
Natalya Rojkova · Finance and capital projects · Sydney
20+ years in finance and construction. Building with AI agents every day and writing about it for people without a technical background.
After six in the evening I hand five agents the work that can be written down. In the morning I review what they did.
None of it is code. It is delegation, and the same rules apply as with people.
Research agent
Gathers and checks the sources for the piece I am writing.
Content maker
Turns my notes into a first draft.
Inbox agent
Drafts replies to my email.
Book editor
Proofreads the book, one chapter at a time.
LinkedIn manager
Prepares my LinkedIn posts and queues them for the morning.

Agents & delegation, 6 min read
An AI agent is software you give a goal, not instructions. It plans the steps itself, uses tools like a browser, files, and email, checks its own work, and comes back with a result. A chatbot answers you. An agent works for you.

Agents & delegation, 5 min read
A chatbot answers questions one message at a time and waits for you. An AI assistant works alongside you, step by step, while you drive. An AI agent takes the whole task, works through it on its own, and comes back with a result. The real difference is not intelligence. It is who does the thinking between the steps.

Agents & delegation, 5 min read
AI agents fail in the first week for management reasons, not technical ones: a vague goal, no boundaries, no definition of what a good result looks like, and no review loop. The model is almost never the problem. The 10-minute fix is rewriting the brief, five lines, and cutting the task's scope in half. Here is exactly how.
A finance leader who builds with AI agents after six in the evening, and writes about it for people without a technical background.