Shopify CEO Tobi Lütke Let AI Read His MRI, and Build the Software to Do It

Why It Matters

The approach shows how executives can leverage generative AI to create bespoke, high‑value tools faster than traditional software procurement, accelerating digital transformation and operational efficiency.

Key Takeaways

• Lütke used Claude AI to build MRI viewer
• Tool automatically annotated images with scan findings
• Demonstrates reflexive AI adoption for niche problems
• Encourages executives to prototype AI solutions quickly
• AI-driven custom tools can outperform commercial software

Pulse Analysis

The incident illustrates a shift from reliance on packaged software to on‑demand AI development. By feeding raw MRI files into Claude and issuing natural‑language prompts, Lütke produced a web‑based viewer that not only rendered the scans but also added diagnostic annotations. This rapid prototyping bypasses licensing costs, reduces vendor lock‑in, and showcases how large language models can act as low‑code developers for highly specialized workflows.

For business leaders, the lesson extends beyond medical imaging. The "reflexive" AI mindset encourages teams to question existing processes and ask whether a generative model could automate or enhance them. Companies that embed this habit gain a competitive edge: they can iterate faster, test concepts without extensive engineering resources, and cultivate a culture of continuous innovation. As AI models become more capable and accessible, the barrier between idea and implementation shrinks dramatically.

However, the approach also raises governance considerations. Custom AI‑built tools must adhere to data privacy, security, and compliance standards, especially when handling sensitive health information. Organizations should establish clear guidelines for prompt engineering, model selection, and validation to mitigate risks. When managed responsibly, the ability to generate tailored software on the fly can transform enterprise agility, turning AI from a novelty into a core productivity engine.