Knowlume Documentation

Knowlume is a knowledge filtering and distillation system. It turns long sources into compact, inspectable knowledge fragments that a person can filter, save, and export.

It is not just a summarizer. A summary compresses the whole source, including weak or generic parts. Knowlume helps you find the parts that are worth keeping: reproducible methods, non-obvious insights, useful cases, source references, and fragments that can be reused later.

Use Knowlume when you need to:

  • process articles, websites, YouTube videos, files, or pasted text;
  • split material into meaningful fragments;
  • score fragments by usefulness signals;
  • select only the fragments worth keeping;
  • export the result to Markdown or automate the flow through APIs.

Production app:

https://app.knowlu.me

Start Here

What Knowlume Produces

The central unit is a fragment: a short, source-backed unit of knowledge. A fragment can be a claim, method, observation, instruction, fact, or synthesis extracted from a larger source.

Each fragment has:

  • a title;
  • source context;
  • Markdown content;
  • scores for reproducibility, originality, and sourcesness;
  • optional rationale explaining why the scores were assigned.

The point is not to summarize everything. The point is to separate signal from noise, then make useful knowledge easy to inspect and reuse.

Reader Paths

If you are evaluating Knowlume, read Quickstart, then Core Concepts.

If you process videos, podcasts, lectures, articles, or news sources, read Use Cases after Quickstart.

If you regularly build a knowledge base, process saved posts, or review conference material, read Best Practices.

If you are building an integration, start with Integrations And API, then read Fragments, Scores, And Export so your client handles source status and export timing correctly.

If something does not work as expected, use Troubleshooting before changing your source or export settings.

Status

This documentation describes the user-facing Knowlume SaaS workflow as of May 28, 2026. It is intentionally small, heavily linked, and optimized for scanning.

Source note: the product framing and filtering examples were enriched from the public Knowlume presentation at https://www.youtube.com/live/qMEKcqoY-H4.