Early access is open — spots are limited. Check availability →

THE PRODUCT

Synthesize knowledge from any source.

NotebookLM and Perplexity answer questions about your sources. SILKLEARN reads the structure through them — what depends on what, where sources contradict each other, what order the material wants to go in — so you get a path, not a chatbot you have to interrogate.

Step 1

Upload

Drop in what you're trying to learn, exactly as it exists. Research papers, codebase docs, textbooks, SOPs, onboarding notes — any format, any structure. SILKLEARN reads it as-is.

Supported: PDF, Markdown, DOCX, Notion export, Confluence export. GitHub and Google Drive coming soon.

Step 2

Review

See the structure the system compiled. The dependency graph shows what concepts build on each other. Contradictions surface where your sources disagree. You check the map before you follow it.

Inspect every connection. Accept it, reject it, or flag it for further review. The compiled structure doesn't become your path until you've verified it makes sense.

Step 3

Follow

Work through the path the system built. Your documents, in the order they actually depend on each other — not the order someone uploaded them, not alphabetical, not by date.

Export as a structured reading list, share with a collaborator, or feed as structured context to an AI system that needs the right order, not a retrieval guess.

What you get out of one reviewed graph.

Learning path

You have a pile of sources with no obvious order. SILKLEARN reads the dependency structure through them and hands you a sequenced path — from the source that unlocks everything else to the one that builds on all of them.

Source map

See which sources depend on which foundations. The structure that was implicit in your material becomes explicit and inspectable before you follow it.

Compliance review queue

Every flagged item carries its source, the reviewer's name, and a timestamp — export-ready when the auditor asks.

AI context bundle

Your AI assistant gets the knowledge in dependency order, from sources a human already reviewed — not a retrieval guess.

Capabilities

Eight things it actually does.

01

Works with your existing docs

PDF, MD, DOCX, Notion, Confluence. No reformatting.

02

Finds the order your docs already imply

Reads the order your documents already assume. Doesn't invent one.

03

Review the graph before you follow it

Visual graph. Inspect every dependency connection. Accept, adjust, or flag before you start reading.

04

Every review decision is logged as it happens

Timestamps and reviewer names logged as you review.

05

Surfaces contradictions between docs

When two docs disagree, a conflict flag appears. You decide.

06

Every output linked to source

Every item traces back to the exact doc section it came from.

07

Integrations

Notion · Confluence · Drive · GitHub · Slack · API (beta)

08

Update and re-publish

Docs change? Re-compile, review the delta, republish. Done.

The structure is already in your sources. You just can't see it yet.