Six teams. One meeting-to-work engine.

Work arrives from two directions at once — monitoring alerts at 3am and client email during the day — and the only question that ever matters is what is on fire and who owns it.

What that is, in the product

• A monitoring alert opens a ticket; the same alert firing again updates the ticket it already made instead of opening a new one every minute, and a recovery closes it.
• Closing that ticket stops the SLA clock through the same code path the button uses, so the clock cannot disagree with itself.
• The built-in CRM's pipeline sits beside the project actually delivering the work, so a renewal is not retyped into a second system.

Sub-contractor dates move as a block, and the cost of finding out late is measured in plant hire. What is needed is not a prettier board but an honest answer to “what does this slip actually hit”.

What that is, in the product

• The Gantt draws real dependency arrows, flags finish-to-start violations and highlights the longest dependent chain.
• How late something is comes from checklist completion against the next milestone, blockers and the assignee's own answer — never from which column a card sits in.
• When a date moves, Autopilot ripples the dependent dates and tells the owners it affected.

Rotas run through the night, and a system that treats 22:00 as “away” is worse than no presence at all. Anything a meeting records has to be safe to keep.

What that is, in the product

• Availability and the presence dot are shift-aware, so somebody mid-night-shift reads as at work rather than asleep.
• Install the redaction extension and the scrub runs before a transcript is stored — on every capture path, including calls made inside Noots.
• Approved time off blocks a slot in find-a-time, and a private calendar event blocks one without ever showing its title.

Twelve people, thirty live projects and one recurring tax: the write-up after every client call, done badly at 6pm by whoever remembers.

What that is, in the product

• The bot joins from the calendar, or the browser extension captures the call with no bot in the room, and the action items land as tasks on that client's project.
• Per-project client invoicing is separate from the internal budget on purpose — “what did this cost us” and “what do we invoice” are different questions and both surfaces say so.
• Every project has its own channel automatically, and the channel and the board link to each other.

Requests arrive with a statutory clock attached, and who did what has to still be provable months after everyone involved has moved on.

What that is, in the product

• The government tools add a records-request exporter, a board that can be published for the public without exposing the workspace behind it, and field-level role controls.
• An append-only compliance log and an audit log record the administrative actions, and the compliance log refuses ordinary edits at the database.
• Every automation firing writes a row — including the ones it skipped, and the exact condition that failed.

Nobody wants to become the person who chases people. The work is real, the plan is in three heads, and the status meeting is the only thing keeping it aligned.

What that is, in the product

• Autopilot on its default setting drafts and proposes: it writes the plan from a five-question brief and waits for you to edit it — and tells you it is waiting, rather than leaving it in a log.
• It books the standup and the milestone reviews itself, writes each agenda from the board, invites the people and sends the bot. That is the status meeting nobody wanted to own.
• A merged pull request proposes closing the task it is linked to, so the board stops drifting from the repository.
• The home screen is a live window on every project and meeting — no status deck, no sitting in on calls, no waiting for a write-up.