Doxbrix is AI-native. AI isn't bolted on as a single chatbot — it runs through the entire documentation lifecycle, helping authors write better content faster an...
Doxbrix is AI-native. AI isn't bolted on as a single chatbot — it runs through the entire documentation lifecycle, helping authors write better content faster an...
Doxbrix is AI-native. AI isn't bolted on as a single chatbot — it runs through the entire documentation lifecycle, helping authors write better content faster and helping readers find answers instantly. This page maps every AI capability so you know what's available and where to go next.
Two sides of AI
This page covers the author-side features. For the reader-side assistant, see Site assistant.
Author-side AI at a glance
/assistant that knows your whole workspace — ask it to find, summarize, audit, or draft content across projects./ palette, plus a side dock that proposes edits and structure changes you approve./agent that drafts pages and sweeps projects for issues on a schedule./health of coverage, freshness, and prioritized quality issues.How the pieces fit together
Each AI feature operates at a different scale — from a single page to your whole workspace.
| Scale | Feature | What it does |
|---|---|---|
| A page (selection/cursor) | **Writing Copilot** | Generates or rewrites content inline from the `/ai` palette. |
| A page | **Document insights** + **Scan with AI** | Scores the page and lists prioritized fixes. |
| A page | **Improve with AI** | Rewrites the draft to fix flagged issues, overall or by area. |
| A project | **AI agent** + **Docs Health** | Drafts pages and finds issues across the project. |
| The workspace | **Workspace assistant** | Answers questions and drafts across all projects. |
| The workspace | **AI policy** | Governs models, budgets, autonomy, style, and privacy. |
Models and tiers
Doxbrix runs on frontier language models and offers two response tiers so you can balance speed and depth:
| Tier | Optimized for | Typical use |
|---|---|---|
| **Spark** | Fast responses, lower cost | Inline copilot, quick rewrites, fast answers |
| **Atlas** | Deeper analysis, higher quality | Quality scans, agent tasks, complex questions |
A Default tier is chosen automatically per feature, and admins can set per-feature overrides (Copilot, Scans, Agent) in the Workspace AI policy.
Governance and control
AI in Doxbrix is governed, not a free-for-all. From Settings → Workspace → AI policy you control:
- Enabled models and tier overrides per feature.
- Token budgets — a monthly workspace budget, an alert threshold, and alert recipients.
- Feature controls — turn each AI surface (Workspace assistant, Writing Copilot, Improve with AI, Scan with AI, Site assistant, AI agent) on or off.
- Agent autonomy — Suggest only, Draft for review, or Autonomous.
- Style & terminology — voice, reading level, preferred and banned terms.
- Data boundaries — no training on your content, plus zero-retention and data-residency options on Enterprise.
See Workspace AI policy for the full set, and Usage for tracking consumption.