// Kedion

Built for working scientists.

Kedion is a growing suite of AI tools for researchers and educators. Discover real cross-domain connections. Structure your thinking. Stress-test your manuscripts. Share finished work as something others can learn from.

Try LatentLink free See the workflow
scroll to explore
kedion / four-tool research workflow
// 01
LatentLink
Discover
Find real connections across fields, not surface ones.
// 02
Notes2Tree
Structure
Turn rough thinking into navigable hierarchy.
// 03
Manifest
Strengthen
Stress-test your manuscript before submission.
// 04
Dekyon
Share
Turn finished work into something others can learn from.

One workflow.
Four tools.

Find connections others miss. Structure your thinking. Stress-test your work. Share what you've made — as something people can actually learn from.

From messy notes
to structured thinking.

Researchers don't think in clean outlines. They think in fragments — reading highlights, margin notes, half-formed observations, paper bookmarks, conversation snippets. Notes2Tree takes that mess and turns it into hierarchy.

Drop in rough notes, PDF excerpts, or recorded thinking. Get a tree you can navigate, restructure, attach evidence to, and ask questions of. Share it as a public dendrogram, or save it back into your own work.

The structural workspace underneath everything else. If LatentLink shows you connections, Notes2Tree is where you build with them.
Open Notes2Tree
What it does
AI from notes or uploads Turn pasted notes, PDFs, or images into a generated hierarchy without hand-building structure from scratch.
Drag to restructure Move nodes onto new parents to reorganize the tree without rewriting the underlying outline.
Ask why branches fit Select branches and ask AI to explain why those nodes belong together — not just where they were placed.
Navigation Mode Present trees in a guided mode that pauses on attachments — usable as explorable presentations or teaching scaffolds.

Stress-test before
you submit.

Peer review catches problems too late. Traditional literature search misses what's been done under different names in adjacent fields. By the time the rejection comes back, you've already invested months.

Manifest sits between you and submission. It runs four checks on your manuscript and surfaces what's weak — so you can strengthen the work before reviewers ever see it. It's not peer review. It's the diagnostic you run before peer review, on yourself.

Manifest's novelty check and cross-domain implications run on the same engine that powers LatentLink. The other rigor tools check citations. Manifest can also tell whether your novelty is real — or a relabeling of work already in adjacent literature.
Request pilot access
What it checks
Citation completeness Surfaces relevant prior work missing from your references — including work in adjacent fields you may not have searched.
Methodology stress-test Compares your study design against discipline norms for sample size, controls, and statistical method.
Real novelty check Tells whether your contribution is genuinely new at the structural level — or a relabeling of existing work.
Cross-domain implications For each finding, surfaces the other fields where it might apply. Often the most interesting result of running Manifest at all.

Turn finished work
into something others can learn from.

A finished paper, a course, a body of recorded talks — most of it sits in a PDF or a video timeline that people read once and never return to. The structure that's already inside the work is invisible.

Dekyon extracts that structure and makes it the interface. Concepts become navigable. Topics become study spaces. Linear material becomes something readers can move through by idea rather than by sequence — and come back to.

Where research becomes teachable. Where a body of work becomes a community's shared scaffold rather than a one-time read.
Pilot Dekyon
What it does
Concept extraction Identifies the ideas, patterns, and relationships already present in the source material.
Topic trees Makes structure visible so users can navigate by concept instead of just sequence.
Study spaces Creates places for guided review, fundamentals, and deeper interaction with the material.
Built on Notes2Tree Structure is generated from what the work already contains — never imposed from outside.

Built for working
scientists.

Kedion isn't a publication venue or a credentialing body. It's tools. You use them to do better work. You publish that work wherever you like — arXiv, Nature, Science, your own preprint server, anywhere. The validation is the work itself.

// 01

Independent researchers

Building rigorous work without institutional infrastructure. The four tools give you the discovery, structuring, rigor-checking, and teaching surfaces a university would otherwise provide.

research outside the university
// 02

Scientists in institutions

Using AI tools to find cross-domain connections traditional literature search misses, and to strengthen submissions before they go out. Affiliation is fine — Kedion just makes the work better.

strengthen the submission
// 03

Research collectives & labs

Teams that want a shared structural workspace across their projects. Notes2Tree and Dekyon make it possible to build, refine, and teach a body of work together — across institutions or outside any.

shared structure across teams
// 04

Grad students & postdocs

The audience most underserved by current tooling. The tools that strengthen senior researchers' work also help early-career researchers learn what rigor looks like — by surfacing gaps before reviewers do.

rigor before peer review

The tools are themselves scientific contributions.

Kedion is run by a working scientist, not a content company. The tools we ship are grounded in peer-reviewable work — and we publish that work openly, with the same rigor we ask of our users.

This matters because the alternative — AI tools for science built without scientific accountability — produces exactly the kind of confident-but-shallow output the field is already drowning in. The framework underneath is supposed to defend itself. Ours does.

// in flight

A framework for finding real connections across scientific fields rather than surface-level ones.

The same framework is the engine behind LatentLink and the novelty checks in Manifest. Currently in revision; preprint available on arXiv.

Read the preprint

Start with what's free.
Add what you need.

Notes2Tree and the basic version of LatentLink are free forever. Manifest and Dekyon are in pilot. Independent researchers, students, and collectives can apply for fee waivers.

LatentLink Pro
Soon
Higher search limits, deeper cross-domain analysis, and priority access to new features as they ship.
  • Higher daily limits
  • Three-domain searches
  • Hypothesis export
  • Citation evidence trails
Manifest
Pilotlimited
The pre-submission rigor diagnostic. Currently rolling out to a small group of early researchers.
  • Citation completeness
  • Methodology stress-test
  • Real novelty check
  • Cross-domain implications
Dekyon
Pilotlimited
For researchers, labs, and educators who want to turn finished work into something others can learn from.
  • Concept extraction
  • Topic trees
  • Study spaces
  • Powered by Notes2Tree

Tools for working science.

Start with LatentLink. Add Notes2Tree at any time. Pilot Manifest and Dekyon when you're ready. The workflow is built to grow with the work.

Open LatentLink →