The screen that asks before anything you write goes to an AI service, and how to change your answer
Before anything you write is sent to an AI service, Streamliner asks you. The screen is called Before the assistant answers, and it appears before your first message in a streamline that uses a service you have not been asked about.
What the screen tells you
- Which service. By name, such as Anthropic (Claude). Your streamlines use it to answer you.
- What is sent. What you write, the results of the tools the assistant uses for you (a customer record, an order, a search), and the chat so far, so it can answer in context.
- Where it goes. In an organization: to the service, through your organization's account and our servers. On this device (free): straight from your device to the service, with your own key, through nothing of ours.
- What is never sent. What you type into secure forms and on sign-in pages: card numbers, ID numbers, passwords. See What is sent, and what never is (PR-2).
Your choice
- Allow lets the assistant answer.
- Don't allow stops anything going to that service. You can still open the app and read your chats; the assistant just won't answer. A message you send gets a calm Not sent line instead.
In an organization, the screen still asks you personally, and names your organization. Your organization chose the service; whether your own words go to it is your agreement to give.
You are asked once per service. If a streamline starts using a different service, you are asked again.
Changing your mind
Settings → Accounts → AI providers lists, for each account, each AI service its streamlines use, with a switch, marked Allowed or Not allowed (with when you chose) or Not asked yet. Change it there at any time.
The Meridian demo never asks: its assistant is scripted on your device and sends nothing to any AI service.
The same manual ships inside the app. This section as Markdown