Ggaetanjaminoningaetanjaminon.hashnode.dev·May 31 · 20 min readAI Ops in a Controlled EnvironmentWhy human-safe is not automation-safe Introducing AI assistance into platform engineering on critical infrastructure is a different problem from adding it to ordinary software work. This is the kind o00
Ggaetanjaminoningaetanjaminon.hashnode.dev·May 31 · 4 min readThe LLM API is stateless: what that actually meansThe most important thing to understand about how LLM APIs work, and the one most often misread. Stateless means: the model remembers nothing between calls Every request to an LLM API is self-contained00
Ggaetanjaminoningaetanjaminon.hashnode.dev·May 31 · 7 min readWorking with AI: The 4D Framework for Effective CollaborationCore idea The 4D framework names what humans actually do when working well with AI. It describes the human side of the collaboration: four competencies (Delegation, Description, Discernment, Diligence00
Ggaetanjaminoningaetanjaminon.hashnode.dev·May 31 · 10 min readWorking with LLMs: Capabilities, Limitations, and the Human RoleCore idea LLMs have a small number of fundamental properties. Each property is both a capability and a limitation: the same trait that makes the model useful is what makes it fail. Working well with A00
