Signal
The world is drowning in information overload and digital "brainrot." Our feeds have become a chaotic blur of distraction where quality is buried under noise.
Core trick
A Gemini computer-use agent scrolls feeds in a browser, and a memory.md file of natural-language preferences rewrites its instructions.
Description
The world is drowning in information overload and digital "brainrot." Our feeds have become a chaotic blur of distraction where quality is buried under noise. Curating this manually is a full-time job that risks biased rabbit holes and radicalization.
Enter Signal.
Signal is an autonomous AI agent built on the Google ADK that acts as your digital filter. Using Gemini’s computer_use capabilities, Signal physically scrolls and curates your Instagram, X, and news feeds on your behalf. Controlled entirely via a simple Telegram bot, it delivers a high-signal briefing every morning tailored to your actual interests.
No complex RL systems or heavy vector databases here. Signal leverages Gemini’s reasoning to update its behavior based on a simple memory.md file. Want to change your focus? Just tell it in natural language. Signal refines its instructions, filters the junk, and restores your focus.
Improve the signal. Kill the noise.
README excerpt
# Signal — Personalised Noise Reducer > Improve the signal. Reduce the noise. We live in a world of information overload. Our feeds are simultaneously our best source of knowledge and our biggest distraction. We scroll through hundreds of posts, newsletters pile up, and algorithms optimise for engagement — not for us. Before we know it, we're stuck in rabbit holes, our feeds are radicalised, and the content that actually matters gets buried. Curating all of this manually? Impossible. Tuning your Instagram algorithm, filtering newsletters, diversifying your news diet — it's a full-time job. And even if you try, you're still one rage-bait click away from a feed that no longer serves you. **Signal** is an autonomous AI agent that does this for you. ## The Idea Signal is built on a simple premise: **what if an AI agent could manage your entire information diet?** Not by building another feed algorithm that optimises for your attention. Not by training neural networks on your engagement patterns. But by giving an AI agent your actual preferences — in plain English — and letting it go to work. It reads your news. It scrolls your Instagram. It filters your content. It learns what you care about. And critically — it watches for the patterns you can't see yourself: echo chambers forming, rage bait creeping in, your feed slowly radicalising. ### What Makes This Different **No complex ML pipeline.** Signal doesn't use reinforcement learning, vector databases, or embedding simi