I’ve noticed many influencers on Twitter posting informative tweets almost every hour, and they seem to be performing really well. I’m building an AI news aggregator that collects the best AI news of the hour and shares it with me. It also scores each news item out of 10. Any news item that scores above 8 will be synthesized into a tweet draft on my behalf. All tweet drafts will follow a specific format inspired by the best Twitter profiles I’ve studied and collected. I'm building this using Codex and not Claude code. I’ve been trying to create my own AI-powered news and content workflow. The goal is simple: I want a system that tracks important AI updates, understands which ones actually matter, and helps me turn them into high-quality X posts that I can review before publishing. Right now, the system does a few things: - Tracks AI updates from official sources like OpenAI, Anthropic, Google, NVIDIA, Hugging Face, etc. - Stores everything in a database - Deduplicates similar news items - Scores each update based on importance, novelty, audience interest, and practical impact - Uses my Notion “second brain” to understand my voice and tweet style - Generates draft X posts - Sends drafts to Telegram - Lets me approve or reject them manually The part I’m most excited about is the second brain layer. I’m building a Notion workspace where I store: - tweet examples I like - Twitter profiles I want to learn from - AI news sources I want to track - my voice profile - content rules - approved and rejected drafts - The idea is that instead of asking AI to write generic content, I’m slowly teaching it my taste. So the workflow becomes: AI news comes in → system ranks it → AI writes in my style → I approve from Telegram → later it can be published. I’m not trying to make a fully autonomous posting bot. I want a human-in-the-loop system where AI does the heavy lifting, but I still make the final call. This has been a really interesting learning process because I’m not a software engineer. I’m learning how systems are built step by step: databases, workflows, APIs, n8n, Notion, Telegram bots, prompts, and automation.