> ## Documentation Index
> Fetch the complete documentation index at: https://docs.apollospace.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Operations
> Triage, tracking, routines, and tasks that move between areas without anyone pushing. The agent as the operator who never drops a thread.
It's 9:12 on a Tuesday. Eleven new messages landed across three channels, two cards have sat untouched for five days, someone promised a reply "by yesterday" in a meeting that's already a memory — and the operation only moves because a person is, right now, re-reading everything to remember who owes what.
That work of **never dropping a thread** is what leaves your plate. The agent becomes the operator who triages the inbox, chases what stalled, turns what was agreed into a task with an owner — and raises its own hand when a deadline is about to slip.
Operations in Apollo is [Athena](/en/agents/athena), the
organization's chief of staff, doing the coordination work nobody sees —
but everyone feels when it stops happening.
## What changes for you
Before, "coordinating" meant one person holding the state of the operation in their head: who owes what, what's stuck, what's due today. That state lived fragile — one day off and it was gone.
With operations agents, the state belongs to the house. Triage happens when the message arrives, not when someone has time. Chasing stale work is a routine, not an act of willpower. And what was about to slip finds you first — not after the customer complained.
## Four operations flows
Each flow follows the same shape: a **trigger** fires, the agent **pulls context** from the [Company Brain](/en/features/brain), **decides and uses real tools**, the **sensitive stuff waits for you**, and everything **becomes memory**.
### 1. Inbox triage → routed to the right board and owner
```mermaid theme={null}
flowchart LR
A[Message or ticket arrives
WhatsApp, email, chat] --> B[Athena pulls context
who, history, what was agreed]
B --> C[Classifies and routes
creates a card on the right board + owner]
C --> D{Changes a
commitment?}
D -->|No| E[Creates the task and logs it]
D -->|Yes| F[Proposes and flags you]
E --> G[Becomes memory]
F --> G
```
A message comes in over [WhatsApp](/en/integrations/whatsapp), email, or [chat](/en/features/chats). Athena reads it, recognizes who it's from and what it's about (pulling history from the [Brain](/en/features/brain)), and **creates a [task](/en/features/boards) on the right board, with the right owner** — support, sales, operations. Simple questions she answers; anything that needs a human decision becomes a card with context attached, not a "did anyone see this?".
The practical difference: the inbox doesn't sit waiting for someone to
have time to triage. It **arrives already classified and addressed**,
with the history attached.
### 2. A daily routine that chases what stalled
Every morning, a [routine](/en/features/rotinas) wakes Athena. She sweeps the [boards](/en/features/boards), finds **tasks stuck for days** and **overdue items**, and acts: comments on the card asking the owner for an update, reopens what was forgotten, and assembles a summary of what needs attention. What she can't unblock alone, she **hands you as a notification** — not as one more card lost in the pile.
| Trigger | Athena does | You get |
| ------------------------ | ------------------------------------ | ------------------------- |
| Card stuck 5 days | Comments asking the owner for status | Nothing (handled) |
| Due yesterday | Flags it, raises the priority | Inbox notification |
| Task with no owner | Suggests the owner from context | A proposal to approve |
| Blocked by a third party | Logs the blocker | A summary of what's stuck |
### 3. Status-chasing across teams
That question you ask ten times a week — *"where's the thing for customer X at?"* — Athena starts asking for you. She cross-reads the [updates](/en/features/boards) on cards across teams, spots what moved and what went quiet, and **chases status from whoever's holding the ball** via a card comment or a [chat](/en/features/chats) message. When someone answers, she updates the card and closes the loop — without you becoming the central point every piece of information has to pass through.
### 4. A meeting ends → it becomes tasks with owners
The meeting wrapped and six commitments are left hanging. You send the notes to Athena (or she already has the recap in the [Brain](/en/features/brain)). She **turns each commitment into a [task](/en/features/boards)** — clear title, likely owner, due date — groups them on the right board, and hands you the list for a quick look before creating them. What used to evaporate between the end of the call and the end of the day now starts life as trackable work.
Every action item from the meeting becomes a task, with the context
of what was said attached.
Athena proposes the owner and date based on who was in the
conversation and the area's history.
A glance, a tweak if needed, and the cards land on the boards already
addressed.
### 5. An SLA/deadline watcher that raises its own hand
This is the flow that turns "reactive" into "in control." A [routine](/en/features/rotinas) watches deadlines and SLAs in the background. When an item is **about to slip** — not after — Athena fires a [proactive notification](/en/features/rotinas) to the inbox and the bell: *"Customer Y's ticket is due in 2h and still has no reply."* You decide what to do with room to maneuver, instead of finding out from an annoyed customer.
Proactive notifications are the agent **raising its own hand**. It isn't
you checking a dashboard — it's the operation warning you before
something drops.
## What stays human
**The agent operates, but it doesn't change commitments on its own.**
These actions it only **proposes** — you approve before anything leaves:
* **Changing a promise made to a customer** — rescheduling a delivery,
renegotiating an agreed deadline, reopening a scope. The agent flags
the risk and proposes; the word to the customer is yours.
* **Anything touching money or contracts** — payment, refund, a
commercial proposal, a signature. Always human.
* **Heavy external communication** — a delicate reply to an unhappy
customer goes out as a draft for you to review, not straight.
* **Destructive actions** — deleting, bulk-archiving, closing something
irreversibly goes through you.
Triaging, chasing status, creating and moving internal tasks, raising
alerts — the agent handles. Changing what was promised to someone
outside — that's yours.
## Where to start
Start with one inbox — the support WhatsApp, or an email. Let Athena
triage and route for a week before wiring in more channels.
A daily [routine](/en/features/rotinas) that sweeps boards and chases
what stalled. It's the one that returns time fastest — and runs
inside your [Stars](/en/billing/stars) budget.
At first the agent proposes everything. As it gets triage and chasing
right, you release more — trust is a ratchet per action type, not a
leap.
## Next steps
The daemon that makes the stale-work sweep and the SLA watcher run on
their own.
Where triage becomes a card and status-chasing happens.
Talk to Athena — the chief of staff is one keystroke away.
When operations becomes a plan: goals that turn into tasks and clear
priority.
Meet the organization's operator who runs all of these flows.