Describe a goal. The agent plans it, runs it, and delivers.

Describe a goal in plain English — "write a research brief on X from my sources" — and the agent breaks it into sub-steps, selects the right operations, and executes them in sequence. You review the result, not each click.

The agent retrieves from your own notes, documents, and uploaded sources — not the open web — so outputs are grounded in your material, with citations back to the exact passages used.

Name an agent run and it becomes a Skill node your whole Space can invoke. Build a "weekly roundup" or "repurpose and schedule" recipe once, then run it with one click or share it across your team.

Point the agent at a schedule ("every Monday draft a roundup from this Space") or an event ("when I publish in this Space, run this skill") and it works unattended — governed by the same role and tier permissions you have.

Every agent run persists its plan and per-step trace in a collapsible timeline on the result node. You see exactly what it did — which ops it called, which sources it retrieved, what it produced at each step.

The agent dispatches the same registered operations you can run by hand, under the same role and tier gate. It cannot do anything you couldn't do yourself — so there are no hidden side-effects.

What can a JustJot.ai agent actually do?

It plans a sequence of registered operations — retrieve from your sources, draft a section, repurpose for a channel, build a study pack, analyze a data table — and executes them in order, returning a deliverable stored as a normal node in your workspace. You can chain operations the agent can't yet do by combining Skills.

Can I run agents automatically without triggering them manually?

Yes. You can schedule an agent run on a cron schedule ("every Monday at 9 am") or set it to trigger on an event — for example, whenever you publish a note to a Space, a skill that repurposes it to social variants runs automatically.

How is this different from a plain AI assistant?

A plain assistant responds to a single prompt. JustJot.ai's agent breaks a goal into a plan, selects the right operations from the control-plane registry, runs them in sequence over your private library, and saves the result as a node you can publish, paywall, or chain into the next pass. It works autonomously, not just turn-by-turn.

What are Skills?

Skills are saved, named agent recipes — a step sequence with parameters that you or your team can re-run on demand. A "research brief" Skill, for instance, can be shared across a Space so every member runs the same quality-controlled workflow, with the same RBAC and tier gate.

Does the agent have access to the web?

Yes, when you use the deep-research op the agent can include web sources alongside your private library. For pure private-library work, you scope it to your own notes and documents and it stays grounded there.

Is it free?

The agent is free to try — run your first plan against your own notes without a credit card. Paid tiers unlock heavier usage, scheduled triggers, and the ability to share Skills across a team Space.