Tolics Engineering Mind

In hybrid AI organisations, developers aren't above the stack — they're stuck in between it, relaying intent like middle managers in a bloated org.

The automatic transmission didn't kill the gearbox — it abstracted it. The same thing is happening to code. The work didn't vanish. It moved up to intent.

Neovim is a maze, Helix flipped the grammar, Zed left the terminal. So here's the editor I want: a tiny vi-like core with a UI you script entirely in Lua.

The agentic AI loop wasn't born in 2024. It was shipped from a Melbourne lab in 1991 — the BDI architecture — inspired by a Stanford philosopher's 1987 book.

Fifteen core AI concepts — from machine learning to agentic AI — each defined in a single sentence, ordered to build on each other into a mental model.

A Malmö carpark, a Melbourne blackout, and the cheerful realisation that systems — not luck or geography — quietly decide what does and doesn't get built.

SUMO tried to encode all of reality into first-order logic. Its directory listing is a philosophical argument most AI engineers have never encountered.

The real bottleneck in software development is not knowledge — it is processing. Ambiguity, cognitive load, and feedback latency are the upstream causes.

Dark factories are coming to software. The shift from artifact-centric correctness to scenario-driven validation changes what you build and maintain.

A 2025 study found Polish outperforms English in long-context LLM tasks. Ukrainian grammar instincts explain why — and how anyone can prompt better.

How RLHF trains models to seek clarification instead of guessing — and a four-agent pipeline that brings the same discipline to your own requirements.

Most teams have no idea if their AI workflow investments are paying off. Here's what to actually measure instead of relying on vibes and excitement.