Shopify CEO Tobi Lütke recently shared how he used AI to create software for an incredibly niche, but important, situation.
His annual MRI scan comes on a USB stick, but requires commercial Windows software to open. Instead of going in search of this existing software, he ran Anthropic’s Claude AI model directly on the MRI files and prompted it to build a web-based viewer.
The result, he said, looked “way better.” With one more prompt, the tool even annotated the images with the findings from the scan.
Lütke called this an example of “reflexively” reaching for AI — training yourself to instinctively use it to build bespoke tools when off-the-shelf software falls short.
“You want to train your brain on this intuition,” he wrote in a thread on X that got more than 7.5 million views.
This type of experimentation takes more upfront time, and it requires challenging your assumptions about how things should be done, according to Bernard Golden, CEO of Navica, a Silicon Valley-based technology analysis, consulting, and investment firm.
“You have to spend some brainpower to reflect on established habits to see how AI could be inserted, but doing so has a snowball effect: the more you try, the more you do,” Golden told Business Insider. “It’s like learning a language. It’s uncomfortable to try speaking when you’re starting out, but doing so accelerates your skills and confidence.”
Sign up for BI’s Tech Memo newsletter here. Reach out to me via email at abarr@businessinsider.com.
