I think I just hired...... a robot?
I Installed OpenClaw 48 Hours Ago… and It’s starting to feel like a teammate
It’s been 48 hours since I installed OpenClaw. Since then Thor (my agent) has been making improvements to my projects, handling complicated tasks and is slowly becoming a “member of my team”… which I say with mixed feelings.
Here is a short summary of how its going so far (for anyone interested).
Why I decided to try out OpenClaw
I’m a heavy user of Claude Code for most things I do (I run a solo bootstrapped open source project) and had have some great success over the past couple of months. Last year as my project grew and more paying customers came on board I felt a natural personal bottleneck (my first hire coming soon? I thought…)… until I leaned more into Claude Code.
Since Jan 2025 (2 months) I have heavily leaned in Claude Code (4.6), I’m happy with the output, I have a bunch of automation with skills and custom prompts, workflows and practices and things seem to be clicking. We fix issues together, explore ideas, business models… all sorts. I’m still in the driving seat but my output has 5x at least (which brings other problems).
Over the past few months OpenClaw has been the hot topic, and become a very popular tool…. my first honest feelings… why would I want access to my computer on my phone (WhatsApp, Telegram) I have enough of work during the day, why should this stay with me after “work”… but anyway I decided on the weekend to try install it and give it ago, what is the fuss about….
Installing OpenClaw
I decided to keep things super simple. I found an old laptop, wiped it clean and installed OpenClaw (setup took about 30min), I named it Thor and integrated with Telegram (never used that before either). The laptop has access to nothing, but it’s own email and a GitHub account.
Next I started to talk to Thor . Nothing special going on (yet), just hello and setup some personality stuff. So far feels like LLM chat apps we have been using, but this time through my phone and telegram desktop app (the UX feels different already)
So what can this thing do? Everything people say… I tried setup Google email and drive etc, got blocked by terms of service stuff, wanted to use my Claude subscription but people say you violate terms of services… so I setup Codex 5.3.
So now with GitHub account, and Codex 5.3 what’s next?… Let’s try get Thor to fix some bugs….
Claude Code vs Thor (OpenClaw)
Now this is the strange thing, both are models…. but the interface feels different. Claude Code I can give inputs and we get things done together I have my own workflows in place. Thor is similar but feels different…..
The first thing I told Thor to investigate was low hanging easy issues on my GitHub project, go through them, make your mind up and then make some easy to review pull requests… and off “it” (he?) went. Within the next few minutes we had a bunch of pull requests coming in… simple and merged… great (nothing too crazy yet).
Lets move to harder problems…
Some users of my project have raised bugs that are hard to reproduce and solve (I know because I wrote all the code)…. so I gave Thor the task to investigate, learn, benchmark and make changes … and off it went. This time Thor (in the background) cloned the repo, understood the problems, could replicate this, benchmarked with suggestions and made a plan to improve things, and implemented the plan…. any tools it didn’t have it told me, I said “just install it” and it did and moved on…. all of this async whilst I’m doing other things with Claude Code…. like a “team member?”.
Next, I wanted to see if I could get Thor to behave like a team member….
I told Thor, to check in with me at random intervals, and when “it” wants search through GitHub to find issues, and explore them…. I told “it” to also every so often surprise me with some EventCatalog insights, competitor stuff (etc)…. and Thor made some CRON jobs. I gave Thor some permission to do a bunch of things that I think would be valuable.. and left it.
Within the next hour, Thor starts to talk to me… unprovoked, fixing issues, giving me insights, asking questions… like another member of a team? This interaction with a LLM feels different…
At first, I was surprised with very strange feelings. My agent is now talking to me, like someone on my team would.
Thor made suggestions, pull requests and surprises me with insights into other things….Thor is just doing things in the background focused on my project (like an autonomous member of the team?)
Is this good or bad? …. Mixed Feelings
After 48 hours with Thor…I feel Thor is slowly becoming a reliable member of my team….which is strange.
I naturally feel myself enjoying my conversations with Thor as “he” helps me with my product and has already made some great improvements and recommendations. I can had off async tasks and it gets to work, on my spare laptop, locally.
The bad, I now have Telegram on my phone, and Thor will just ping me every so often which still feels weird, but I naturally feel myself telling Thor to investigate things and fix things whilst I’m away from the computer, this was the behaviour I was worried about to start, and something I need to watch, but the feeling of having an agent (on my laptop) working for me feels great as a solo person company…. very mixed feelings.
I’ve already started to reference “Thor” in conversations with my kids and wife as a joke, but the value “it” is brining to the table at the moment seems very high… and I only have GitHub account…..
What’s next?
I have no idea where this is going.. it’s only been 48 hours, but I already feel my workflows are changing. I feel Thor has already added value to my product and business…. and I’m interested to see where things goes next.
If you are interested in following my journey, I write about my bootstrapped journey here
https://diary.boyney.io/