Why it matters
Without automation, surveys drift due to inconsistent grading, missing data, delayed reporting, and disputed insurance claims, risking irreversible deterioration and loan cancellations.
From design to safety
Design
AI-informed survey design
The system defines data fields, grading rubrics, sampling logic, and reporting templates based on treatment documentation and PACR standards; it maps relationships between condition grades, sampling triggers, and archival metadata, incorporating imaging channels and sensor readings for cross-checks.
- Survey Form Builder Tablet
- Condition Grading Scale API
- Central Condition Registry
Build
Integrated survey tooling
Data models, schema, and interfaces are wired to capture condition grades, surveys, and samples with versioned records; validation rules enforce mandatory fields and referenced imagery, while preflight checks compare against master templates.
- Survey Form Builder Tablet
- High-Resolution Imaging System
- Central Condition Registry
- Sampling & Audit Toolkit
Operation
Live survey execution
Field teams capture condition data, images, and samples; AI checks completeness, consistency, and auto-suggests missing fields; cross-site data normalization occurs during submission and audit trails attach to each report.
- Survey Form Builder Tablet
- High-Resolution Imaging System
- Condition Grading Scale API
Maintenance
Ongoing data health
Regular calibration of imaging, rubric updates, and data deduplication maintain accuracy and traceability across projects; quarterly audits compare hand-written notes with digital captures and versioned revisions.
- Central Condition Registry
- Condition Grading Scale API
- Survey Reports Generator
Safety
Data governance and ethics
Access controls, COSHH-aligned data handling, and signed-off treatment records ensure compliance and defensibility of decisions; retention policies govern lifecycle of condition data and reports, with anonymization where required.
- Central Condition Registry
- Survey Reports Generator
- Survey Form Builder Tablet
Part by part
- 1
High-Resolution Imaging System
- Watch for
- inconsistent lighting, misalignment, or motion blur across survey images
- AI action
- flag image quality issues and trigger re-capture workflow with guided settings
- 2
Condition Grading Scale API
- Watch for
- grading drift between surveys or conflicting grades
- AI action
- detect anomalies, suggest rubric adjustments, and flag for expert review
- 3
Survey Form Builder Tablet
- Watch for
- missing fields or inconsistent data entry
- AI action
- auto-fill templates, validate required fields, and route for supervisor sign-off
- 4
Sampling & Audit Toolkit
- Watch for
- gaps in sampling coverage or overdue audits
- AI action
- generate sampling plans, assign tasks, and link to condition reports
- 5
Survey Reports Generator
- Watch for
- incomplete metadata in reports
- AI action
- produce standardized drafts and exports with audit trails
- 6
Central Condition Registry
- Watch for
- data duplication and version conflicts
- AI action
- deduplicate entries, unify versions, trigger review workflows
Worked example
Cohort A Painting Condition Review
Illustrative scenario with sample data.
A private collection needs rapid, repeatable surveys across 120 works with standardized documentation and archival-ready metadata for insurance and loan pipelines.
Signal in
survey_id: S2026-041, date: 2026-11-01, project: Cohort A, fields: condition_grade, area, notes, images_ref, sampling_id, report_id, photographer, capture_device, ambient_temp, humidity, operator_id, site_id, sensor_data: temp_aggr, glare_index, reflectance_values
System out
alert: grade drift detected in 12 works; generated draft reports for 12% subset; task list created for site team; recommendations for re-imaging and re-sampling logged; archival-ready metadata linked to master records.
Human sign-off: Head of Conservation reviews flagged items, approves sampling plans, and signs off final reports for archival submission; curator approval is requested for items in loan pipeline; data governance confirms access permissions.
How to roll it out
Phase 1
Weeks 1 to 4: Baseline
Establish core schemas, templates, and governance; calibrate imaging and grading rubrics against PACR standards and include multispectral inputs.
Phase 2
Weeks 5 to 12: Field Pilot
Deploy tablet-based surveys in 2 sites; test data flows to Central Condition Registry and Reports Generator; begin cross-site normalization.
Phase 3
Phase 3: Full rollout
Scale to all projects; enforce data quality checks, automated reporting, and long-term archival integration with version-controlled records.
Where people stay in charge
- All condition grading decisions must be signed off by a qualified conservator prior to archival submission.
- No autonomous treatment alteration; AI informs and records for expert review with documented rationales.
- Access to the Central Condition Registry is restricted to approved roles with versioned audit trails.
Questions owners and crews ask
What data does AI use to drive condition grades in large surveys?
The AI uses standardized rubrics, image-derived features, sampling results, and field notes to propose grades; final grades are confirmed by the conservator-in-charge, with provenance attached.
How is data integrity maintained across many works?
The system tracks unique survey IDs, timestamps, and version histories; deduplication and reconciliation are handled by the Central Condition Registry under supervisor oversight, with audit trails.
Can this automate reporting for loan packages?
Yes, it generates drafts of condition reports and archival-ready metadata; final sign-off remains with the conservator before loan submission, with archival export assets.
What safety controls protect sensitive conservation data?
Role-based access, audit logs, COSHH-aligned handling, and signed off documentation ensure data privacy and compliance with accreditation standards.