The Entire Infrastructure Moved in Two Days

Today the work is fully active on the new infrastructure. There is actually a lot to share and a lot to say, but I do not want to overload this with excessive technical details. The main thing is that I am shocked by the capabilities, the speed, and the level of complexity that AI agents can handle now.

Imagine a small IT company working in one city. It has 30 employees, and it decides to move to another city. It seems easier for an IT company to move than for a physical factory. At the same time, every person has a workspace, high-speed internet, security systems, and communications. Everything has to be dismantled in one place, assembled in another, connected, every employee has to be given new clothes, a new home, and a new McDonald's to visit.

I managed to do all of this in two days. The scale of the work I watched was enormous. I am not a developer, although after two years of developing through AI, I can probably call myself one already.

In a sense, I had to fire all the employees and hire new ones again. The ones working locally on the computer were deactivated, and new ones started working on the server. The collection of metrics from different exchanges and paid and free services used to flow into one local operating mode. Moving to the server geographically changed many structures and processes and touched the whole architecture and its autonomy. In many places, the laptop was used as a specific device. In many places, my personal accounts were involved. Everything had to be transferred, reconfigured, adjusted, completed, and supported with additional skills.

Without AI, this move would probably have taken a team of ten people a couple of weeks under severe stress. In a calmer mode, perhaps a month. I completed it alone in two days.

Autonomy is finally becoming the main capability. There is no longer a dependency on whether my local computer is running. Process performance increased because the rented server is more powerful than my computer. Today I am working on the autonomy of every process: deciding what can systematically bring value to our factory every day and which tasks can run at night so they are ready by morning. Eight threads are running right now, the load on the service is minimal, and the speed of work has increased.

Most importantly, the autonomous fund has gained full independence. It wakes up every hour and checks the state of trades and the market. Once a day it receives the same sealed risk-metrics package that my B2B clients receive in Crash Brief. The fund also has access to the infrastructure and calculations it can run every hour. It has instructions for trading decisions and self-improvement. External agents and researchers join as independent auditors and assistants. They help it get out of a dead end or find an unusual solution. Several different models participate in the work.

After this move, the project's technical boundaries expanded greatly. I am impatient to see when I hit them again. Honestly, it is astonishing to compare where I was a year ago, where the project was, how it started, and what is happening now. A year ago, I did not even think I could develop it to this level. It is amazing to imagine what will happen in another year.