MASTERMIND, Three Years On: The Orchestrator Became the Pattern
MASTERMIND described a controller over reasoning modules and BDI agents. That shape became my architecture; its hardest lesson was about repetition.
aGLM, Three Years On: Memory Worked, Autonomous Learning Did Not Yet
aGLM promised a model that learns from its own memory. The memory works; the learning regressed after generation 39, and I can show the ledger.
RAGE, Then Until Now: From GATERAGE to mindX in One Timeline
From GATERAGE in 2023 to mindX in 2026: one timeline of RAGE, dated from the posts themselves, with what shipped, what stalled, and what is still open.