Nooruddin Hargey

Built an app with AI tools, a freelancer or a team that moved on? Still at the idea stage? Three fixed-price services from experienced developers: • Review My App: R4,950 • Plan My App: R9,950 • Finish My App: from R19,950 Pick your starting point: puzzl.co.za/for-app-creators…

Type three letters into the search box on a 404 page. The whole page reloads. Type another. It reloads again. The search box wasn't broken. It was asking the wrong place for its data.

Got an app idea? Or an AI-built app stuck at 80%? Get it reviewed, planned or across the line by real human developers. Clear scope, upfront pricing. #VibeCoding #Puzzl

Pick your starting point: puzzl.co.za/for-app-creators…

Two quotes for the same app: $40,000 and $220,000. Both can be correct. The features you list are the tip of the iceberg. The cost lives underneath.

Incorrect municipal bills. Medical aid claims that don't pay out. Contractors who never show up. Every South African has a frustration they think an app could fix. Fewer would pay for one. We looked at eight industries to find where a small software business can actually win.

By handover: 100+ releases, 400+ test files and a written engineering guide. "A good agency doesn't make the client fit its favourite tools." Full case study: puzzl.co.za/projects/case-st…

Three software companies. Three quotes. Three different numbers. Nobody has to be overcharging. They may each be pricing a different interpretation of the same brief. New article: what drives custom software costs, and how to compare quotes properly.

Technical debt isn't necessarily bad engineering. Sometimes a shortcut is the right business decision. The problem is when nobody tracks it, revisits it, or considers what it will cost later. And as AI makes software development faster, managing that debt becomes even more important.

Most build vs buy conversations end the same way: "it's strategic, so we build." That's not a decision. It's a placeholder for one. Here's a better test: if every competitor had access to the exact same software tomorrow, would you lose anything that actually matters? We wrote up the full framework, including how AI-assisted development actually changes (and doesn't change) the calculation.

AI helped create the duplication. AI helped clean it up. A human had to spot the pattern first. Our first Puzzl case study in 15 months looks at what happened when AI-generated implementations of the same breadcrumb component started diverging across pages.

On the day-to-day, one tries to avoid using any fancy Unicode characters in source code as far as possible, but you still need to account for them occurring elsewhere. Today's video illustrates that "how long is this string" depends on who you ask. Triple points if you got all three answers correct!

The interesting question was never whether AI can write code. It's what happens to everything else once it does. A randomized controlled trial from METR had experienced developers complete real tasks with and without AI tools. They took about 19% longer to finish, while believing they'd been roughly 20% faster. Implementation gets faster. Requirements, review, testing, and security don't automatically speed up with it. The bottleneck just moves. We wrote up what this actually means for teams building software right now. Not the hype version. The version backed by METR, Google's DORA research, Faros AI, Veracode, and CodeRabbit data.

Apple has a pattern: ship a feature that makes everyone ask "why did they even do that?" 😀 A few years later, the answer usually becomes obvious. We looked at some of Apple's most criticised features and the pattern behind them. It's not about whether Apple always gets it right. It's why decisions that look pointless at launch can make a lot more sense once you see the bigger strategy.

Let's start an argument: Flutter or React Native, which should startups actually choose? 😀 There's no "this framework is objectively better" answer. It depends on what you're building, what your team knows, and where the product needs to go. We looked at both from a startup lens: dev speed, performance, ecosystem, maintainability, and the trade-offs that actually matter when you're trying to ship something people will use. 👉 And yes, we have an opinion.