Critique Agent

This agent acts as an autonomous reviewer that simulates audience reception of generated presentation content. It evaluates each slide’s tone, structure, density, and language clarity using LLM-based quality assessment, then suggests improvements and drives iterative refinements through a reflective feedback loop. Operating within that loop, it decides whether to approve or refine content, showing self-assessment and goal-checking behavior.

The agent is built for professionals who need to turn raw input such as text, URLs, and documents into polished, audience-tailored slide decks. It reduces the need for human feedback loops and peer reviews, improves presentation quality and clarity for stakeholder meetings, and lowers the risk of errors or misaligned messaging. It uses Vertex AI LLMs, including Gemini Pro and Text-Bison, for quality review and content refinement.