Jarvis Daily Briefing Agent
Generates tailored morning briefings from news, stocks, and calendar events using LLMs and text-to-speech pipelines.
Tech stack: Python, Telegram Bot, GitHub Actions
Live: https://placeholder.com
Repo: https://github.com/placeholder
Problem
Starting the day productively means knowing the weather, top news, calendar commitments, and market moves before opening any app. Checking five sources manually every morning wastes 15-20 minutes and often gets skipped.
Results
Running fully automated since April 2025 via GitHub Actions. Delivers briefings to Telegram and Gmail every morning. A live static site on GitHub Pages archives past briefings.
Challenges
- Aggregating data from 4+ heterogeneous API sources reliably, with graceful fallbacks when any individual source is down or rate-limited.
- Prompting the LLM to produce a brief that feels personal and concise rather than a wall of AI-generated text.
- Formatting the same content for both Telegram's MarkdownV2 and Gmail's HTML without duplicating the rendering logic.
Architecture
A Python script pulls data from weather APIs, news RSS feeds, a Google Calendar integration, and stock tickers. It synthesises a structured brief and passes it to an LLM for natural-language summarisation. The final output is delivered simultaneously over Telegram (formatted message) and Gmail (HTML email). GitHub Actions triggers the script on a cron schedule — no server required.