You don’t have to be an engineer to think like one

The text explores the challenges of integrating AI into internal workflows, emphasizing that the main issue often lies not with the AI models or prompts, but with users' lack of precise definitions of good work and clear processes. It draws parallels between AI work and engineering practices, suggesting that non-engineers can benefit from adopting engineering habits like thorough documentation and context clarity. Engineers typically prepare their systems to be easily understood by others, a practice that is crucial when working with AI, which lacks memory and institutional context. The article recommends treating AI prompts, documents, and workflows as executable code to maintain and update them rigorously, ensuring they are as reliable as engineering systems. It highlights the importance of defining success criteria upfront to guide AI outputs effectively, and the necessity for organizations to make implicit knowledge explicit for better AI integration. The piece argues that the teams succeeding with AI will have the most explicit processes, not necessarily the most tools or budgets, and encourages adopting an engineering mindset to enhance AI-driven workflows.