AI can handle the mechanical parts of code review, like style, naming, and null checks. Build a lightweight PR review bot for .NET and keep humans in the loop.
LLM output is non-deterministic, so traditional asserts break. Test AI features in .NET with mocks, snapshot testing, and evaluation frameworks instead.
Running AI locally in Open WebUI or OpenCode? Weigh model size against your hardware, quantization, compatibility, and benchmarks when choosing a model.