Why it matters
Without it, misinterpretation of object types, missed deadlines, and inconsistent client communications risk delayed or disputed insurance claims and damaged client trust.
From design to safety
Design
Triage design for intake
Define data fields, decision trees, risk flags, and escalation criteria for enquiries, including object type, provenance codes, photos, urgency, and location. Establish rules for referrals and preliminary quotes with audit trails.
- enquiry emails
- client intake forms
- quotes and invoices
Build
Enquiry workflow setup
Integrate intake channels, configure AI classifiers with standardized metadata, and create templates for responses, quotes, and site-work planning, all with access controls and versioned templates.
- photography setup
- referral records
- site visit reports
Operation
Real-time intake processing
AI analyzes new enquiries, assigns conservators, schedules preliminary assessments, flags follow-up items, stores structured metadata for the project file, and creates a preliminary task list for the team.
- enquiry emails
- photography setup
- quotes and invoices
Maintenance
Data hygiene and calibration
Periodic review of AI rules, updating taxonomy for object types, provenance, and owner data; verify referrals and updated contact records; run quarterly QA on routing and quotes.
- referral records
- site visit reports
- client intake forms
Safety
Ethics and client data safety
Ensure compliant handling of sensitive owner information, with access restricted to accredited staff, documented consent workflows, encryption in transit at rest, and breach response procedures.
- enquiry emails
- quotes and invoices
- client intake forms
Part by part
- 1
enquiry emails
- Watch for
- New inquiries with vague object descriptions, missing photos, unusual urgency, conflicting ownership notes, or ambiguous provenance that hinder routing.
- AI action
- AI classifies urgency levels, identifies incomplete data, flags potential conflicts, and routes to the most appropriate conservator queue with audit log.
- 2
client intake forms
- Watch for
- Incomplete forms, conflicting ownership details, missing provenance notes, unverified contact data, or inconsistent object naming that impede quotes.
- AI action
- AI prompts data completion, creates a preliminary client profile, assigns a case ID, and enforces mandatory fields before submission.
- 3
photography setup
- Watch for
- Low-resolution images, inconsistent lighting, missing scale, lack of multi-angle coverage, or absent measurements that compromise object reading.
- AI action
- AI requests high-resolution images, standardized scales, multi-angle capture, and embedded metadata for every image before analysis.
- 4
referral records
- Watch for
- Multiple referrals for similar object types or conflicting recommendations affecting triage.
- AI action
- AI consolidates referrals, cross-checks provenance, and suggests best-fit conservator or specialist, with confidence scores and source tagging.
- 5
site visit reports
- Watch for
- Inconsistent condition notes, missing measurements, safety concerns, or undocumented alterations that affect planning.
- AI action
- AI flags data gaps, suggests a standardized site-visit template, links to the project file, and records visit time and observer identity.
- 6
quotes and invoices
- Watch for
- Quote discrepancies, unclear scope, or delayed invoicing that risk cash flow and client trust.
- AI action
- AI generates standardized preliminary quotes, attaches scope definitions, lists materials and treatments, and tracks approval status with version history.
Worked example
Private inquiry triage for a 19th-century cabinet
Illustrative scenario with sample data.
A private collector submits an enquiry with a damaged enamel panel, 3 photos, a request for on-site assessment, and urgency due to upcoming exhibition.
Signal in
enquiry_id: EQ-472, owner: Smith Family, object_type: cabinet, photos: 3, urgency: high, location: Paris, site visit: requested, prelim_quote_needed: yes, provenance: documented
System out
Alert: high-urgency intake with enamel panel damage. Assign to senior conservator, schedule site visit, request additional high-res photos, generate preliminary quote template, log enquiry in CRM.
Human sign-off: Head of conservation reviews AI routing, approves site visit, and signs off on the preliminary quote.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline intake mapping, configure classifiers, develop templates, and test data routing with mock enquiries.
Phase 2
Weeks 5 to 8
Live intake with monitored performance, refine categories, train staff on templates and escalation paths.
Phase 3
Weeks 9 to 12
Full governance setup, audits, and rollout to all new client engagements and ongoing site work planning.
Where people stay in charge
- All decisions affecting client acceptance, pricing, scope, or loan/insurance implications must be signed by the accredited conservator or head of conservation.
- Data privacy and consent must be verified by a designated Data Protection Officer before any client data is stored or processed.
- Only trained personnel can override AI-generated routing or quotes; escalation paths must be documented and auditable.
Questions owners and crews ask
What types of enquiries can the AI triage handle?
The AI triage handles written enquiries, photographs, urgency signals, provenance notes, and structured metadata to route to the appropriate conservator and draft initial communications.
How does the system protect client confidentiality?
Access is role-based, with encrypted storage, consent workflows, and audit trails ensuring only authorized staff view sensitive owner information.
Can the AI suggest a site visit date?
Yes, it analyzes staff availability, travel constraints, venue access, and urgency to propose feasible site-visit windows for approval.
What happens if data is missing or inconsistent?
The AI flags gaps, prompts for completion, logs reason codes, and routes to a human for corrective data entry and reconciliation.