---
schema: 1
type: term
title: '**provenance**'
description: >-
  **Provenance** — in this material — is the record of _where each piece of text
  in a document came from_: specifically, whether it was written by a human or
  by an AI agent. It is tracked through color-coded authorship (purple for
  AI-written content, green for human edits), so readers can see what a human
  actually wrote, thought about, and stands behind [^6fa3b152@397].
canonical: 'https://liquidgold.cc/channel/everyinc/explore/term/provenance-b84aadf9be00'
markdown: 'https://liquidgold.cc/channel/everyinc/explore/term/provenance-b84aadf9be00.md'
llms: null
parent: 'https://liquidgold.cc/channel/everyinc/llms.txt'
channel: everyinc
updated: '2026-10-23T22:12:35.346Z'
language: null
children: []
---


# \*\*provenance\*\*

Canonical: https://liquidgold.cc/channel/everyinc/explore/term/provenance-b84aadf9be00

Parent: [Channel index](https://liquidgold.cc/channel/everyinc/llms.txt)

## Definition

**Provenance** — in this material — is the record of _where each piece of text in a document came from_: specifically, whether it was written by a human or by an AI agent. It is tracked through color-coded authorship (purple for AI-written content, green for human edits), so readers can see what a human actually wrote, thought about, and stands behind [^6fa3b152@397].

**Provenance** here means the ability to trace _where each piece of text in a document came from_ — specifically, whether a human wrote it or an AI agent wrote it. It is the original motivating problem behind **Proof**, Every's collaborative editor: as planning docs get written increasingly by agents (like Notion agents producing 2026 plans), how do you tell what a human actually wrote, thought about, and stands behind?The first version of the tool was a Mac app with a **colored gutter** that marked authorship directly in the text:  - **purple** = AI-written text - **green** = human-written text  The live demo showed the same idea in the web editor: agents joining a document write content marked purple, while human edits show up green. For one team member this was the deciding factor — in Google Docs it was clunky, even impossible, to track what he wrote versus what the agent wrote. Provenance also survives into everyday workflows: a writer deliberately keeps his drafting mostly his own words by only letting the agent draft after he has monologued into the doc, so the colored record reflects his voice. Watch the turning point when provenance-tracking evolved into real-time collaboration.As the product matured, the team widened provenance into a broader mission. Proof's job is _not_ to organize documents — it is to **communicate about writing, ideas, and where writing came from**. Once a doc is mature, it graduates to heavier tools like Notion; Proof remains the place where authorship and ideas are visible while the writing happens.- **Provenance = tracing the origin of text**: whether a human or an agent wrote each part of a document. - It was the **original motivation** for Proof, driven by agent-generated planning docs. - The implementation is **color-coded authorship**: purple for AI, green for humans. - Provenance helps writers stand behind their work and track their own contributions versus agent contributions. - The idea grew into Proof's broader purpose: communicating **where writing came from**.