Replacing a car engine is hands-on work no AI agent can touch.

Every engine replacement varies by make, model, year, and the specific failure mode. The task requires adaptive physical judgment at each step, making it structurally unique every time.

The stated symptom — overheating — doesn't necessarily mean the engine needs replacement; it could be a thermostat, coolant leak, or head gasket. Success criteria are undefined and require physical diagnosis to even establish.

An AI agent has no access to the physical vehicle, no ability to use wrenches or hoists, and cannot inspect components. There are no APIs or digital tools that bridge this gap.

A botched engine replacement can destroy a vehicle worth thousands of dollars, create safety hazards, or cause injury. Errors are expensive and largely irreversible.

Diagnosing the root cause, deciding whether replacement is warranted, and executing the physical work all require expert mechanical judgment and hands-on skill that AI cannot replicate.