Glitter AI 对比 Lettria

Glitter AI turns screen recordings, videos, and live walkthroughs into polished step-by-step guides and SOPs. Record a process with the desktop app or browser extension, narrate as you go, and the platform captures screenshots, transcribes your speech, and writes the documentation for you.

You can also upload existing videos, including Zoom recordings, mobile screen captures, or footage of physical tasks, and Glitter converts them into editable guides. The guide editor lets you polish steps, screenshots, and copy before sharing via link, PDF, HTML, Markdown, or embed.

It supports 99 languages for recording and one-click translation. Teams at companies like Google, Walmart, and Salesforce use it for employee onboarding, IT runbooks, customer help articles, and compliance SOPs. The platform is SOC 2 Type II certified.

Lettria 构建了一个图上下文层，将企业文档和数据库转化为本体支持的知识图谱，供自动化团队使用。其 Knowledge Studio GraphRAG 产品解析 PDF、表格、SAP 导出文件和扫描附件，然后通过带有来源引用的多跳问题回答。Perseus 是一个开发平台，用于生成本体和图谱，无需手动建模每个实体。

向量 RAG 工具检索相似片段；Lettria 则遵循结构化和非结构化数据之间的明确关系。公司发布的基准比较显示，图代理的整体准确率为 81.7%，而向量代理为 57.5%，此外文本转图的准确率提高了 30%，提取速度比通用大语言模型快最多 400 倍。该权衡更适合需要审计轨迹而非快速语义搜索的受监管团队。

金融、医疗、法律和工程团队使用 Lettria 处理 ESG 报告、生物医学文献、合同语料库和技术手册。案例研究引用了 Alfa Laval（提取准确率提升 30%）、AP-HP（研究速度提升 60%）以及 Wisecube 将 108 GB 生物医学文本处理成 500 多个监控类别。