Automated Crypto News Pipeline
A modular AI-powered crypto news pipeline using n8n, PostgreSQL, Browserless, OpenAI, OpenRouter, Telegram approval workflows, and automated publishing.
Overview
Cryptocurrency media platforms require rapid content turnaround to capture market movements. This project replaces manual monitoring and drafting with an automated system with multiple agents that aggregate international crypto news (or any other kind of news), summerize it to be concise, categorize and rate it based on the overall importance of the article, and then prepare structured drafts for editorial review.
The Problem
The editorial team was spending critical hours tracking disparate RSS feeds, manually translating foreign market updates, and copying data into their content management system. This friction led to delayed publishing times and high resource expenditure for routine curation tasks.
Architecture
The system utilizes an event-driven automation framework built inside n8n to orchestrate data ingestion, processing, and publishing:
Ingestion Layer polls monitored cryptocurrency RSS feeds and news APIs at regular intervals to capture high-priority industry developments.
Processing & Parsing Node filters out duplicate and spam articles, extracts clean text from raw feeds, and formats incoming metadata for downstream consumption.
Translation & Localization Engine routes the extracted text through an LLM automation step to translate the context accurately, ensuring technical blockchain terminology is preserved correctly in the target language.
Publishing Integration connects directly to the WordPress API, automatically mapping the translated content, tags, and source metadata into a structured draft ready for final human approval.
Results
- Complete automation of the news gathering and translation pipeline
- Zero manual copying required to move external news into WordPress drafts
- Real-time updates for international cryptocurrency market news
Technical Challenges
Maintaining terminology accuracy during translation was the primary challenge. Naive translation engines misinterpret specific blockchain terms. This was resolved by implementing precise localization prompts within the automation workflow to preserve technical context.
The second challenge involved handling API rate limits and structural inconsistencies across diverse RSS feeds. The workflow was optimized with data filtering nodes to standardize payloads before sending requests to the content management system.
Stack
n8n, WordPress API, LLM Automation, JavaScript, RSS/Feed Parsers.