Research agent
How the AI research agent bootstraps a new project with proposed documents, tasks, fields, and context that you review before saving.
The research agent is a standalone AI service that gives a new project a running start. Instead of building everything by hand, you let the agent propose a starting structure, then review its proposals and decide what to keep.
Bootstrapping a project
In its bootstrap mode, the agent sets up a new project by proposing a coherent starting set:
- Documents — suggested files to include.
- Milestones and tasks — a first cut of the work plan.
- Fields — structured details to record on the project.
- Context entries — background notes fed into the project's AI context.
Nothing is applied automatically. Each proposal is something you review and choose to save or discard, so you stay in control of what actually lands in the project. Save the pieces that fit, adjust or drop the rest — the same review-before-save flow you'll see whenever the agent suggests changes.
Cataloger mode
The agent also has a second, cataloger mode, geared toward catalog- and template-related cataloging work rather than bootstrapping a fresh project. If you're getting started, the bootstrap flow above is the one you'll use most; cataloger mode supports the catalog side of the system.