OmniBox (小黑 OmniBox) collects, parses and answers questions across your own material. A browser extension snapshots webpages, blogs and tweets into editable documents; PDFs, Word files, PowerPoint, Markdown, audio and video are parsed into the same searchable form; and the question-answering layer decides what to do with a query and traces its answers back to the documents they came from. The company is Hangzhou Xiaohei Intelligent Technology Co., Ltd.
It is open source, with the code on GitHub, and that is the reason to look at it before its closed competitors. A personal knowledge base is the single worst category in which to be locked into a vendor – the value compounds over years and evaporates entirely if the company folds or changes terms – and source-available code with self-hosting means the archive outlives the business. Snapshotting rather than bookmarking is the other correct decision: a bookmark to a page that has since changed or disappeared preserves nothing, which is most of what a years-old reading list turns out to be.
It is early, and the roadmap says so – a subscription tier, a tag system and mobile apps are all listed as coming, which means today’s version is browser and desktop with no way to capture from a phone, where a lot of reading happens. Documentation and interface lean Chinese-first, which is a real friction for English-speaking users even where translation exists. No rates are published yet.
