Dialor

From Debate to Clarity.

Most AI tools summarize documents for you. Dialor debates them with you — question any passage and a cast of AI agents answers in role, building real comprehension instead of surface familiarity.

Every annotation is saved as a plain file in a folder you own — your library compounds into a personal knowledge base, and Dialor surfaces connections across documents as you read.

You ask. They argue back.

Highlight any passage, drop a question, and Dialor routes it to the right specialist. You're not getting one flattened summary — you're getting a mode of thinking, in role.

Clarify

Define terms, paraphrase, deconstruct methods.

Connect

Surface parallel passages and prior notes.

Challenge

Probe inconsistencies, premises, or argument strength.

Consolidate

Test recall with synthesis questions.

Challenge runs deeper — three specialists who disagree differently. Dialor picks who answers, or you choose:

ContradictorInternal inconsistencies across the document.
PhilosopherUnstated premises and normative commitments.
Structural CriticHidden premises and argument strength.

Your notes are plain files. In a folder you own.

Dialor never locks your thinking inside an app. Every document you open gets a folder on your own disk — open it in Obsidian, grep it, commit it to git, or feed it to your own scripts. It's text. It's yours, and it outlives Dialor.

Plain Markdown, everywhere: Source text and exported note threads are .md — open them in Obsidian or any editor.
No database, no export button: The files are the format. Nothing to extract or migrate.
Local-first: Your reading stays on your machine.

my-workspace/
  my-paper.pdf/
    source.md          ← clean Markdown
    annotations.jsonl  ← highlights + AI
    figures/           ← cropped images
  notes/
    my-paper--….md     ← Obsidian-ready