Local/Web AI-Driven Content Extraction & Publishing Pipeline

Job ID: 40573755

Budget: $100 – $250 AUD

Project Brief: Local AI-Driven Content Extraction & Publishing Pipeline
Developer Scoping Brief (Automated Content Engine)
Objective:

Build a localized, automated pipeline on a macOS environment (MacBook Air) that processes a raw 500 GB video library and web data into optimized short-form videos (9:16), generates marketing copy, and schedules them to YouTube, Instagram, and X.

Core Tech Stack:

Agent/Orchestration: Max Hermes (running locally with terminal/file access) + n8n (for cloud API routing/webhooks).

Cognitive Layer: Minimax API (for text reasoning, transcript analysis, and copywriting).

Media Processing: FFmpeg & OpenAI Whisper (or similar open-source local transcription).

High-Level Workflow to Scope:

[500GB Local HD] --> [Local Whisper Transcribe] -- > [Minimax API Hooks Selection]

[n8n Scheduler] <--- [n8n Social Publish] <---- [Local FFmpeg Video Cut/Crop]

Ingestion & State Tracking: A local storage indexing system (e.g., SQLite) to track which source videos have been processed to prevent duplicates.

AI-Driven Slicing: A local workflow where video audio is extracted, transcribed, and the raw text is sent to the Minimax API. Minimax must analyze the transcript, identify high-engagement hooks, and return a structured JSON array containing precise start/end timestamps.

Local Media Render: An automated terminal routine where FFmpeg reads the Minimax timestamps, physically cuts the clip, crops it to a 9:16 vertical format, and places the final file into a "Ready to Publish" folder.

Copywriting Engine: Minimax generates platform-specific captions, hashtags, and SEO titles based on the context of the winning clip.

Publishing Handoff: A local n8n agent monitors the output folder, extracts the video and text metadata, and securely publishes them via the Instagram Graph API, YouTube Data API, and X API.

Developer Deliverables Requested:

Estimated development hours split by Local Processing (FFmpeg/Hermes) vs. Cloud Orchestration (n8n/APIs).

Recommended local database/state setup for video inventory tracking.
API integration architecture mapping Max Hermes' local JSON outputs to cloud webhooks.