How I Read with My Personal AI Assistant
We all save more links than we can ever read. My “read later” queue used to be a graveyard of good intentions. Now, my personal AI assistant reads them for me.
Here is the exact workflow I use to process articles, extract what matters, and route those insights directly into my second brain.
The Reading Skill
My assistant (running via OpenClaw) operates using a modular skill system. One of these modules is the Reading Skill.
When I find an interesting article or YouTube video on my phone or laptop, I drop the link into a dedicated Discord channel (#reading) and tell the agent to read it.
Behind the scenes, the agent uses browser tools or text-extraction APIs to pull the raw content. But it doesn’t just give me a generic summary. It filters the content through the lens of my current projects.
Because the agent has access to my global context—it knows I manage SRE at Billie, teach DevOps at Harbour.Space, and consult on AI agent infrastructure—it highlights the exact pieces of the article that matter to me. It ignores the fluff and gives me a dense, practitioner-focused breakdown.
Connecting to the Second Brain
Reading is only half the battle. The real value is retention.
Once the agent processes an article, it doesn’t just leave the summary in chat. It reaches into my local file system and modifies my Obsidian vault.
Specifically, it appends a one-sentence punchy insight and a reference link directly into my Daily Note under a ## Reading & Learning section.
If the content is highly relevant to a long-term project—say, a deep dive on Kubernetes multi-cluster routing or a new framework for agent observability—the agent asks if I want a dedicated note. If I say yes, it generates a clean Markdown file with proper frontmatter, tags, and structure, saving it directly into my active knowledge base.
Why This Matters
This workflow changes the fundamental economics of consuming content.
- Zero Context Switching: I don’t have to leave the chat to save a note. The agent does the filing.
- High Signal, Low Noise: I only read the dense summaries tailored to my actual work. If a summary proves the article is irrelevant, I just saved 15 minutes.
- Compound Interest: Because every insight is automatically linked in my daily notes, my Obsidian vault grows organically. When I sit down to write a LinkedIn post or a technical design document, the research is already formatted and waiting for me.
The goal isn’t to read faster. The goal is to extract value without the cognitive overhead of manual filing. And with a personal agent, the filing system runs itself.