Automation (Triggers, Pipelines, Webhooks)

Automation in Rankion bundles three tools: Bulk Generations (a list of keywords → a list of articles), Autopilot (run modules on a cron schedule, e.g. "AVI tracking every Monday"), and Pipelines (multi-stage workflows with named stages, e.g. Generate → Humanize → Detect → Optimize → Publish). Plus webhooks for external systems (review sources, later CMS sync). Required module for anyone publishing more than occasionally — manual clicks don't scale.

What it can do

• Bulk generations — drop a list of keywords in, Rankion generates N articles in the background.
• Autopilot — periodic module runs: frequency (daily, weekly, monthly), module (tracking, audit, freshness...), each trigger with its own config object.
• Pipelines — named workflows with stages[]. Each stage is a module call with inputs from the previous one. A failing stage stops the pipeline; the retry endpoint resumes from the failed point.
• Async + persistent — all runs execute as queue jobs (see ARCHITEKTUR-REGEL: Long-Running Processes in CLAUDE.md), status lives in the DB, the UI shows live progress.
• Webhooks (inbound) — external systems can post into Rankion, e.g. review aggregators via /webhooks/reviews/{source} (public, key auth).
• Webhooks (outbound) — pipeline stages can hit external URLs, e.g. to notify your CMS of a new article.

When to use

• You publish 10+ articles/week and manual Generate→Optimize→Publish costs too much click time.
• You want your AVI/audit tracking to run automatically every Monday without having to remember.
• You're building a real "content factory" — keywords in, published articles out.
• You're integrating Rankion into a larger stack (Make, Zapier, your own backends) and need webhook triggers.

Workflow

1. Start bulk run — POST /bulk-generations with {project_id, keywords[]}. Status GET /bulk-generations/{id}.
2. Configure Autopilot — POST /autopilot with {module, frequency, config}. Cron runs server-side (Laravel Scheduler, ticking every minute).
3. Define pipeline — POST /pipelines with name + stages array. Each stage: module, params, optional on_fail.
4. Run pipeline — via a trigger event (e.g. bulk-generation completed) or manually. The run gets its own ID, every stage update streams to the UI.
5. Retry on fail — POST /pipelines/{id}/retry resumes from the last successful stage.

API

Body of POST /pipelines:

{
  "name": "Generate-Humanize-Publish",
  "stages": [
    {"module": "ai-content-editor", "action": "generate", "params": {"length": "medium"}},
    {"module": "humanizer",         "action": "humanize", "params": {"level": "medium"}},
    {"module": "ai-detector",       "action": "detect",   "params": {"max_score": 40}},
    {"module": "content-optimizer", "action": "analyze"},
    {"module": "publish",           "action": "send_to_blog", "params": {"blog_id": 12}}
  ]
}

Body of POST /autopilot:

{
  "module": "ai-visibility-tracking",
  "frequency": "weekly",
  "config": {"day": "monday", "hour": 6}
}

Credits & Limits

• Bulk generation: 15 credits base + standard generate cost per article.
• Pipeline setup: 20 credits, one-time per newly created pipeline.
• Pipeline run: no base cost — you pay for the stages (each stage costs its normal module credits).
• Autopilot: no setup cost; every executed run costs the normal module credits.
• Async required: all operations >10 s run as queue jobs (Supervisor rankion-worker, 2 processes). On worker stall: supervisorctl restart rankion-worker:*.
• Retry limit: 3 automatic retries per stage, then manual via /pipelines/{id}/retry.

Related modules

• Webhooks API — the webhook endpoints in the API reference.
• AI Content Editor — most common pipeline stage (Generate, Score, Publish).
• Humanizer — a classic stage in any "quality-gated pipeline".
• AI Detector — natural quality gate before the publish step.
• Action Center — pipelines can auto-generate action items on certain events.
• Agentic Chat — pipelines can also be triggered via the master agent in natural language.