AI agents, automation, web3 experiments, and the occasional side project that gets out of hand.
See what I’m working on ↓( a few things here are interactive )
02 / work in progress
A mix of tools, experiments and projects that made it past the first idea. Some are live, some are paused, and some are still being figured out.
LOG 003
A custom assistant for a small law firm, designed to help with intake, document handling and early drafts while keeping a person responsible for the final work.
What I learned: The useful part is not flashy AI. It is making the next step clearer and keeping a record of what happened.
LOG 001
A tool for turning long, messy documents into information that is easier to search, check and use.
What I learned: Documents are rarely as tidy as they look. Getting the structure right is most of the job.
LOG 005
Automations that watch for routine work, organise information and help move tasks along without asking someone to do the same thing twice. Also made Hermes and OpenClaw coexist on the same rig without stepping on each other.
What I learned: Automation is helpful when it removes a decision, not when it creates another dashboard to check.
LOG 002
Experiments around on-chain data, wallets and market information. Still exploring where the useful ideas are.
What I learned: A useful tool needs a clear question to answer. More data is not automatically a better answer.
LOG 004
Small projects built for the fun of making something interactive, learning a new tool, or scratching an idea that would not leave me alone.
What I learned: The best side projects teach you something before they ask for a roadmap.
03 / paused, not forgotten
Not everything needs to be finished immediately. These are ideas worth keeping, along with the reason they stopped.
The idea
A simple way to track and make decisions about personal web3 activity.
Why it stopped
It became more complicated than the problem it was meant to solve.
What would change
If there is a smaller version that answers one question well.
Keeping a note of what did not work makes the next attempt quicker.
04 / notes from the workbench
Short thoughts from building, testing and trying to make tools behave.
AI
The best agent workflow is usually the one that gives you fewer things to review.
BUILDING
If a tool needs a guide before it is useful, the tool still needs work.
SECURITY
A small security check is much cheaper than a large clean-up.
GAMES
A good side project can be a break from work and still teach you something useful.
WEB3
Good tools make the underlying system easier to understand. They do not add more noise.
FIELD NOTE
Shipping is useful. Knowing what to change next is more useful.
05 / about
I’m retr0x. I build practical things with AI, automation and web3.
I like taking a rough idea, making it work, then finding out where it breaks. Recently that has meant experimenting with AI agents, document workflows and tools that save people time. I care about useful systems, clear outputs and learning by doing.
The work is the timeline.
Still exploring.
06 / contact
Have an idea, a problem worth looking at, or something useful to build? Send it over.
( edit links in lib/content.ts )
If it sounds useful, I’d like to hear about it.