Astrono works in two modes, and you're always in charge of which one is active. Research mode is look-but-don't-touch; Agent mode is hands-on. Nothing in your workspace changes unless you've deliberately asked for it.
Research mode: read-only
Research mode is the everyday default. Astrono answers, researches, and suggests — grounded in your files, projects, posts, and performance — but it never changes anything in your workspace on its own. You can poke around, brainstorm, and dig into your content with zero risk of an assistant quietly rearranging your work. Research mode conversations cost 0 AI minutes and are unlimited.
Agent mode: Astrono does the work
When you want something done, not just discussed, switch to Agent mode. Astrono makes a quick plan, gathers what it needs from your content, carries out the steps, and then double-checks its own work before reporting back. If it says it did something, it verified the change actually happened rather than assuming it worked.
Everything Agent mode can create or edit
In Agent mode, Astrono can create and edit:
- Projects and subtasks
- Documents
- Post drafts
- Folders
- Comments
- Tags
- Workflow stages
DeepThink for harder problems
For gnarlier problems — a messy workspace cleanup, a plan with lots of moving pieces — turn on the DeepThink toggle. It gives Astrono deeper reasoning on the task at hand, so it thinks a problem through more thoroughly before answering or acting.
What agent runs cost
Chat and Research mode are free and unlimited. Each Agent mode plan-execution costs about 12 AI minutes, drawn from your plan's monthly pool. That covers the whole run: planning, executing the steps, and verifying the result. See AI minutes explained for how that pool works and what everything else costs.