Imagine getting instant design feedback on your charts, not from a human expert, but from an AI. That’s the promise of Visualizationary, a new tool that uses Large Language Models (LLMs), the same technology behind ChatGPT, to help people create better data visualizations. Creating effective charts isn’t just about plotting data—it’s a design challenge. Picking the right chart type, color scheme, and layout can make the difference between confusion and clarity. Visualizationary tackles this by scanning uploaded charts and using perceptual filters to analyze elements like visual hierarchy, color contrast, and text placement. Then, it translates these technical findings into plain-English feedback anyone can understand. The system analyzes the chart image using several perceptual metrics, clarifies the interpretations in natural language for easy comprehension by the visualization designer, guides designers by offering natural language suggestions to address any issues found during the analysis and tracking component shows the progress of the design over time, offering a historical record of alterations and upgrades implemented. The tool then provides design knowledge to the LLM to guide the designer with practical advice on how to improve their visualization and also keeps track of the design revisions over time. In a user study with novice, intermediate, and expert visualization designers, participants created visualizations from scratch and used Visualizationary for feedback over several days. They generally found the AI’s advice helpful, especially for catching details they might have missed. While the AI sometimes struggled to provide very specific, actionable advice, its role as a design assistant was generally well-received, even making the design process feel less isolating. Visualizationary offers a glimpse into the future of design tools where AI can boost the creative process with personalized guidance. While challenges like ensuring data privacy and handling potential AI “hallucinations” (generation of nonsensical or unfaithful content) remain, it’s a promising step toward democratizing good design.