AI Academy Episode 3
The repository as memory and a line of control
In productionThis episode is in production. Leave your e-mail address below and we will let you know when it is out.
Why an AI assistant's memory should live in files you control, and how the same repository sets the boundary an agent does not cross without your consent. The pattern in four layers, the implementation as of November 2026, a real incident and the safeguard it led to, and an exercise for applying the pattern to your own tools.
New episodes and the pitfall catalogue
Leave your e-mail address and we will let you know when a new episode is published. The catalogue of 17 pitfalls (PDF) will reach you as soon as it is ready.
BluePoint
Bringing this into your organisation
Workshops for leadership and delivery teams: the same patterns applied to your own processes and tools, documented in line with the AI-literacy requirement of the EU AI Act.
In some videos the presenter is an AI avatar with a cloned voice, used with Patryk's consent. The content, the experience and the mistakes are his own.