MemSlides
MemSlides is a hierarchical, memory-driven agent framework for personalized presentation slides that separates persistent preferences from working memory and tool experience, editing only the smallest affected slide region during a revision instead of the whole deck.
MemSlides treats presentation generation as a stateful authoring process instead of a one-shot conversion from source material to slides. It separates personalization signals by how long they last: persistent user-profile memory captures preferences that recur across jobs, working memory carries active constraints across revision rounds within one deck, and tool memory stores reusable execution experience for reliable localized editing.
Long-term user-profile memory routes personalization by presentation intent to theme, visual style, layout and content-strategy choices at the start of a job, while during revision the system projects a user's feedback onto the smallest affected slide region and applies a scoped local patch instead of regenerating the full deck. A public trial is available, and the project asks users to check that they are on its official site rather than an unaffiliated look-alike domain.