Artificial Intelligence

Why the Best AI Tools Remove Work, Not Just Add Features

By Shandre Botha, Founder

I spent fifteen years marking, planning, and writing report comments long after my students had gone home. Somewhere in that decade and a half, I got quite good at spotting the difference between a tool that helps and a tool that just adds another thing to check.

Every year brings a fresh wave of apps promising to transform how teachers work. Dashboards. Analytics. Another login, another notification badge, another feature nobody asked for bolted onto a product that was already doing fine. I tried a lot of them. Most of them made my evenings longer, not shorter, because now I had to learn the tool on top of doing the job the tool was supposedly helping with.

That's the trap most software falls into, and it isn't unique to education. Somewhere along the way, "AI-powered" became a feature to advertise rather than a problem to solve. Companies add a chatbot to a product that never needed one, call it innovation, and ship it. The person using the product is left with more buttons, more choices, more things demanding their attention — and no more time in their day than they had before.

The only question that matters

When I started building Penny, I gave myself one rule: every feature has to delete something from a teacher's evening, not add something to their morning. Not "help you write report comments faster." Delete the task of writing them from scratch entirely. Not "organise your marking workflow." Take the marking workflow off your plate.

That sounds obvious written down. It is genuinely rare in practice. Most software is built by people optimising for engagement — more time in the app, more clicks, more screens visited — because that's what venture-backed growth metrics reward. But the people paying for these tools, teachers, small business owners, cleaning services, don't want to spend more time in an app. They want their evening back. Engagement is the wrong thing to optimise for when the actual goal is giving someone their life back.

What this looks like in practice

When we built Penny's report comment generator, the test wasn't "does this look impressive in a demo." It was "does a teacher who used to spend four hours on a Sunday now spend forty minutes, and does the output actually sound like them, not like a robot." When we scoped Xtramurals, the test wasn't "how many features can we list on a pricing page." It was "does the school admin who currently manages sign-ups on a spreadsheet and a WhatsApp group get to delete both of those things."

This is the same lens we bring to every AI project at Heavenly Glow Journeys, whether it's our own products or a client's. Before we recommend a single line of code, we ask what specifically gets removed from someone's day. If the honest answer is "nothing, it just looks clever," we don't build it — no matter how impressive the technology behind it might be.

Anyone can bolt a chatbot onto a website now. That was true five years ago too, with a different set of buzzwords attached. What's still rare, and what we think is actually worth paying for, is the judgment to know which problems are worth solving and which features are just noise dressed up as progress.