AI video creation & multi-platform publishing
End-to-end n8n automation that generates short-form video from daily triggers, runs Veo3 generation, logs outputs to Sheets, and publishes across major social platforms—hands-off after setup.
Results at a glance
Daily unattended generation and publishing cadence
Script, captions, and structured Veo3 prompts from LLM agents
Eight-platform distribution from one orchestrated graph
Challenge
A content-heavy programme needed repeatable short-form video without a daily production team: scripts and prompts had to stay on-brand, rendering had to tolerate async API latency, and every publish had to land in the right channel accounts with traceability.
What Habrig built
- Operator-facing visibility through Sheets and execution logs rather than a bespoke portal—aligned with how creators already review assets
- Scheduled trigger kicking off LLM steps for script, captioning, and Veo3-ready prompt formatting
- Dedicated subgraph for Veo3 submission, polling/wait handling, and persistence of the final asset URL
- Branching publish paths per platform with shared guardrails (rate limits, retries, failure alerts)
- Credential isolation for LLM, Veo3, Sheets, and each social integration
- Execution history retained long enough to audit failed publishes without rerunning expensive renders
Outcomes
- Reduced manual editing to exception handling instead of daily assembly-line work
- Predictable daily drops across Instagram, TikTok, YouTube, Facebook, Threads, X, LinkedIn, and Pinterest-style endpoints
- Single source of truth in Sheets for “what shipped today” and downstream analytics hooks
Technology
frontend
Lightweight review surfaces (Sheets + notifications); no heavy custom UI required for operators
backend
n8n orchestration with GPT-class agents for scripting and prompt shaping, Veo3 API integration, and multi-connector publishing fabric
database
Google Sheets as the operational ledger for generated URLs, captions, and publish status
infrastructure
Hosted n8n with encrypted credentials and queue-friendly pacing for API-heavy steps
monitoring
Per-node failure routing, execution replay paths, and owner alerts when a platform rejects a payload
cicd
Workflow JSON exported and promoted through staging workspaces before touching production credentials
Execution detail
Product & frontend
- Operator-facing visibility through Sheets and execution logs rather than a bespoke portal—aligned with how creators already review assets
Backend & data
- Scheduled trigger kicking off LLM steps for script, captioning, and Veo3-ready prompt formatting
- Dedicated subgraph for Veo3 submission, polling/wait handling, and persistence of the final asset URL
- Branching publish paths per platform with shared guardrails (rate limits, retries, failure alerts)
Platform & delivery
- Credential isolation for LLM, Veo3, Sheets, and each social integration
- Execution history retained long enough to audit failed publishes without rerunning expensive renders
Gallery
Plan your next release
Tell us what shipped, what is at risk, and what success looks like. We will respond with a practical path.
Book a consultation