Automated Review
Review Otomatis
An agent reviewing another agent's work, often with a different model or system prompt. It forms a judgement.
An automated review is an agent reviewing another agent's work, often with a different model or a different system prompt. The reviewer forms a judgement, and that judgement is non-deterministic in the way a test is not.
The value is a fresh perspective. The agent that wrote the code cannot see what it misunderstood, because it is too close to its own reasoning. A reviewer with a clean context and an adversarial instruction catches things the implementer structurally cannot. This is why review agents work better when told to look for problems than when asked to confirm the work is fine.
The weakness is the same as its strength: the reviewer is also a model, with the same blind spots and the same tendency to be agreeable. A review that says everything is fine is not evidence, and an automated review should be calibrated against human review on a known set before you trust it to gate anything.
Related terms
Human Review
The user reading the code the agent produced and forming a judgement on it. Reading the diff counts; reading the summary does not.
Code Review
A second pass by someone who did not write the code, looking for what the author missed.
Automated Check
A deterministic verification that runs in the environment: tests, type checks, lints, build. Pass or fail, no judgement.
Evaluation
Measuring whether the system does the job, on a fixed set of cases, repeatedly.
LLM Judge
Using a model to score another model's output against a rubric.
Want this working inside your team?
It starts with one conversation about your workflow.