A person sitting at a desk with multiple code-filled monitors, head in hands, looking toward a bright forest path glowing through a window.

AI fatigue is real

Running agents instead of writing code felt like 10x productivity. Then I broke, and needed 4-6 weeks to recover.

June 30, 2026 · 2 min · 306 words · Agoston Fung
Four panels showing an evolution from a terminal window, to a file browser, to a web page, to a glowing AI robot connected to chat, database, analytics, and integration icons.

GUI is dead, long live agents

Command line, thick clients, the web, and now agents. The interface itself is starting to look optional.

June 15, 2026 · 1 min · 136 words · Agoston Fung
A crumbling wall labelled legacy code, outdated stack, and slow delivery on one side, with a small robot carrying a code icon flying through into a bright modern laptop dashboard on the other side.

AI can rewrite your legacy app in hours

Big-bang rewrites used to be a bad idea for good reasons. AI agents just changed the math.

May 27, 2026 · 1 min · 164 words · Agoston Fung
A person at a desk holding their head, with a cloud of thoughts above showing security warnings, a glowing brain, a legal scale, a dollar sign, and a document.

The unexpected side effect of LLMs: the burden of knowledge

AI is supposed to make us more productive. Once you know something, though, there’s no going back - and that has a cost of its own.

May 20, 2026 · 2 min · 237 words · Agoston Fung

I vibe-coded a Sudoku app and learned that product taste doesn't come free

AI can write all the code. It still can’t magically create a great product - the taste has to come from somewhere else.

May 12, 2026 · 2 min · 262 words · Agoston Fung
Diagram titled 'Safe. Powerful. Direct.' showing an AI agent routed through guardrails (dedicated role, read-only transactions, connection pool, query timeouts) into a read-only schema and read replica, separate from the primary production database.

Seven strategies for safe AI agent database access

Unrestricted database access for AI agents is dangerous. Excessive restrictions defeat the purpose. Here’s how to find the middle ground.

May 6, 2026 · 1 min · 182 words · Agoston Fung

Being a software engineer is weird when it comes to family

My team spent years automating £50 million in promotional income. A three-day Sudoku app got me a standing ovation at dinner.

April 30, 2026 · 2 min · 231 words · Agoston Fung
A pendulum swinging back from 'AI writes the code' toward 'Humans write the code', under the headline 'The AI honeymoon is over - the pendulum is swinging back'.

The AI honeymoon is over

The real cost of AI writing code is starting to show - what the data is telling us about the human/AI balance in software work.

April 25, 2026 · 1 min · 131 words · Agoston Fung

How can an LLM write code?

It can feel like AI is thinking. What’s actually happening is closer to pattern recognition - and that frame makes LLMs much easier to reason about.

March 29, 2026 · 1 min · 197 words · Agoston Fung
A wide-eyed programmer in bed at 3:17 AM, sweating and clutching a phone showing a fake 'CLAUDE AI $9000/MO' notification, while a menacing cartoon AI looms in the dark room behind him with binary code and dollar signs.

Claude separation anxiety

A small clinical case study of a new programming-related condition. Symptoms may include reviewing code like a museum visitor observing ancient artifacts.

March 26, 2026 · 1 min · 133 words · Agoston Fung