~/projects/ai-news-briefing $ cat README.md
Automated AI news briefing
A daily news briefing that puts itself together and sends an alert when a run fails.
- n8n
- RSS
- Brave Search
- LLM APIs
- Telegram
// the problem
In media relations you start the day by reading a lot: newsletters, feeds, search results. At Deutsche Telekom that collection was the same routine every morning.
// what I built
I built a workflow in n8n that pulls from around 40 sources and search queries (newsletters, RSS feeds, Brave Search), lets an LLM summarise what matters and sends the briefing out on a schedule. Every run gets checked, and if something fails, a Telegram alert goes out instead of a silent gap.
// how it works
Simple flow, with the important part at the end: checks and alerts.
┌──────────────────────────────────────────┐
│ Newsletters · RSS · search │
└────────────────────┬─────────────────────┘
↓
n8n → LLM summary
↓
Validation
├── pass → delivery
└── failure → alert// result
A small side project during my internship, built mostly in n8n. It was fun to build, and it replaced the manual collecting with a briefing that arrives on schedule.
// what I took away
Having an LLM write the summary was quick to set up. Most of my time went into making it reliable, so every run gets checked and I hear about it when one fails.