5 scheduled posts, previewable inline

    The safety nets

    • Manual refresh requires confirmation. If you hit "run scan now" mid-day, we require an explicit credit-cost confirmation.
    • Scan pauses. If a project runs out of credits, scans pause and we email you before continuing.
    • Publish failures. If a channel publish fails (expired token, provider outage, image regeneration failed), we retry with backoff and log the failure — nothing publishes silently corrupted.
    • Rate limits. All provider calls have retry-with-exponential-backoff wrappers so a transient provider hiccup doesn't take down your day.

    Manual mode

    If you turn Autopilot off, the loop still runs — scans, gaps, drafts, calendar. The only step that changes is publish: nothing goes live until you click Publish on a draft. The daily brief will highlight what's waiting.

    Most customers run Autopilot in trusted mode after ~3 weeks of manual review — enough time to calibrate the drafts to their voice.

    What "world-class" content generation means in practice

    The loop is only as good as the drafts it produces. That's why so much of GeoNexo is invested in brand memory, style profiles, and per-platform prompting (see How our AI content generation works). The loop is the delivery mechanism; the drafts are the product.