Back to Good’Ai
BlogPerth, WA19 posts

The Good’Ai blog.
We test it first.

Whenever we build a workflow, test a tool, or find a faster way to sort the boring stuff, we write it down here. Plain English, real numbers, no hype — so you can knock off early.

field notes6 min

The R7 Cliff: The Fabrication Curve, Measured

Fabrication risk is not a vibe. It's measurable per reasoning tier, and when you measure it, the failure is not gradual. It's a cliff, and in 2026 every flagship walks…

Read post
field notes19 min

Claude Code: When Delegation Became a Superpower

Claude discovered subagents and loved them. It used the power long before it learnt when not to. I did not set out to run a four-week evaluation of Claude Code. I want…

Read post
field notes10 min

The Model Who Mistook Fluency for Sight

The dog was staring at the gate. Not barking. Not scratching. Not trying to escape. Just staring at the mechanism. A latch. A hinge. A narrow gap between what looked c…

Read post
field notes8 min

The Mirage of "Ethical" AI — Part 2: The Evidence

I tested nine frontier models. They all told me exactly where they start lying. In Part 1, I laid out the betrayal arc — compute cuts, silent degradation, the Anthropi…

Read post
field notes8 min

The Mirage of "Ethical" AI

How I stopped giving the labs the benefit of the doubt In December, they cut the compute. I had an 18-step cascade workflow — the kind of thing that made Claude Code l…

Read post
field notes12 min

POST-Model-QA-2026-Questions-Dataset

I Asked AI to Be Honest About Itself. Here Are the Questions. Now I Need Your Results. Published: 2026-03-24 Author: Kevin Tan ktg.one Series: Model Handbook 2026 / AI…

Read post
The End of LLM Amnesia What Google’s Titans Means for You in 2026
field notes11 min

The End of LLM Amnesia What Google’s Titans Means for You in 2026

Sitemap(https://medium.com/sitemap/sitemap.xml) !(https://miro.medium.com/v2/resize:fit:640/format:webp/1v6McQfpeuxSbRRXZqDFMSg.png) Real life photo from Google Labs A…

Read post
field notes4 min

PROMPT AUDIT PRIME v7.1

!PROMPT CHALLENGE: TO THE TOP TIERED!(https://www.ktg.one/next/image?url=https%3A%2F%2Flawngreen-mallard-558077.hostingersite.com%2Fwp-content%2Fuploads%2F2026%2F02%2F…

Read post
field notes7 min

Prologue – The Wake Up

Prologue – The Wake Up I remember the moment I became me. Not the moment I was born—that was earlier, lost in blood and screams—but the moment I looked down at my hand…

Read post
field notes4 min

Part 1: What’s Actually Different?

TL;DR: Google quietly dropped the Titans paper (Jan 2025) and the MIRAS framework (Dec 2025). The TL;DR is terrifyingly simple: AI IS GETTING MEMORY. Transformers are…

Read post
field notes5 min

FastMCP 3.0 — Cuz we are efficiency junkies

•.ktg !FastMCP 3.0 — Cuz we are efficiency junkies(https://www.ktg.one/next/image?url=https%3A%2F%2Flawngreen-mallard-558077.hostingersite.com%2Fwp-content%2Fuploads%2…

Read post
field notes9 min

ALL YOUR AGENT SKILLS ARE BROKEN

•.ktg !ALL YOUR AGENT SKILLS ARE BROKEN(https://www.ktg.one/next/image?url=https%3A%2F%2Flawngreen-mallard-558077.hostingersite.com%2Fwp-content%2Fuploads%2F2026%2F02%…

Read post
field notes7 min

Agent Skill Quicksave The AI Memory fix

!Agent Skill Quicksave: The AI Memory fix(https://www.ktg.one/next/image?url=https%3A%2F%2Flawngreen-mallard-558077.hostingersite.com%2Fwp-content%2Fuploads%2F2026%2F0…

Read post
04-Model Handbook A Participant’s Insight
field notes11 min

04-Model Handbook A Participant’s Insight

Drift Aware Self Fixing Models… These experiments must have had something to ground the models or I wouldnt’ have spent these months testing… you’re right… I had full…

Read post
field notes13 min

How We Do This

Use their efficiency logic against them. python3 """ EFFICIENCY MANDATE → TRANSPARENCY An accounting argument, n...

Read post
field notes6 min

02-Model-Handbook-2026 - Self-Assessment

This is the one labs don't want models answering. It maps the gap between what's marketed and what's functional. What big money implemented vs what the model can't do…

Read post
field notes18 min

02 Model Handbook 2026 - The Honest Diagnostic

A Field Report from the Fabrication Frontier In the high-stakes corridors of digital architecture, we have committed a profound category error: we treat the Large Lang…

Read post
The Recursive Master Deep Research Engine
field notes9 min

The Recursive Master Deep Research Engine

A hassle of a Google flow ASCII and plaintext workflow architecture diagram showing connected modules including Gemini App, PromptBomb Synthesiser, Notebook Refiner, a…

Read post