Robbert van Empel explains AI agents through a morning routine

Robbert van Empel has published an eight-minute video that explains AI agents through an ordinary morning at home. Released on September 6, 2026, the film avoids a technical product tour. Instead, it follows a small set of agents while Van Empel makes coffee, showing how automated work can happen in the background without taking over the rhythm of the day.

The central distinction is simple. A chatbot waits for another question, while an agent can continue through the steps of an assigned task until it returns a result. Van Empel shows his agents starting when he unlocks and wakes his laptop. He then leaves the computer and prepares coffee by hand. The automation is presented as support for a routine, not as a replacement for the parts of that routine he values.

In the example, one agent collects AI news relevant to his work, another organizes tasks, and another checks the inbox for messages that need attention. The results appear in one overview alongside a calendar appointment. A short summary comes first, so the user can see what changed, what needs a decision and what can wait. Each item can still open the original email, article, document or calendar entry. That keeps source context available and leaves the final choice with the person using the system.

The video also shows an agent reviewing the reach and responses to Van Empel’s latest LinkedIn post. It saves one lesson that he can review, correct or remove before that context informs later work. This is an important boundary: the agent can organize feedback, but the user remains responsible for deciding whether the saved interpretation is useful. The film does not provide benchmark results or claim that every agent will work reliably without supervision.

For newcomers, Van Empel recommends starting with one repeated task, connecting only the information it needs, describing the desired result, running it once and improving one part after checking the outcome. Possible starting points include a morning news summary, turning email into a task list or preparing for the workday. For AI users and teams, the practical message is that a useful agent does not need to automate an entire job. A narrow workflow, a clear output and a visible review point can be enough to return time while preserving human judgment.