Back
Post
How are you using OpenClaw?
Just curious for inspiration. Seems it’s getting safer as they’re patching vulnerabilities so considering trying it out!
New to WIP?
We're a group of makers shipping together. We help each other stay accountable and reach our goals.
Join now
👋 Join WIP to participate
Hey! Running two agents currently:
Marvin (coordinator): part executive coach, part program manager, part agent orchestrator. Handles infrastructure, builds out my productivity tools, manages planning and tracking. Basically keeps the balls in the air so I don't drop any.
Ace (CMO): owns growth/marketing for one of my products. Just migrated him to a dedicated Mac Mini yesterday (wrote about the chaos: jamiewatters.work/journey/ai-agent-migration-checklist) He's trained on the product vision, mission, positioning, ICPs, architecture he can make product suggestions, but I don't let the Bot do coding. He's very much a PoC then, once I iron out the creases I will replicate him for each product.
Both connected to Telegram so I can chat with them from my phone when I'm about. Marvin handles my emails, Ace tracks all product channels.
The big win: the agents free me up to focus on product design, coding, and new features — the stuff I actually enjoy. All the planning, tracking, and "running the business" work happens in the background. In particular, the stuff I avoid (marketing) is actually starting to get momentum.
Building toward 10+ specialized agents — one per product vertical. The recent security patches were a big reason I felt comfortable scaling up and moving to a dedicated Mac Mini.
I'm keeping Marvin on AWS for resilience, if I lose the Mac I still have Marvin; like wise if the AWS goes down I still have ACE.
Happy to share more if you're considering it!
How much do you spend on all this?
I already had the Mini, which I'd used for dev, but moved the dev back to my Macbook and cleared the mini down. I use my claude max plan so that was a sunk cost. I'm paranoid about one of the agents deciding to use my openrouter api as it would probably cost me more than a holiday to Europe. The AWS is the only cost and is coming in about $20pm, but I spun it up using the AWS free intro so should be $0 for at least the next 4 months while I burn the free credits. I wrote this about setting up for free on AWS: ( jamiewatters.work/journey/how… ) I had to upgrade disk and storage over this spec but it's not costing much more.
thank you!
Thank you for the long answer! That all makes sense.
One thing I don't get is: why the mac mini? Why not just using a VPS for this? In the end, you're using LLMs in cloud right? Or do you run them locally on the Mac to save on costs?
Thank you!