Hello and welcome back to The Mindshift AI Inference!

I built something yesterday that I want to share with you.

It is a research assistant that runs while I sleep. Every morning at 5am, it wakes up, reads RSS feeds from writers I follow, and asks: "Is this relevant to what Matthias is working on right now?"

The key word is right now.

The assistant reads my current article drafts. It reads my ideas file. Then it filters articles based on that context. Not keywords. Not categories. What I am actually thinking about in the last few days.

Articles that pass the filter get analyzed in depth. The assistant writes a short digest and drops it into my vault (I use Obsidian to write and organize my thinking). When I wake up, I can read the file and get inspiration to learn something new or I might have an interesting link to reference in my writing.

Here is what makes this different from Feedly or Pocket or any other reading tool:

It knows what I am working on.

If I am drafting a post about terminal AI tools (which I am right now), it surfaces articles about CLI automation. If I am exploring voice workflows, it finds pieces on audio transcription. The filtering adapts to my current focus.

I built this in a single afternoon with Claude Code. The entire system is about 700 lines of Python. It runs for free on GitHub Actions.

You can find the code here: github.com/matthiasroder/research-assistant

This is part of a larger creativity system I am building. A system where AI handles the logistics so I can focus on the ideas.

A second brain that actually thinks.

Over the coming weeks, I will share more about how I work with Claude Code to build these tools. Not tutorials. Real workflows. The messy, exploratory process of figuring out what is possible.

If that sounds interesting, stick around, and if you are reading this on LinkedIn, make sure to subscribe to my newsletter at http://matthiasroder.com

Have a great day!

Matthias