The question that killed most of his ideas, and how he found one the next model can’t replace.

Building on AI is easier than ever, which also means the next model can copy what you built in an afternoon.

We asked Magic Safari founder and CEO Philip Joubert, who ran OfferZen for about ten years, how he chose an idea that would survive…

When Philip planned his next company, he set three rules: keep the team tiny, build something that wasn’t possible five years ago, and build something that doesn’t get killed when the next Claude or ChatGPT model comes out. The third rule caused him the most worry, because he believes the next models will be very good.

“I could imagine a next model where you need a pretty thin wrapper to build the entire company.”

Start from the position that the next release will be far more capable than today’s. Philip describes himself as pretty AI-pilled, and that assumption killed many of his ideas early. It’s better to lose an idea on paper than after a year of building it.

For each idea, picture the next model with a simple app on top. If that combination does what your startup does, the idea won’t last. Keep the ideas where the model alone isn’t enough, no matter how good it gets.

Philip landed on luxury safaris, a global industry worth around R320 billion that he knew nothing about. In his words, there’s not a single book ever written about how the industry actually works behind the scenes. The models have nothing to train on, which is why he feels safe. The back end runs on emails, calls and huge human effort, even though the average trip costs around R500,000.

AI has reset every industry, so what used to be a whole product is now often just a feature. Philip says Magic Safari is really three or four startups in one, because you can’t plan a safari without every part of the process working. A narrow tool is easy for a model to absorb. A whole business with many moving parts is much harder to replace.

Philip found the safari industry by helping a company out, at his wife’s suggestion. He ended up in a room full of people who knew the industry inside out, showing him how it worked and what the numbers looked like. He calls consulting an underrated way to find real problems in industries you’d never see from the outside.

An idea that passes this test can use AI heavily without being replaced by it. That’s also why Philip doesn’t call Magic Safari an AI company: in his words, it’s a safari company that uses a ton of AI.

Philip’s full keynote from The Open Letter Cape Town 2026 is available to members inside the Founder Collab, where he goes further than we could cover here:

You’ll also get access to 40+ masterclasses from SA founders and operators on sales, fundraising, UX, paid media and more inside The Founder Collab.

Get more SA tech and business news, tips and business-building workflows, and subscribe to The Open Letter.