Why it matters
Without AI welfare scoring, misreads of health signals can lead to unnoticed distress, delayed interventions, failed licence inspections, and ad hoc action plans that miss systemic risks within enclosures and tethering regimes.
From design to safety
Design
Design AI welfare scoring framework
Define data streams, sensors, and record layouts for Five Domains, enclosure scores, and tethering times; establish audit checklists, governance, and data retention policies across licenses and welfare protocols.
- flying grounds telemetry
- mews tethering hardware
- enclosure scoring tablet
- welfare audit forms
Build
Build data capture and rule engine
Install weatherproof telemetry nodes at flight lines, install tethering sensors on each post, configure secure data pipelines, compile domain rules, and set alert thresholds with red/amber/green codes.
- telemetry sensor node
- tethering hardware assembly
- enclosure pressure/temperature sensor
- audits app interface
Operation
Operate continuous welfare scoring
Ingest live sensor streams from perch sensors, motion counters, environmental sensors, and tethering monitors; run domain scores every 15 minutes; auto-suggest welfare actions and schedule vet checks.
- perch telemetry
- enclosure scoring device
- tethering line strain gauge
- audit mobile app
Maintenance
Maintain data quality and models
Schedule quarterly sensor calibration, test data pipelines on a staging environment, annotate edge cases, refresh domain thresholds for seasonal cohorts, and archive historical risk trends.
- sensor calibration kit
- data warehouse
- health checks dashboard
- maintenance log book
Safety
Safeguard welfare governance
Enforce role-based approvals, mandatory licensing cross-checks for actions, and maintain an audit trail with signatures from head falconer and avian vet for all flagged welfare risks.
- risk register
- cloud backup system
- caretaker training records
- biosecurity log
Part by part
- 1
Telemetry sensor node
- Watch for
- Sensor drift or intermittent connection in outdoor weather; failure reduces data fidelity.
- AI action
- Detect drift, trigger calibration; alert supervisor.
- 2
Tethering hardware assembly
- Watch for
- Wear on tether lines, fraying, or jess wear at attachment points.
- AI action
- Flag risk, pause tethering, schedule replacement.
- 3
Enclosure scoring device
- Watch for
- Changes in substrate cleanliness, perch height, clutter, or escape routes.
- AI action
- Compute enclosure score, issue cleaning or reorganization task.
- 4
Audits app interface
- Watch for
- Incomplete welfare audit entries, missing signatures, or overdue reviews.
- AI action
- Prompt auditor, generate compliance reminder.
- 5
Perch telemetry
- Watch for
- Inadequate perch spacing, temperature hotspots, or wear.
- AI action
- Warn keeper, adjust roosting plan.
- 6
Audit and records cabinet
- Watch for
- Missing Article 10 paperwork or inconsistent tagging
- AI action
- Flag licensing risk, request document update.
Worked example
Welfare scoring check before a new display season
Illustrative scenario with sample data.
A centre with 22 birds across mixed-species yards plans a high-traffic seasonal display and must verify welfare risks across the entire yard and transport routes.
Signal in
domainScores: {nutrition: 0.88, movement: 0.81, hydration: 0.94, social: 0.62, environment: 0.75}; tetherTime: 2h 15m; enclosureScores: 0.74; lastAudit: 2026-11-01; birds: 22; species: {Saker:3, Harris:5, Red-tailed:7, Falcon:7}
System out
alert: high risk in social domain for Harris hawks; action: reduce group size to 2 per flight; assign welfare plan; report to avian vet; update enclosure plan; log in licensing package.
Human sign-off: Head falconer reviews AI alert, confirms action with senior keeper, signs off on revised flight groups, welfare plan, and updated enclosure maps for the display season.
How to roll it out
Phase 1
Weeks 1 to 2
Baseline data capture and sensor checks; establish scoring thresholds and governance across all Five Domains for licensing alignment.
Phase 2
Weeks 3 to 6
Pilot AI rules on a subset of enclosures and tether lines; calibrate alerts with escalation paths and vet sign-off rules.
Phase 3
Weeks 7 to 12
Full deployment; integrate with licensing and audit workflows; formalise reporting cadence and cross-department dashboards.
Where people stay in charge
- All welfare risk decisions must be signed off by the head falconer or licensed avian vet, who has final responsibility for animal welfare.
- No AI-generated action plan may override mandatory licensing requirements or legal reporting without human review and sign-off.
- Action plans must be documented in the welfare management system with timestamps and responsible keeper IDs.
Questions owners and crews ask
How does AI welfare scoring integrate with licensing audits?
The AI system generates welfare risk scores and flagged items, which feed into the licensing audit package; licensed officers review the scores, verify records, and sign off on corrective actions.
What data sources drive the Five Domains scoring?
Data sources include nutrition and hydration readings, activity and movement data, environmental conditions, social groupings, and sensory or medical indicators collected by sensors and keeper observations.
Who validates AI alerts before actions are taken?
Head falconer and avian vet validate and approve high-risk alerts; routine alerts are reviewed by senior keeper teams before action is logged.
What records are needed for ongoing compliance?
Digital welfare scores, tethering durations, enclosure scores, audit checklists, Article 10 certificates, and action plans maintained in the welfare records system.