Executive Command: what changed this week, in one screen

It would have been easy to build Executive Command as a chat box wrapped around a language model, pointed at your workspace data, and call it "AI insights." We built something narrower and, we think, more trustworthy: a deterministic advisor that answers a fixed set of real management questions from live counts, and says so plainly when there's nothing to report.

Ordered true sentences, not generated prose

Executive Command's advisor builds its answers from live aggregates -- what needs attention, what's overdue, what's at risk, what's waiting, who owes what, unresolved decisions, unassigned meeting actions, meetings today, what changed this week, and what's worth investigating. No model is called to produce these sentences. If a question is about departments or performance and your workspace has no department records, the honest answer is "no department records" -- not an invented deterioration story dressed up to sound insightful.

Every answer links to where the number actually lives

Each counter and answer in Executive Command links back to Actions, Brief, Meetings, or Recordings -- the place where that number is actually computed. A pending meeting action links directly to the recording it came from. Command is a lens on data that already exists elsewhere in the workspace, not a second copy of it.

It's gated the way the rest of accountability tooling should be

Command is only visible to workspace managers -- owners, admins, and VIRO staff -- with the right plan-level access. A Starter workspace is redirected to Actions instead of seeing a locked, teased screen. Members without manager access see an explicit access-required message rather than a broken page. The nav item itself is hidden entirely unless both gates pass, so the feature doesn't advertise itself to people who can't open it.

If we don't have live data for a question, Command says so. That's a harder discipline than generating something plausible-sounding, and it's the one we chose.