<aside> 👉
ps; give this guide to claude code/ cowork and ask it to explain anything youre confused about, and to set this up for you :)
make sure you understand tho, what its doing, don’t just blindly set it up.
</aside>
Every morning, Claude plans my day. Automatically. I wake up, open my notes, and there's a briefing already written.
Last week it said: "You've been talking about rewriting your welcome email since January. You've mentioned it in four different notes. You haven't started. That's your priority today."
Or the other day I was building a landing page and Claude went into my vault, found a transcript from a client call three months ago, and pulled out a sentence to use as a testimonial on the page. I'd completely forgotten that call even happened.
That's what happens when your AI actually has context.
The tool behind all of this is Obsidian. It's a stupid simple note-taking app. But when you plug it into Claude Code, it becomes something else entirely. An entire neural network for your AI agent. It connects all your thoughts (projects, people, meetings, brand docs, strategies) in a web of linked markdown files.
And Claude can traverse all of it. I connect this same vault to Claude Code, Hermes agents, and OpenClaw. Same markdown files, same context, different harnesses.
This is my full setup.
Most people don't think about this when they start using AI agents. Every time you start a new session, your AI remembers nothing. Your brand voice, your customer profile, your meeting notes, your projects. Gone. You're re-explaining yourself constantly.
Context is the real bottleneck, not the models.
The models are smart enough. They've been smart enough for a while. What they're missing is you. Your business, your decisions, your history, your thinking. Without that, you're just talking to a very capable stranger every single time.
I tried solving this a few different ways. I tried cramming everything into CLAUDE.md files. I tried keeping context docs in every project folder. I even tried pointing agents at Notion.
The Notion approach was the closest. Keep everything in one central place and just point each agent there. But that meant every time your agent needed to understand something about you, which is every single chat, it had to hit the Notion API to go find it. Slow & Token-heavy. And it wouldn't be proactive finding new information (like my meeting notes for the testimonial example from earlier).