Away from the bare chat window
An empty text field that spits out a finished answer after a few seconds feels like a black box. We've extended the chat — both for you and in the Anwaltsportal — with four things that together create a different feeling: the AI works visibly on your behalf, not anonymously in the background.
You can see what the AI is currently working on
While an answer is taking shape, a small activity feed shows the real working steps: "Reading document", "Checking playbook", "Researching". Only steps that were actually carried out appear — no fake progress. If a step fails, it stays visible as failed rather than disappearing without a word.
Finished results as their own card
When the AI delivers a complete result — an email, a letter, a legal citation, a table — it no longer appears as running text but as a typed card with the matching actions: open the email straight in your mail app/Gmail/Outlook, download the document as DOCX/PDF or file it in your case, check the citation against its source.
Lexi thinks one step ahead
After an analysis or answer, Lexi actively suggests the next sensible step — with a visible origin ("based on your document analysis", "based on playbook …") and the price before you click, if the step costs credits. If you decline, the system remembers that permanently.
Everything in one place: the work status
New is the work-status sidebar next to the chat — on a smartphone, a panel that swings in from the side. It bundles together which document/which case is currently in context, all the answer cards from this conversation (clickable), open next steps and open recommendations. Beneath every answer you'll also consistently find Copy, Rate and "Turn into" — adopt as a letter, export as a document, hand over to a lawyer or save the conversation.
Also in the Anwaltsportal
The AI assistant in the Anwaltsportal gets the same work status, supplemented by the session's context usage — a shared look and feel for both sides of the platform.
Staying honest
None of this replaces your own review: the activity feed and artefact cards show WHAT the AI has done, not that it's error-free. Citations remain verifiable, recommendations name their source, and at every step you decide whether to continue. More on this: How reliable is the AI?.
