Nora, the assistant

1. Opening the chat and its modes

Nora opens with the round button in the bottom right corner (“Ask Nora”). On the «Dashboard» it is also built in at the top, in the “Nora” block.

The chat opens floating, with the title “Nora” and, below, “Context: …”, the screen you are on. The header icons are, from left to right: back (starts a new conversation), “Nora guide” (this guide), “History”, “Dock” (pins it on the right, beside the screen; “Undock” returns it), “Maximize” (full screen) and “Close”. With an empty chat it suggests example questions.

2. Asking

Type in “Type your message...” and press Enter or “Submit”. While it works you will see “Thinking…” or “Consulting data…”. Nora looks up your company's data (workplaces, assessments, measures, incidents, inspections, PPE, training, contractors, documents…) and takes into account the screen shown in “Context”: if you ask about «this» without naming it, it understands you mean what you have open.

For documents it looks up the list (title, type, status, version), not the content of the files; you cannot attach files in the chat. Under each answer, the thumbs are for rating it.

3. Changes Nora prepares: confirm or reject

If you ask it to create or change something, Nora replies with a “Suggested action” card with the record type, the operation and the data. With the “Manual” autonomy (the default) nothing is saved until you click “Confirm” or reply just «yes», «ok». “Edit” changes the data on the card and “Reject” discards it. Once saved, “View record” opens it: check it and complete whatever is missing.

In “Preferences” → “Assistant” you can change the autonomy to “Semi-auto” (runs what is reversible and not sensitive) or “Auto” (runs everything your permissions allow); Nora then tells you with “I did this automatically:”.

4. How to judge an answer

When Nora can link a record, a tag appears under the answer that opens it. If there is no link, check the figure on its screen. Check the figures in the module and the legal references in the official source before using them: Nora can be wrong, and its answer does not replace the judgement of a prevention technician or of the medical service.

5. Permissions and privacy

Nora works as your user and in the active company: it only sees what you can see and only prepares what you could do. If it does not find a record, you may not see it with your permissions. With a read-only user it answers but does not prepare changes. Clinical data is only available to those with medical access.

Each user only sees their own conversations, in “History”. What you type is sent to the installation's AI service: do not write health data or personal data that is not needed.

6. Long conversations

Nora takes into account the latest messages, a summary of what came before and the data you have corrected (the last value counts). In a very long conversation it may lose details from the beginning: start another with the back arrow when you change topic, workplace or company. The earlier ones stay in “History”, where you can archive or delete them. If it replies with an error, send the question again; if it happens again, tell support.