Wispy

Accepts text, documents, or web content and generates concise summaries using a combination of extractive (key sentence selection) and abstractive (neural paraphrasing) techniques. The system appears to process input through a content normalization pipeline before applying summarization models, preserving semantic meaning while reducing token count by 60-80%. Supports variable summary lengths (bullet points, paragraph, executive summary) with configurable detail levels.

Generates original written content (articles, blog posts, social media copy, emails) by accepting a topic, outline, or brief description and applying user-specified tone, style, and format templates. The system likely uses prompt engineering or fine-tuned language models to enforce stylistic consistency across generated content, with support for multiple content types and audience personas. Includes iterative refinement where users can request rewrites with different tones or emphasis.

Automatically generates quizzes, multiple-choice questions, and assessments from provided source material (documents, articles, or web content) using question-generation models that extract key concepts and create pedagogically-sound test items. The system likely analyzes content structure to identify learning objectives, then generates questions at varying difficulty levels (Bloom's taxonomy alignment) with distractors that are semantically plausible but factually incorrect. Supports multiple question types (multiple-choice, true/false, short-answer) and can generate answer keys with explanations.