Command tier only ($50K/yr)
200 interfaces with weekly incidents means 200-800 engineer-hours per year diagnosing and fixing at 2am. With Command, the system heals autonomously and the engineer reviews next business day. Messages keep flowing while the fix is generated and applied.
Self-Healing Pipeline
Inbound Message → Parse + Validate
↓ PASS → Deliver normally
↓ FAIL
Deviation Detection (compare against vendor profile)
↓ Active rule exists? → Apply rule, deliver, log
↓ No rule
Generate Micro-ETL Remediation Rule
↓
Playbook: Autonomous?
├── YES → Apply + deliver + queue review
└── NO → Queue for 1-click approval + hold message
Deviation Types
| Type | Example |
|---|
| Field format change | NDC-10 → NDC-11 format |
| Segment reorder | PV1 before PID (expected PID before PV1) |
| Missing segment | EVN segment dropped after vendor upgrade |
| New field | New OBX observation type added |
| Value set change | Gender code “MALE” instead of “M” |
Rules are micro-ETL transformations applied to incoming messages:
| Rule Type | Description |
|---|
| FieldTransform | Map field values (e.g., “MALE” → “M”) |
| SegmentReorder | Reorder segments to expected sequence |
| FieldDefault | Insert default value for missing required fields |
| ValueMap | Map between code systems |
| FormatNormalize | Normalize date/ID formats |
Review Queue
Engineers disposition healed messages through the review queue:
- Approve permanently — Rule becomes permanent
- Accept temporarily (TTL) — Rule expires after N days, auto-escalates if unresolved
- Reject and investigate — Rule removed, held messages released for manual review
- Enquire with vendor — Pre-drafted email/ticket sent to vendor contact
Playbook Configuration
playbooks:
- interface: surescripts-pharmacy
self_heal: true
approval_mode: autonomous
default_ttl: 7d
max_concurrent_rules: 5
escalation:
- level: 1
notify: ["[email protected]"]
channel: slack
after: 0m
- level: 2
notify: ["[email protected]"]
channel: email
after: 4h
vendor:
name: "SureScripts"
contact: "[email protected]"
ROI
Without Command: 200 interfaces with weekly incidents = 200-800 engineer-hours/year diagnosing and fixing at 2am.
With Command: System heals autonomously, engineer reviews next business day = ~17 engineer-hours/year. The $50K subscription pays for itself.