I’ve been experimenting with different AI agent workflows for a while now, and I think I’ve landed on something that just works. The combo of OpenClaw and Obsidian has become my go-to setup for having a research agent that quietly does its thing in the background while I get on with my day.

The setup

The core idea is simple: I have an Obsidian vault that syncs across two machines. On one of those machines, OpenClaw controls the vault — filling it with research, organizing notes, handling backlinks automatically, and keeping everything tidy. When it’s done with a task, it pings me on Telegram so I know there’s fresh stuff waiting for me.

The beauty of it is that Obsidian syncs everywhere, including my phone. So no matter where I am, I’ve got a fully organized knowledge base at my fingertips, built and maintained by my agent. Waiting for coffee? I can skim through the research OpenClaw pulled together that morning. On the train? Everything’s right there.

The async todo list workflow

This is the part that gets me the most excited. I’ve started sharing an instructions file and a todo list with my OpenClaw agent through the Obsidian vault. The workflow looks like this:

You
Add research tasks to your todo list
Obsidian
Shared vault syncs the todo file
OpenClaw
Picks up tasks, researches autonomously
Telegram
Pings you when each task is done
Anywhere
Read results on phone, laptop, anywhere

Throughout my day, whenever I come across something I need researched, summarized, or looked into, I just jot it down in one Obsidian file — a simple todo list. OpenClaw picks it up, starts working through the items, and reports back when each one is done.

It turns idle moments into productive ones without me having to actively sit down and prompt an AI. I update a list, and by the time I check back, the work is done. It feels like having a research assistant who never sleeps.

What’s next: the Obsidian CLI

Obsidian recently released their CLI, and I suspect this is going to make the whole pipeline even smoother. Having programmatic access to vault operations could tighten the integration between OpenClaw and the vault significantly — think better automation for creating notes, updating indexes, or triggering specific workflows without any manual steps at all.

I haven’t fully explored it yet, but it’s high on my list.

Why this works so well

What I love about this setup is how low-effort it is once you get it running. There’s no complex infrastructure, no custom apps. It’s just a synced folder, an agent that knows how to write good notes, and a messaging integration for notifications. The tools already exist — you’re just connecting them in a way that creates something genuinely useful.

If you’ve been looking for a practical way to have an AI agent working alongside you without the overhead of building custom tooling, I’d highly recommend giving this combination a shot.