#FAQ

Straight answers to the questions people ask most about working with Nourva.

#Overview

This page answers the questions that come up most often: how Nourva decides what to do, whether it is safe to let it act, what it remembers, how to stop it, how credits work, what happens to the things it makes, and why an update is sometimes required.

Every answer here stays at the level you actually experience — what you ask for and what you get. If you want the longer version of any topic, follow the links at the end of each answer.

#Questions & answers

#How does Nourva decide what to do?

You tell Nourva the result you want, not the steps to get there. Nourva understands the goal, plans an approach, carries it out across the right surfaces, checks its own work, and reports back with something you can see and reuse. When a goal is ambiguous or risky, Nourva asks first instead of guessing.

You never have to specify the individual steps. You describe the outcome and let Nourva find the path. See Outcomes, not steps and How Nourva works.

#Is it safe to let Nourva act on my behalf?

Yes, and you stay in control of how much freedom it has. You choose a permission level for any task:

• Look only — Nourva reads and inspects but never changes anything.
• Plan first — Nourva proposes a plan and waits for your approval before acting.
• Ask before acting — Nourva pauses for your permission before each change.
• Work in your workspace — Nourva acts freely inside your project or workspace, and nothing outside it.
• Act automatically — Nourva carries the whole task through on its own.

Start cautious when you are exploring, and give more freedom for long, well-understood tasks you trust. You can change the level at any time. See Permissions & control.

#How do I stop Nourva once it has started?

Stopping always works immediately, regardless of the permission level. You can stop a running task at any moment, change its permission level, or step in to steer it. In the Shell, you can also stop just the command that is currently running and keep the session, or close the session cleanly when you are done.

See Permissions & control and Built-in session commands.

#Can I trust it when Nourva says a task is done?

Yes. Every action Nourva takes is automatically checked before it is reported as done. If a step does not actually succeed, Nourva notices, says so honestly, and retries or adjusts instead of pretending it worked. A "done" from Nourva means the outcome was verified, not just attempted.

See How Nourva checks its work.

#What does Nourva remember about me?

Nourva keeps a layered memory that is encrypted and private to you. It remembers what matters about your work and preferences so you do not have to repeat yourself, and it leaves out what does not matter. This is what lets Nourva feel continuous instead of starting from zero each time.

You can browse what it holds, export it, and tell it to forget anything. Your memory is yours. See Privacy & memory and Viewing, exporting & forgetting.

#What happens to the results Nourva produces?

The things Nourva makes — documents, research, sites, media, and results — are kept as artifacts you can find again, reopen, reuse, and export or download. Results are not throwaway; they are saved so you can return to them, build on them, or share them.

Results are presented in whichever of eight formats best fits the work: table, chart, report, comparison, dashboard, media gallery, alert, or source pack. Nourva chooses the format automatically, and you can ask for a different one. See Results & artifacts and Artifacts & where they go.

#How do credits work?

Some of Nourva's heavier work uses credits. You have a balance, certain kinds of work draw from it, and your balance updates as you use Nourva and as you add to it. Lighter, everyday use is simple; the more intensive surfaces — like a deep multi-source investigation — are where credits come in.

See Plans & credits.

#Why am I sometimes required to update before continuing?

Nourva keeps itself up to date, and new versions arrive smoothly. Sometimes an update is required before you can continue so that everyone is on a current, supported version. Updating is quick and keeps you on the latest, most capable Nourva.

See Updates.

#Which surface should I use for what?

A quick guide:

Chat is the simplest place to start; when a job is bigger, Chat points you to the right surface.

#Can Nourva use current information from the web?

Yes. When you need current information, Chat can search the live web and answer from what it finds, telling you what it based the answer on. For deeper questions, Arachne and Deep Research gather from multiple real sources and give you a source pack so you can check the origin of any claim.

See Asking for current information and Source packs.

#Will I lose my work if I close the window or my machine restarts?

No. In the Shell, every session keeps a durable receipt of what it did, so if you close the window or your machine restarts, the session's history and result are not lost — you can return to it and see exactly what happened. Long jobs keep running in the background, and you are told when they finish.

See Durable receipts and Running in the background.

#Tips

• When you are unsure how much freedom to give, start at a lower permission level and watch the first run — you can raise the level once you trust the task.
• If a result arrives in a format that does not suit you, just ask for a different one of the eight; the same result can be shown as a table, chart, report, and so on.
• Check what Nourva remembers from time to time, and tell it to forget anything you would rather it not keep — your memory is fully under your control.
• For anything that needs verifiable sources, prefer a research surface so you get a source pack alongside the answer.

#Related

• Welcome to Nourva
• How Nourva works
• Permissions & control
• Plans & credits
• Command reference