One pull request, three AI reviewers: how Claude, ChatGPT, and Gemini review the same code

Aido, the open-source multi-provider AI assistant for GitHub code review, has published a public demonstration of the idea at the center of the tool: that no single AI model catches everything, and that running more than one, each covering the others' blind spots, produces a more holistic review.

The demo is a single pull request on Aido's own repository. It changes one file, a small user-service module seeded with a set of deliberate, classic mistakes: a hardcoded credential, a SQL injection, missing null and error handling, weak randomness in a token, and a money-transfer endpoint that is missing both an authorization check and an await. It is published as a demonstration and is not meant to be merged.

The same diff, three models

Aido reviews that same change three times, once each with Claude, ChatGPT, and Gemini, using the same personas and instructions and differing only in the provider. The three reviews sit side by side on the pull request, so anyone can scroll through and see where the models reach the same conclusion and where they diverge.

Why more than one reviewer

Running several models is not about redundancy. Each tends to have its own strengths and blind spots, and a finding one model misses another often catches. For a developer using AI to write code, especially someone moving fast, a review from a single model can feel like a green light while a real issue slips through. A second and third opinion, at the cost of an extra API call, narrows that gap.

Aido is built for exactly this. It is multi-provider by design, and a team can point it at Gemini, ChatGPT, Claude, or any OpenAI-compatible model on its own keys. The demonstration is open on GitHub for anyone to read.