Ask your business anything, and keep the thread.
An AI assistant that answers from your live business data and lists the sources behind every answer. Conversations are saved so you can pick them up next week, and it always asks before changing anything.
A leave request, an overdue invoice, a contract needing a signature and an expense claim. Four systems everywhere else. One list here, because they are four rows in one database.
The surface
A chat you come back to.
Your recent threads sit on the left, with four starter prompts when you need one. Every answer lists its sources, any change arrives as a card you confirm, and the AI's colour appears only on the message box, so the rest of the screen stays quiet.
It has not moved since the handover review on 14 July. The last thing anybody did was attach the revised schedule of rates; nothing has been logged against it since.
Read from 2 deals, 1 meeting and 4 tasks you can see.
The thread is kept. Pick it up next week and it still knows what you were asking about.
It asks before it changes anything.
The thread is still there next week.
Every conversation is saved, titled from your first message, grouped by week and pinned if it matters.
- Pick up Monday's question on Friday, with the context intact
- Pin the threads worth keeping and archive the rest
- Start from one of four prompts, or just type
- Talk to it out loud on the same thread, where voice is enabled
- Each department chooses which AI model answers
Nothing changes until you say so.
A proposed change shows the task, the date and the owner in plain words. Nothing happens until you tap confirm.
- Review every proposed change before it happens
- See the sources named under every answer
- It can only see what your account can see
- If it has nothing to go on, it says so instead of guessing
- Every action stays on the thread, so decisions can be revisited
Access
Who can do what.
Everyone can ask. Answers respect each person's permissions, so two people asking the same question can rightly get different answers.
Ask questions, and revisit your own threads.
Confirm the changes it proposes.
Choose the AI model your department uses.
Set the persona, the guardrails and the usage caps.