Language models have transformed coding by automating implementation, making tasks like building REST APIs or React components faster. However, the article argues that coding is only partially solved. While AI excels at translating clear specifications into code, organizations face deeper challenges. Enterprises operate with complex contexts: existing architectures, security requirements, business constraints, and historical decisions. These factors shift the bottleneck from writing code to making strategic decisions.

Software engineering involves multiple layers: business objectives, product design, solution design, and implementation. AI is now automating the implementation layer, but higher-level tasks remain critical. For example, determining which team owns a functionality or how it integrates with legacy systems requires organizational knowledge. This context is absent in hobby projects but essential in mature organizations.

The article emphasizes that AI’s true value lies in revealing the complexity above coding. As implementation becomes cheaper, engineers must focus on aligning solutions with business goals, regulatory requirements, and user expectations. The same model that simplifies weekend projects struggles with enterprise-scale decisions. Organizational context—unwritten conventions, security policies, and technical debt—creates barriers that AI cannot overcome without explicit guidance.

Thus, while coding may be automated, software engineering as a discipline remains multifaceted. The real progress is in recognizing that code is just one layer of a broader problem-solving process.