Getting started with Consilience

Lesson 1 of 37 min

What you'll learn

you can say what Consilience is in a few words: an ontology agent workspace
you can name the three layers (workspace, graph, agent) and say which one is the source of truth
you can explain how it differs from Obsidian and Notion
you can find your way around this course

Write notes long enough and your knowledge ends up scattered. The person you met is in one note, the company they work for is in another, the product that company makes is in a third. Each note is fine on its own, but ask "so how do these connect?" and neither search nor folders can answer.

Consilience finds those connections for you. It reads the notes you write and pulls out the people, companies, products, and ideas inside them, along with how they relate. Your scattered notes become a connected knowledge graph.

A built-in AI agent walks those connections to answer questions no single note holds on its own. You don't have to add wikilinks. Just write the way you always do.

Use a folder as a workspace

A Consilience workspace is just a folder on your disk. There's no import step and no special file format. Your markdown notes, PDFs, code, and images stay as ordinary files. The app watches that folder, so an edit you make in Finder or the terminal shows up right away.

Your notes stay plain markdown from start to finish. Extraction (the AI pass that reads your files into the graph) never rewrites them.
Removing a workspace from the recent list never touches the folder. Only the list entry goes away.

Your files are always the source of truth

Facts extracted from documents can be extracted again. To preserve the same ontology, export the graph, vocabulary, and human decisions along with the documents. Conversation history is stored separately.

Drop a .consilience/graph.nq file in the folder and it restores its graph on another machine, with no AI run needed.

reasoning

extraction

writes notes & receipts

Knowledge graph

A graph of entities, relationships, and statements extracted from notes, with their sources. Source facts can be extracted again; preserve the graph, vocabulary, and decisions to keep the existing ontology.

Three layers: your markdown files are the base, the knowledge graph is extracted from them, and the agent reasons on top. This is a preview; a later lesson walks through each layer in detail.

What an ontology workspace is

Open the graph view ("Knowledge graph", ⌘⇧G) and you can see what Consilience does. Every note, every entity it pulled out (a person, an organization, a place, an idea), and every relationship between them shows up as spheres joined by lines.

Click an entity to open its profile: its type, what it connects to, and the notes it appears in. Hover over a connection and Consilience shows you exactly which sentence in which note that relationship came from.

Layer	What it is	If it's lost
Markdown workspace	Plain files in a folder you own: notes, PDFs, code, images	The source of truth. Extraction never changes it
Knowledge graph	Entities and typed relationships pulled from your notes	Document facts can be extracted again. Export the graph to preserve the existing result.
AI agent	Answers questions by following graph connections across many documents	Conversation history is stored separately from documents and the graph.

What's an ontology?

An ontology is a way of writing down the world as objects and relationships. "Minjun works at Acme" becomes three pieces: (Minjun) → works at → (Acme). Instead of searching through whole documents, you walk the objects and relationships inside them.

Analysts at large organizations model huge amounts of material this way, then follow the connections to trace a lead. If one raw material runs short, they can jump to every order that depends on it. Consilience points the same idea at your own notes.

How it differs from Obsidian and Notion

If you already keep notes in Obsidian, you can open that same folder as a Consilience workspace. Same markdown, same [[wikilinks]], no conversion. The difference shows up next.

Obsidian only shows the links you typed by hand. Consilience needs no manual links; it reads the text and pulls out the entities and relationships itself.
Notion AI finds and summarizes your pages. That's good for a question about one page, but it tends to miss questions that need to connect facts across many pages.
Consilience reasons over the graph it extracts from your notes. It walks the connections across many notes to answer, and shows you the path it took. You ask questions and get work done on top of your documents.

How to use this course

Modules run in order, from your first steps to real workflows. Lessons run from about 7 to 18 minutes.
Try it boxes tell you exactly what to do in the app right now.
Quizzes check that an idea landed. Every answer comes with a short why.
Progress saves in this browser, so you can stop anytime and pick up where you left off.

Workspace: A plain folder on disk that Consilience opens, one per window.
Knowledge graph: The web of entities and relationships pulled from your notes, drawn as a live 3D map.
Entity: Something your notes mention: a person, an organization, a place, an idea.
Extraction: The AI reading a file into the graph. It reads only, and never rewrites the file.
Agent: The built-in AI that answers by following graph connections across many documents.

See it for yourself

Download Consilience and launch it. On a genuine first run there's no workspace to reopen, so you land straight in the main window on "Open a workspace to get started", with the titlebar selector reading "No workspace".
Stop there. Which of those buttons to press, and what happens once a folder is mounted, is the next lesson's job.

You reached the app's main window with no account, no credits, and no setup.

You've got the one-line version now: a markdown folder grows into a knowledge graph, and the agent reasons on top of it. Next you point the app at your first folder, learn how to switch between workspaces, and watch the first extraction fill that graph in.

The graph gets smarter the more it reads

Consilience only reasons over files the AI has read. The colored dots in the file tree are your status lights: a red dot means "not read yet", an amber dot means "changed, not caught up yet". Click a dot to add or update a file, and from that moment the agent knows more.

The more you feed in, the wider the agent's reach. Keep your workspace current and it keeps getting sharper.