Telegram Autoposter
Automated content aggregation, enrichment, and publishing system for Telegram channels.
Client:
Cooking Company, Poland
Project Overview
The client engaged Axis to build a Telegram autoposter capable of autonomously curating and sharing news articles. The system collects articles from RSS feeds, processes them through AI for summarization and filtering, manages duplicates, and posts the final content to a Telegram channel on a defined schedule. It is deployed on a virtual machine and designed for continuous, low-maintenance operation.
Challenge
Reliance on external AI provider (together.ai) created billing and availability risks.
RSS parsing broke when websites changed feed delivery methods, requiring parser rework.
Client requested future functionality beyond RSS, including image + text parsing from Pinterest, demanding more complex scraping and content formatting.
Tech Stack
Backend & Orchestration: Node.js with TypeScript, task scheduling, and async job orchestration.
AI / LLM Processing: together.ai API integration for summarization, filtering, and generation.
Data Storage: PostgreSQL (article tracking, freshness control, deduplication).
Message Delivery: Telegram Bot API for secure and reliable channel posting.
Deployment & Runtime: Linux-based Virtual Machine with background process management (PM2/forever).
Content Acquisition: RSS/XML parsers with fallback handling for non-standard feed formats.
Solution
Axis delivered a configurable autoposter with:
Adjustable retrieval and posting frequency (recommended 1:4 ratio).
Content filters to enforce minimum length and quality.
Database cleanup logic (automatic purging after
REMOVE_HOURS).Error resilience – continues running after terminal closure, robust scheduling.
