You can see what the AI is currently doing
While the AI is working, an activity feed shows the actual steps – e.g., reading your document or looking up a legal norm.
When you ask a question or have a document analyzed, the AI does not work in a "black box". Above the emerging answer, a small, collapsible activity feed appears, showing the actual work steps – even before the first word of the answer appears.
What you see there
- Your document is being read – if your question refers to an uploaded document.
- A statutory text is being looked up or a deadline is being calculated – if the AI is using a tool for this.
- Each step initially shows a spinner (currently running) and then a checkmark (completed).
Only real steps – no fake progress
The feed only reports steps that actually happen. If a step fails – such as a research task – it remains visible as failed instead of secretly disappearing. This allows you to assess what the answer is based on (how reliable the AI is).
Why this helps
You can see at a glance whether the AI has actually consulted your document and which sources it has checked. This makes the answer comprehensible – in line with the basic principle that the AI discloses its derivation (how the AI works). You can ask questions directly about the document in the document chat.
Related topics: how the AI works · document chat · how reliable the AI is · what the AI is · answer cards · proactive recommendations · work status
Was this article helpful?
