OpenClaw is insane
29 Jan 2026 · 2 min read · updated 15 Feb 2026
Obligatory OpenClaw post.

I had to jump on the clawdbot moltbot OpenClaw bandwagon. This thing is crazy.
TL;DR
The elevator pitch is the AI that actually does things and it’s not wrong. Out of the box you get:
- Runs on your machine
- Interact with the bot from almost any chat app
- Browser control
- Persistent memory
- Full system access (if you enable it, start sandboxed)
My Journey
I’ve been playing for a few weeks now. Started on a VPS, since moved to an old linux machine. Some of the workflows and things I’ve set up:
- A daily briefing with the things I care about
- Quick crypto price update
- Recent web search and shares top articles on crypto, ai and OpenClaw itself
- Local news and headlines plus weather (lol)
- Any open reminders I have
- Reminder of things I’ve asked to discuss later with the agent
- Daily notes
- Pings me every night with a summary of what we’ve done that day
- Weekly cron to summarise the dailies
- Monthly cron to summarise the weeklies
- Quarterly, yearly and you get the idea
- Every evening it runs its own mini retro
- Evaluates what’s been working well and what it’s struggled with
- Suggests improvements to
MEMORY.md(and other alignment files) for me to approve
- Periodic checking of calendar and email
So what
None of this is mind blowing. It’s safe, boring stuff that speeds me up and adds real value. Yeah I had a play and got the agent to order something from Amazon, call some friends and family for fun. But my actual use case is having the agent automate the boring stuff on my behalf.
There’s a lot of hate and a lot of mania around this. Both are probably warranted. I’d recommend playing around in whatever limits are comfortable for you. Most people will find at least one workflow worth automating.
What’s next?
Iterate. Keep improving my workflows. Particularly interested in doing things onchain with agents.