Why Are We Still Not Able to Get AI to Do Work Properly?

As AI takes on more coding and review work, the key skill is shifting from doing the work yourself to delegating it well to AI.

The article argues that simple prompt writing is no longer enough; providing context, constraints, and rules matters more.

This evolution is described as moving from prompt engineering to context engineering and then to harness engineering.

In practice, AI must be given a clear environment and execution setup, such as through tools like Claude Code, MCP servers, terminals, and IDEs.

For developers and organizations, productivity and output quality will increasingly depend on how well they design tasks for AI.