editorzero
Design system · “Aperture” vibrant · humane · self-hostable
A
Screen A
Dashboard / Home
Acme/ Home

Start from zero.
Write with everyone — human and agent.

A self-hosted home for your team's knowledge, where people and AI agents edit side by side as peers — fully owned, fully auditable.

248
Documents across 4 spaces
▲ 12 this week
6
Agents with editor access
2 editing now
31
Teammates · 4 teams
▲ 3 invited

Live activity

Atlas agent suggested 3 edits in Q3 Launch Narrative
just now
JT
Jordan accepted an agent suggestion
4 min ago
NA
Nomi published Brand Voice Guide
1 hour ago
Scribe agent created Release Notes 2.4
3 hours ago
B
Screen B
Documents in a space — tree + list, human & agent editors
Acme/ Engineering/ Architecture

Architecture

18 documents · sorted by last edited
Document
Last editor
Status
Updated
🏗️
Sync Protocol — CRDT Notes
/engineering/architecture
Atlasagent
editing
4 min ago
Dispatcher Write-Path Atomicity
/engineering/architecture
RPRiya P.human
draft
1 hour ago
🔐
Permission Invariants & Tenant Isolation
/engineering/architecture
NANomi A.human
published
yesterday
📡
Event Outbox & Render Consumer
/engineering/architecture
Scribeagent
internal
2 days ago
🧩
Capability Registry — SSOT
/engineering/architecture
JTJordan T.human
published
3 days ago
🗄️
Dual-Backend Conformance (SQLite ↔ PG)
/engineering/architecture
SKSam K.human
draft
4 days ago
C
Screen C
Document reading view
Acme/ Design / Foundations / Brand Voice Guide
Published
Design · Foundations

The editorzero Voice: warm, exact, never beige.

How we sound when we write — for docs, product copy, and the words an agent puts on the page on our behalf. One source of truth, co-maintained by the Design team and our writing agent.

Our voice is the through-line that makes a CLI error, a tooltip, and a 40-page handbook all feel like the same company talking. It is colourful without being loud, precise without being cold. When an agent drafts on our behalf, these are the rails it writes on.

Three principles

Every sentence we ship — human-written or agent-drafted — should pass these three tests before it reaches a reader.

  • Lead with the useful truth. Say the real thing first; context second. Never bury the answer under throat-clearing.
  • Be specific, not safe. Concrete nouns and real numbers over hedged abstractions. “Converges in under 40ms across replicas,” not “fast.”
  • Warmth through clarity, not decoration. We earn friendliness by being easy to follow — not by sprinkling exclamation marks.
“If a reader has to slow down to decode us instead of the idea, we've failed — no matter how polished the prose looks.”
Agent guidance. When Scribe drafts copy, it cites this guide as its system context and surfaces any sentence that scores low on specificity for a human to confirm before publish.

Worked example

The same release note, rewritten to the voice. Notice how the second version trades abstraction for a concrete, checkable claim:

# before — vague, safe
We've made significant improvements to sync reliability.

# after — specific, warm
Edits from people and agents now converge in <40ms
across every replica — even offline-then-online.

Read the companion Terminology & Capitalisation sheet for the exact word list, and the Inclusive Language guidelines before you publish anything public.

D
Screen D · signature
Document editor — human & AI agent co-editing live
Acme/ Product / Launches / Q3 Launch Narrative
Draft
saved · synced to 3 peers
Editing now NA Nomi A. JT Jordan T. Atlas 2 humans · 1 agent · peers
/head
Blocks
H
Heading 2
section title
Paragraph
body text
Quote
callout / pull quote
AI
Ask an agent to draft…
delegate this block
Q3 Launch Narrative

Why this launch matters

This release is the moment editorzero stops being “docs with an API” and becomes a place where people and agents do the work together. The narrative below is what we'll tell customers — and it has to be exactNomi

Our sync layer is the differentiator. Edits from humans and agents are usually fast converge in under 40ms across every replica, even after going offline and reconnecting✦ Atlas — a claim we can actually defend in a demo.

AtlasAI AGENT · SUGGESTING
Replaced a vague phrase with a measurable claim → “converge in under 40ms”. Cites Sync Protocol.

Type / for blocks · to delegate a block to an agent · drag to reorder

E
Screen E
Admin — members, teams & agents
Acme/ Admin / Members

Members & agents

31 people · 6 agents · everyone is a peer principal with its own audit trail.

Principal
Role
Teams
Status
NA
Nomi Ali
nomi@acme.dev
⚑ Admin
PlatformDesign
active
JT
Jordan Tan
jordan@acme.dev
✎ Editor
Product
active
Atlas AI AGENT
agent:atlas · rate 60/min
✦ Agent editor
ProductEngineering
editing
RP
Riya Patel
riya@acme.dev
✎ Editor
Engineering
active
Scribe AI AGENT
agent:scribe · rate 30/min
✦ Agent author
Design
idle
SK
Sam Kim
sam@acme.dev
👁 Viewer
Handbook
invited

Teams

📦

Product

9 members · 1 agent
NAJT+6
2 spaces
🛠

Engineering

14 members · 2 agents
RPSK+11
3 spaces
📐

Design

6 members · 1 agent
EM+4
1 space