ABOUT
Trustworthy knowledge behind every AI decision
Re0 turns public documentation and private material you are entitled to use into versioned libraries an AI agent can search, retrieve and cite. What matters to us is not whether a model sounds convincing, but where the passage it cited came from, which version it belongs to, and whether someone else can check it.
WHY
What we are solving
The docs an agent gets are usually stale
Knowledge baked into model weights has a cutoff date, and a web scrape is a snapshot of one moment. Neither can answer "which version is this passage from". Re0 organises libraries by version, so every result points at one definite, traceable version.
Citations cannot be checked
Text stuffed into a context window usually arrives with no source, no timestamp and no proof it was not rewritten on the way. Every Re0 result keeps its source and citation, and points at one specific published version.
Billing is tangled up with content
Per-token pricing turns long context into a cost risk, and selling by library charges twice for knowledge that is already public. Re0 bills one accepted query, and the amount of content returned never changes the price.
PRINCIPLES
What we hold to
Billed per call, nothing else
One accepted API or MCP query is 1 call, whatever the chunk or token count. Every public library is queryable without buying access library by library.
Publishers get a share
Claimed public libraries earn a share of platform revenue based on successful retrievals — an initial 20% rate, allocated linearly by call volume rather than weighted by Trust Score.
Boundaries stated plainly
We provide retrieval and citation, not conclusions. Payouts run through an external payment provider — we store no bank account numbers and hold no user funds.
SCOPE
What the platform is made of
Retrieval, integration and ownership, over one versioned dataset.
Libraries
Sources, versions and indexes
GitHub repositories, websites, llms.txt and OpenAPI documents are parsed and chunked on connect, then published by version, public or private. Public libraries submitted by users go through review.
Integration
REST, MCP and CLI
The same retrieval rules are exposed as a REST API and an MCP endpoint that Claude, Codex, Cursor and others connect to directly. The CLI writes and cleans up local configuration, removing only what it created.
Ownership
Claiming and revenue share
Ownership of a public library is proven by a DNS TXT record or a well-known file; a declaration inside a config file changes nothing. Revenue share begins only once a claim completes.
Want to talk about partnering, claiming or integrating?
Content ownership, enterprise integration and security disclosure each have their own address, and we route by channel.