HELP & LEGAL
How Labs works, and the rules behind it.
Guides for reading experiments, decision records and the model card; the legal texts of the service below.
TYPE SOMETHING · 15 ARTICLES
Topics
Each guide explains one screen of the Labs.
Reading the Labs
The example notebook on the home page: experiments, states, owners.
Measurement
Morning run, gate, adoption.
Decision records
How to read a decision record.
Model card
What is sent, what is kept.
Changelog & roadmap
The example notebook's versions and quarters.
Requests & data
Deleting facts, data requests, wrong suggestions.
AI DATA POLICY
What B.O.S.S. sees, keeps and never trains on.
A suggestion is produced from a context selected from your records: recent documents, upcoming and pending work, contacts from the last six months and the facts in memory. Internal notes, costs, margins and file contents are not sent. Nothing is used to train models.
Chat text is not stored; the one-sentence facts in memory come from the morning run.
Every stored fact can be deleted one by one from the B.O.S.S. screen.
Any app can be removed from B.O.S.S.'s view on the Connections screen; connected mail and calendar accounts are not covered by this setting.
Recent documents, upcoming and pending work, recent contacts, facts in memory.
Internal notes, cost, margin, file contents, credentials.
No customer data is used to train or fine-tune any model.
Suggestions and deletable one-sentence facts; chat text is not kept.
Legal
Questions about a record or a suggestion? Write to the lab.
Say which sentence on which page you are asking about; the form goes straight to the lab team.
Write to the lab