How it works

Everything you feed it becomes
a knowledge graph you own.

Here's every step stoa runs for you, start to finish.

The pipeline

Six moves, from dropped-in to living knowledge

Everything that makes a “second brain” a chore — cleaning, tagging, linking, keeping it current — stoa does automatically. Nothing is saved until you approve it.

You drop it in

Drop it in. stoa does the rest.

A PDF, a messy transcript, a forwarded email, a web link. stoa reads it, strips the boilerplate, and figures out what it actually says.

No tagging, no folders, no templates. stoa infers the document's type, cleans it into readable text, and writes its own summary — the work you'd normally do by hand, done for you.

You drop in anything
Acme Q3 launch review.pdf
Weekly sync — transcript
Re: pricing tiers (email)
an article you saved

No tagging. No folders. No templates.

Signal over noise

Strip the noise, keep the signal.

Headers, footers, page numbers, email quoting, navigation cruft — the boilerplate that pollutes retrieval is removed first, leaving clean, readable text.

You get back a tidy markdown version of every source, so the reading surface is yours too — not just an index you can never look at.

Boilerplate removed
CONFIDENTIAL — page 7 of 24
Managed billing flows through Acme Billing;
usage-export is a hard dependency.
— — — — — footer • acme.example — — —
©2026 · printed 06/24 · do not distribute
Clean markdown, kept for you to read — not just indexed.
Claims & entities

It breaks each source into its smallest claims.

People, projects, decisions, dates — the discrete facts inside the prose, pulled out and made queryable.

Instead of storing a wall of text, stoa extracts the individual claims and the entities they're about — so you can ask precise questions and get precise, sourced answers back.

“Sam will drive the EMEA reseller sign-off — it’s the gate on the Q3 launch.”
broken into
Entities
Sam Carter Q3 Launch Acme
Claims
claimSam owns the reseller sign-off
claimEMEA go-live gates the launch
Reconcile — the part everyone skips

One name for one thing — across everything.

stoa does the synthesis you'd otherwise do by hand: it recognizes “Bob” and “Robert” are one person, retires outdated facts (with history), and cross-links everything — once, then keeps it current. That maintained depth, not just search, is the product.

One name for one thing
“Acme”
“Acme Corp”
“Acme Billing”
resolved to
Acme CorporationOrganization
Contradictions surface — with history
EMEA go-live date
Sep 1older source · retired
Sep 15flagged · both kept

This is the hard part everyone else skips. stoa recognizes when two sources mean the same thing, merges them, and when a new claim clashes with a settled one, it flags the disagreement rather than silently stacking both.

It stays coherent

A knowledge graph that maintains itself.

Entities link to entities. Pages rewrite themselves as you add more. The picture stays current without you curating it.

Every source you add updates the graph and the cited wiki pages built on top of it. Nothing goes stale in a drawer — the whole picture re-coheres each time you feed it something new.

Q3 Launch

Bundles the managed edition with hosted MCP. Gated on pricing sign-off, the Acme Billing integration, and Sam Carter’s sign-off.

cited · rewrites itself as you add more
Reach it from anywhere

Not another app to switch to.

stoa does not replace the AI assistants, clients, and front-ends you already work in. It's the knowledge layer underneath them — reachable in whichever place you already do your best work: your terminal, your AI chatbot, your editor, all over MCP.

Access it headlessly from your tools
one MCP server · over MCP or CLI
AI assistants
Claude, ChatGPT, Amazon Quick + any other
Terminal tools & agents
Claude Code, openclaw, Hermes + any other
Editors & IDEs
VS Code, Cursor, Windsurf + any other
Secure auth, your way

Browser sign-in for connector clients (opt-in on your instance).

Or open the portal that ships with it
app.stoa.wiki
stoa/Product & Strategy
Search ⌘K
Your brains
Ask Product & Strategy anything…
How stoa derives an answer

Let's walk through it together.

Curious how a question becomes a cited answer? Step through it at your own pace — each move explained in plain language, no algorithm background needed.

Step 01 / 07
Your question

“What's blocking the Q3 launch?” — no keywords, no query syntax. stoa takes it from here.

Why

You shouldn't have to phrase things the way a database wants.

“What's blocking the Q3 launch?”
plain language — no keywords to guess
On your machine or managed by us
A self-evolving wiki you fully own

A cited wiki that rewrites itself as you read and write.

Every source you add re-coheres the graph and the wiki pages built on top of it — linked, cited, and current, with nothing to curate. It's a folder of plain files you own: open it, move it, or delete it, no lock-in.

The wiki pages are stored in the — an open standard for knowledge that AI agents can read directly, so your whole brain can be used as-is by other tools, with no custom integration.

stoa.wiki/brains/greek-philosophy Wiki
Pages
Concept24 claims

Propositional logic

The logic your computer runs on — invented 2,000 years early. Propositional logic treats whole statements as single true/false values joined by operators like "if… then", "and", "or".

Why it matters today

This is exactly how a computer thinks: every if/else a program runs is propositional logic in action. It became Boolean algebra and the switching circuits inside every chip.

Who founded it

The Stoics — systematized by (3rd c. BCE), over two millennia before the transistor. (Aristotle founded the other branch, term logic.)

  • Built on — "if… then", "and", "or"
  • Codified as the five — the first rules of inference
Related

See it on your own documents.

Free, open source, and running on your machine in one command.

Open source·Self-hostable·stoa.wiki