AI policies in BrainBox: control your team's models and capabilities
How an admin decides which AI models the team uses and what the assistant can do in each Box.
Lesson 20 of 21 · 1:53
What it covers
As an admin you decide how your team uses AI. In AI Policies you turn each model on or off for the whole workspace, by lab or one by one, and in each Box you choose what the assistant can do: search the web, run code, edit files or use plugins.
Frequently asked questions
›Can I limit which AI models my team uses?
Yes. In AI Policies every model has a switch. When it's off, members need an individual exception to use it.
›How do I enable a model for just one person?
In Members, open that person's AI access and enable the model. Their settings override the workspace policy. You can also set an intelligence unit limit there.
›Can I stop the AI from searching the web in a Box?
Yes. In Box settings you turn off Web Search, or any other capability, for everyone who chats there.
Transcript
0:00The concern
Andrés is worried: what if someone uses the most expensive model for everything, or the AI goes searching the internet for things it shouldn't? Relax: in BrainBox, you set the rules.
0:13Two levels
The rules have two levels. In the workspace you decide which models the team can use. And in each Box, what the assistant can do.
0:24The team's models
In AI Policies you see every model, and how many are enabled. Filter by lab: Anthropic, OpenAI... Every model has its switch: if you turn it off, nobody on the team can use it, unless you give them an exception. And you can turn a whole lab on or off, in one click.
0:45Exceptions per person
And if you turn a model off for everyone, but Julián, from IT, does need it? In Members, open his AI access. You see how many units he has used this period, and if you want, you give him his own limit. Below are all the models: what you enable here applies only to that person, above the team policy. Save, and done: everyone with the general rule, and Julián, with his exception.
1:13Capabilities per Box
In each Box, open its settings. Here you decide whether the assistant can search the web, run code, edit files or use plugins. Whatever you turn off is off for everyone who chats in that Box.
1:29Recap
To recap: in AI Policies you choose your team's models, in AI access you give each person exceptions and limits, and in each Box you decide what the assistant can do. Andrés can rest easy. AI works for your team, by your rules.