SLA Monitor

Tracks commitments and SLAs made to customers -- from formal contractual obligations to informal promises made in calls and emails. Surfaces upcoming deadlines, flags breaches before they happen, and maintains an accountability record. Part of the Lifecycle Orchestration pillar.

This is a tracking and alerting skill. It monitors commitments against deadlines and surfaces deviations. It does not make commitments, renegotiate deadlines, or communicate with customers about breaches. Those are human decisions.

When to Run

• Continuous: Monitors all active commitments against their deadlines
• Triggered: When a new commitment is logged (manually by CSM or extracted from call transcripts/meeting notes)
• On-demand: When a CSM requests the commitment status for a specific account

Commitment Types

Core Execution Logic

Step 1: Capture Commitments

Commitments enter the system through three channels:

Automated capture:

• Support ticket SLAs (response time, resolution time) from support platform
• Contractual SLAs from contract terms in CRM
• Action items extracted from call transcripts (if NLP extraction is available)

Semi-automated capture:

• Post-call summary parsing: when the CSM generates a call summary, the skill identifies action items with owners and deadlines
• Email parsing: detect commitment language ("I will," "we will," "by [date]") in outbound CSM emails

Manual capture:

• CSM logs a commitment directly (account, description, owner, deadline)

Step 2: Track Against Deadlines

For each active commitment:

1. Calculate days remaining to deadline
2. Classify status:

Step 3: Alert on Upcoming Deadlines

Step 4: Maintain Accountability Record

For each account, maintain a running record of all commitments:

• Total commitments made (trailing 90 days)
• Completion rate (% delivered on time)
• Average time to fulfil
• Most common breach type (if any)
• Overdue items (currently open)

This record feeds into QBR preparation, renewal assessment, and escalation context.

Output Format

Active commitment record (per account):

{
  "account_id": "string",
  "active_commitments": [
    {
      "id": "commit-001",
      "description": "Send benchmark report comparing account usage to segment peers",
      "owner": "Jane Doe (CSM)",
      "created": "2026-03-07",
      "deadline": "2026-03-14",
      "status": "on_track",
      "source": "QBR call on 2026-03-07",
      "days_remaining": 4
    },
    {
      "id": "commit-002",
      "description": "Engineering fix for API latency issue",
      "owner": "Engineering (via escalation)",
      "created": "2026-02-28",
      "deadline": "2026-03-10",
      "status": "at_risk",
      "source": "Escalation TICKET-4521",
      "days_remaining": 0
    }
  ],
  "accountability_summary": {
    "commitments_90d": 12,
    "completed_on_time": 10,
    "completion_rate": 0.83,
    "currently_overdue": 0,
    "currently_at_risk": 1
  }
}

CSM commitment dashboard (weekly):

{
  "csm": "Jane Doe",
  "week_of": "2026-03-10",
  "due_this_week": 4,
  "at_risk": 1,
  "overdue": 0,
  "completion_rate_30d": 0.91,
  "accounts_with_overdue_items": [],
  "top_priority": {
    "account": "Acme Corp",
    "commitment": "Engineering fix for API latency",
    "deadline": "2026-03-10",
    "status": "at_risk"
  }
}

Handoff to Human

Confidence and Limitations

• High confidence for deadline tracking and alerting -- deterministic calendar logic
• Medium confidence for automated commitment capture from call transcripts and emails -- NLP extraction may miss or misinterpret commitments
• Low confidence for commitments that were made verbally and never logged -- the system can only track what it knows about
• Cannot assess whether a completed commitment was delivered to the customer's satisfaction (only that it was delivered)
• Cannot prioritise between competing commitments when resources are limited -- that is a human judgment call

Dependencies

Required:

• CRM API (commitment logging, account context)
• Calendar or task management integration (deadline tracking)

Strongly recommended:

• Support platform (for SLA tracking on tickets)
• Call transcript integration (for automated commitment extraction)
• Email integration (for commitment language detection)
• bi-account-brief (for customer context when a breach occurs)

References

• references/commitment-capture.md -- Patterns for extracting commitments from different sources