Experiment: Turn Papers into Self-contained Interactive Posters
I have been experimenting with Claude to turn some of our recent papers into interactive “poster pages”: single, self-contained HTML pages that summarize a paper, explain it in plain language, and let a reader play with a live in-browser demo of its core idea.
Everything runs in the browser as a single file, with no build step and no external libraries. The page is a fully self-contained HTML file. I packaged the workflow as a Claude skill so making new posters are always consistent with posters that were made before. It does require a bit of tweaking and a lot of validation - for this type of deep content LLMs sometimes oversimplify things or add information that is incorrect.
The posters so far
- BeatriXR: Comprehensive and Adaptive Feedforward Support for Guidance in Virtual Reality (EICS 2026, paper)
- RaDIA: Visual Analytics for Systematic Sample Association in Nuclear Weapon Test Verification Workflows (EICS 2026 Late-Breaking Work, paper)
- From Embeddings to Exploration (LAPEX): Interactive Latent Space Visualizations for AI Model Sensemaking (EICS 2026, paper)
- DIVERSE: Disagreement-Inducing Vector Evolution for Rashomon Set Exploration (ICLR 2026, paper)
Each poster page links back to the full paper, so it complements the publication rather than replacing it. The goal is to make our work approachable for a broader audience while keeping the interactive demo honest about what the paper actually shows. Since many ACM papers use the Creative Commons BY or BY-SA licenes, which allow remixing, adapting and extending the material from a paper, this makes a perfect fit to “remix” papers into interactive posters.