When Cortex does substantial work, the output is a deliverable: a structured document you can trust, trace, and share - not a chat message that scrolls away.
Anatomy of a deliverable
- ›The content: sections, tables, and charts appropriate to the play that produced it.
- ›Citations: which records and documents the analysis drew on, so claims are checkable.
- ›Status and history: queued, running, ready; every version kept as inputs change or you request revisions.
- ›Discussion: comments and reactions from your team, in place.
Where they live
Cortex, Deliverables lists everything produced - by play, requester, and date. Deliverables produced by Autopilot or agents land in the same place, so there is one shelf for AI output.
Working with one
Ask Cortex to revise a section, tighten an assumption, or re-run with a different period; each revision becomes a new version with the old one preserved. Export to share outside Brainis.
Trust, but verify
Tip: The citations block is the fastest quality check. An analysis grounded in 40 records reads very differently from one grounded in 3 - if the data behind a conclusion is thin, the deliverable says so.