AI agents: Deep Research, Curator, and Generation

Contents Lab has three AI agents. Each does real, scoped work — they are not “chat” bots.

Deep Research

The Deep Research agent takes a research question and produces a brief, an outline, and a list of real, findable sources.

When to use: Before you draft a topic, to gather sources and context.

How to use: Open the AI panel in the editor, pick Deep Research, type a question, and run. The agent streams progress over SSE; when it is done, the brief and sources are saved to your Source Library.

Quota: Each run costs ~10k tokens. Your plan’s monthly quota is consumed. When the quota is exhausted, you can use wallet balance (pay-as-you-go) or wait for the next billing cycle.

Curator

The Curator agent takes a topic and produces a structured outline (sections, sub-sections, key points).

When to use: After you have research and before you start writing.

How to use: Open the AI panel, pick Curator, point it at a topic, and run. The outline is inserted into the topic’s outline field, where you can edit it.

Generation

The Generation agent takes an outline (or a section heading + brief) and drafts the section.

When to use: When you have an outline and want to draft a section.

How to use: Open the AI panel, pick Generation, select a section, and run. The draft is inserted into the editor at the cursor, with a “Drafted by AI” annotation. You can accept, edit, or reject.

Quotas and wallet fallback

All three agents respect your plan’s monthly AI token quota. When the quota is exhausted, you can:

• Wait for the next billing cycle (subscription renews)
• Top up your wallet (pay-as-you-go pricing)
• Upgrade your plan

Audit records

Every agent run creates an audit record (in the ai_generations collection) with the prompt, completion, token counts, and cost. You can view these records in the project’s Activity log.