Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.teeem-ai.com/llms.txt

Use this file to discover all available pages before exploring further.

Without anyone asking each time, Teeem AI builds reports on a schedule and proactively alerts when something significant changes.

​
Four patterns

Daily executive brief

Every weekday morning, yesterday’s key changes in a single message.

Customer risk report

Every Monday, accounts at risk + recommended actions.

Monthly operations report

Last business day of each month, company-wide ops indicators consolidated.

Anomaly alerts

Real-time, fires the moment a threshold is crossed.

​
1. Daily executive brief

Every morning, before the CEO is at their desk, yesterday’s key signals arrive in a single message. 
# Schedule definition (set in admin console)
schedule:
  cron: "0 7 * * 1-5"   # Weekdays 07:00
  timezone: "Asia/Seoul"
  agent: "executive-brief"
  channel: "@kim_ceo"   # CEO DM
  prompt: |
    Summarise yesterday's (last business day) key operational changes.
    - Revenue, orders, inventory
    - Customer CS issues (P1/P2 only)
    - Executive calendar (where the CEO is attending)
    - Yesterday's social/news mentions of the company
    Up to 8 sentences. Attach one chart.
Every weekday at 07:00 sharp, the CEO’s Slack DM receives: 
Yesterday (Thu, 2026-05-01) brief

Revenue ₩142M (+8% vs average). Orders: 47 normal.
Inventory turnover: within range.
1 P1 CS issue — account A shipping delay (resolved).
Exec calendar: 14:00 meeting with B Co. (deck in #c-sales).
No social mentions.

[Chart: Daily revenue, last 30 days]

─── Sources ───
📄 sales_orders model
📄 cs_tickets model
📄 hr_calendar

​
2. Customer risk report

Every Monday morning, accounts showing risk signals get summarised. Example signals:
  • Revenue dropped 30%+ vs the last 4-week average
  • Payment day slipped 2+ weeks beyond average
  • CS issues 3× the same-quarter average
  • N days since last order
schedule:
  cron: "0 9 * * 1"     # Mondays 09:00
  agent: "account-risk"
  channel: "#sales-leadership"
  prompt: |
    Pull this week's top 5 at-risk accounts.
    - Reasons (which signals are firing)
    - Recommended actions (sales meeting, payment terms, CS priority)
    - Monitoring points for next week
The report is posted to the channel, and an account insight card is generated for each flagged account.

​
3. Monthly operations report

Last business day of each month at 18:00. Company-wide indicators consolidated. Sections included:
  • Revenue, orders, inventory (this month vs last vs same-period last year)
  • Per-department activity summary
  • New / churned accounts
  • HR / approval / sign-off throughput
  • Risk signals (at-risk accounts + CS trend)
  • Recommended priorities for next month
A PPTX report + Excel dataset + Slack summary all generate together. Suitable for executive review as-is.

​
4. Anomaly alerts

Real-time monitoring. Threshold crossings fire immediately. 
alerts:
  - name: "P1 CS issue"
    condition: "cs_tickets where priority='P1' and created_at > now() - 15min"
    severity: high
    channel: "#cs-on-call"
    template: "New P1: {customer} - {summary}"

  - name: "Inventory threshold"
    condition: "inventory where qty < min_threshold"
    severity: medium
    channel: "#ops"
    template: "{sku} stock at {qty} (threshold {min_threshold})"

  - name: "Payment overdue"
    condition: "invoices where due_date < today and status != 'paid'"
    severity: medium
    channel: "@finance_lead"
    digest: daily   # Daily summary (avoid alert spam)
Delivery channels:
ChannelPurpose
Slack channelTeam-level alerts
Slack DMIndividual on-call
SMSField staff without mobile data
EmailExternal partners (with approval)

​
Digest vs immediate

If the same alert fires too often, people stop noticing. Two modes solve this:
  • Immediate (severity=high) — P1 incidents, system-down events, anything truly urgent
  • Digest (digest=daily/hourly) — minor threshold misses bundled into one daily (or hourly) message
Tune frequency in Operations → Alert policy.

​
Creating a scheduled job

Admin console → Scheduled jobs → New schedule → fill in:
  1. Name (e.g. “Weekly KPI digest”)
  2. Cron schedule
  3. Timezone
  4. Agent or inline prompt
  5. Output channel
  6. Active toggle
The skill library ships with schedulable templates so first-time tenants can stand up their first schedule in five minutes.

​
Safety rails

  • Channel permission verification: the bot must be a member to post
  • Data permissions: scheduled jobs run within the defining user’s permissions
  • Failure notifications: failed runs DM the configured user
  • Rate limit: identical alerts within 5 minutes are auto-debounced
  • Audit trail: every scheduled run is recorded in the audit log

​
Next

Use cases

How the outputs of scheduled jobs are typically consumed.

Playbook: Weekly report

Step-by-step setup of a weekly report automation.