Exponential Intelligence
This page is for the owner who decides whether the team uses the writing assistant, and how. Exponential Intelligence
is built into the administration: editors ask in plain language for new content or changes, and the assistant looks
up what it needs and writes drafts, as the signed-in editor and with that editor's permissions (README.md,
"Exponential Intelligence"). How to use it well is on Using AI drafts responsibly.
Contents
- What it does for editors
- The rule that protects you
- Where text goes
- Costs and limits
- Who gets access
- Decisions before switching it on
What it does for editors
From README.md and expai: Exponential Intelligence:
| Where | What |
|---|---|
| The Intelligence tab of an item | a conversation per user and item; each lookup and draft is a step; the drafts it wrote are listed with links to edit, view and discard |
| The edit form | "Fill the fields" from a brief; Write, Rewrite, Shorten, Translate and Suggest per field; presets for title, summary, meta description, keywords; an Improve menu; values go into the form for review |
| Images | alternative text suggestions from a vision model, and a "Missing alt text" report for a subtree |
| Vector | a workspace for larger jobs: a plan with a cost estimate, review of every change per field before anything is written, What changed with revert |
| Chains and translation sync | write, review, revise and translate in steps with a human checkpoint; translations brought up to date after the source changed |
| A dashboard | cost, tokens, requests, drafts and limits |
The rule that protects you
- Drafts only. It never publishes on its own. "Send for approval" runs the normal publish operation as the
editor, so an approval workflow holds the version for the approver (
README.md). - The editor's own permissions. Every lookup and draft runs with the editor's content permissions.
- Labeled. Every version it wrote carries the label "Exponential Intelligence" in My drafts, version views, edit forms and approval items.
- Audited. Every request, draft and state change is an event of the audit channel
ai; prompts, answers and keys are never recorded.
Where text goes
| Provider | Region |
|---|---|
| Anthropic (Claude) | as stated by the provider |
| An OpenAI-compatible service (Melious.ai in the EU region, Hugging Face, OpenRouter, Mistral and others) | as stated by the provider |
| A local open-weight model through Ollama | on your own server; the data never leaves it |
Each provider states its region (us, eu, local) and retention; rules per section and content class decide which
content may go where, and every request records where it went. EI: Summary is one printable page for a data
protection officer (README.md). Only models that passed a live tool-calling probe, or are trusted by
configuration, may write drafts.
Costs and limits
A usage log keeps the estimated cost of every request; limits can be set per user, day, month, site and provider, and
budgets for queued work (README.md). Ask your administrator to set limits before the team starts, and look at the
dashboard monthly. Provider prices are set by the providers, not by Exponential.
Who gets access
The assistant's features are separate policies (expai/write, expai/vector, ...). On a new installation only the
Administrator role holds them (README.md). Give expai/write (and ezjscore/call) to the editors who should use it;
keep Vector and the administration of providers with fewer people. The switch [AISettings] Enabled turns it off
everywhere at once, or per siteaccess.
Note
A new installation activates the extension for the administration siteaccesses, but it sends nothing anywhere
until a provider key is added (README.md). The key belongs in an override settings file, never in a committed
file. Cron parts expqueue and expai run its queued work.
Decisions before switching it on
- To do: Which provider, in which region, with which retention: decided with your data protection officer. This is not legal advice.
- To do: Which sections or classes must never be sent to an outside provider.
- To do: Who gets
expai/write, who gets more. - To do: Monthly and per-user limits.
- To do: That approvals stay in place where they were, for AI drafts too (Approvals and review).
- To do: The team's rules: Using AI drafts responsibly.
Related pages
- Using AI drafts responsibly
- Assisted translation with Exponential Intelligence
- expai: Exponential Intelligence
- Free documentation: this page is free to copy, translate and improve under the GNU FDL 1.3.
Become an Exponential Partner
Partners share their information, provide funding, contribute back and develop the products with us, and that is how Exponential grows. A partner website and dashboard are planned for mid-2027: please become a partner today to speed up our growth rate and land new features faster.