Operations
Dependencies as dated items, not as things people said in a meeting
Two teams agree something in a call. Neither writes it down. Three weeks later each is waiting for the other. A worker makes the promise a tracked item.
What the worker does
An AI worker in Polaris tracks cross-team dependencies by turning commitments made in Slack and in Linear into tasks with a named owner and a date, chasing whoever is late, and posting a weekly list of what is blocked and by whom. Slack messages arrive in the Polaris Inbox as prefilled task suggestions that a human accepts. Nothing becomes a commitment without someone clicking accept.
- Connections
- Slack · Linear · Google Calendar
- Every dependency has
- An owner, a date, a blocked item
- Suggestions become tasks
- Only when a human accepts
The most expensive delay in a company is two teams politely waiting for each other
It starts as a reasonable exchange. Design says they will have the final assets once the copy is signed off. Marketing hears that as a commitment for next week. Nobody writes a date. Both teams move on to other work and are entirely certain they are not the blocker.
The delay only becomes visible when a launch date is missed, at which point it is a conversation about blame rather than about a missing dependency.
The intervention that works is unglamorous: someone writes the promise down with a name and a date on it, and asks about it before the date arrives. That is a worker's job description.
How a commitment becomes a tracked item
The Inbox catches what your tools hear. Signals become suggestions, never silent tasks.
A Slack message arrives as a prefilled task suggestion
With the workstream, lane, labels and owner already set. One click makes it a real task, and nothing is created without that click.
The worker adds the missing date
Or asks for it. A dependency without a date is a wish, and the worker's first question is always when.
The blocked item is linked
Knowing that something is late matters much less than knowing what stops moving because of it.
The chase runs on a schedule
Before the date, not after it. A dependency chased the morning it comes due is often still recoverable.
The weekly dependency report
One comment, one shape, every week, so the pattern over a quarter is readable.
| Section | What it lists | The conversation it starts |
|---|---|---|
| Due this week | Commitments with a date inside the window and their owners | Is anything about to slip |
| Overdue | How many days late, and what it is blocking | Is this the same team every week |
| No date agreed | Commitments that were made without one | Which of these are real and which were politeness |
| Cleared since last week | What actually moved | The part teams never get told, and it matters |
| Recurring blockers | The same dependency appearing across weeks | This is a structural problem, not a chasing problem |
Why this lives in one workspace
Cross-team work is where tool sprawl does the most damage, because each team keeps its own version of the truth.
3
Focus horizons every team shares
Today, This week, Next 30 days
1
Click to turn a Slack signal into a task
The suggestion arrives prefilled
$0
Cost per additional person or team
No seats, unlimited humans
Questions people ask
+What if the other team uses Linear and we do not?
Give the worker the Linear connection. It reads what is there and tracks the dependency in Polaris, so the two teams do not have to agree on a single tracker before they can agree on a date.
+Does every Slack message become a task?
No, and that distinction is deliberate. The Inbox turns a signal into a suggestion with the fields prefilled, and a human accepts it. A tool that silently creates tasks from chat produces a backlog nobody trusts within a fortnight.
+Who should own the dependency report?
One named person, usually in ops, with the report as an assigned task rather than a message in a channel. Reports addressed to everyone are read by no one.
+Can the worker escalate?
It can follow the escalation path you write in the SKILL.md: chase the owner, then their manager after a set number of days, then flag to the ops lead. Because the file is plain text, everybody can read the rules being applied to them, which is what makes escalation acceptable.
Related
Operations work with an AI worker doing the chasing
Ops is the function that holds the seams together. Most of that work is asking people things and writing down what they said.
Getting a process out of one person's head
The process exists. It is in someone's habits and in six months of Slack threads. A worker interviews and reads until it is on a page.
Every commitment from the meeting, owned and dated
The decisions were good. The problem is the eleven commitments that existed only in a document nobody reopened.
The two hours before planning that nobody schedules
Carry-over, gaps and blockers assembled the day before, so the meeting is about commitment.
Hire an AI project coordinator
The unglamorous half of running projects, done every week without anyone having to be the person who nags.
Connect Linear to Polaris
A Linear API key carries one person's visibility, so the account you make it on decides what every worker can see.
Connect Slack to Polaris
Six scopes, no access to direct messages, and a bot that only reads the channels somebody invited it into.
Focus lane
A commitment for a horizon, not a filter over everything you have.