Software · AI · Agents

I build software together with AI.

TadeCode is my one-person business. Day to day I write software working with Claude models, and alongside that I develop AI agent farms — teams of agents that work together on larger tasks.

What I do

Currently fully committed to a long-term engagement.

What I do

Code, AI and agents

For me AI is a working tool, not a slogan. It takes over the repetitive part of the work, while I stay responsible for architecture, quality and decisions.

Building software with AI

I write web and mobile apps and internal tools, working daily with Claude and Claude Code. The model writes, tests and refactors alongside me — faster, but still under control.

Integrations and automation

I connect AI models to the tools where work actually happens — documents, email, databases — via APIs and MCP, so repetitive tasks get done on their own.

Current focus

AI agent farms

Instead of a single chat — a team of specialised agents, each with its own role: analysing, writing, reviewing, reporting. An orchestrator coordinates them, and a human approves what matters. This is the direction I'm investing in most.

  • Agents built on Claude models
  • Many agents working in parallel on one large task
  • Roles: developer, reviewer, tester, analyst
  • Logs, oversight and cost control — you can see what an agent did and why
  • A human in the loop for decisions that matter

How I work

A human sets the direction, AI does the work

My everyday cycle of working with models and agents.

  1. Goal

    I describe precisely what should be built and how I'll know it works.

  2. Execution

    Agents write code, tests and documentation — often several tasks at once.

  3. Verification

    Tests, code review and checks against real data.

  4. Decision

    I approve what moves forward. Responsibility stays with the human.

Principles

What I stick to

Competence

AI speeds up the work, but I'm responsible for code quality and results.

Simplicity

Simple solutions where they're enough. Complexity only where it's needed.

Progress

I keep testing new models and tools — and keep the ones that genuinely help.

Transparency

It's always clear what an agent did and what a human did.