Responsible AI & Technology Governance

EntireFM leverages artificial intelligence and machine learning to optimize facilities management logistics, synthesize building telemetry, and assist our technical team. We operate under clear principles: AI must be transparent, secure, bounded by safety limits, and subject to human oversight.

We deploy AI assistance across specific, bounded facilities management workflows:

• Service Request Triage: Categorising incoming maintenance requests by trade (HVAC, Electrical, Plumbing) and extracting asset codes.
• Priority Recommendation: Suggesting urgency tiers (P1-P4) based on safety keywords and client SLA terms for dispatcher confirmation.
• PPM Matrix Drafting: Synthesising asset lists and SFG20 guidelines to draft recommended planned maintenance schedules.
• Contractor Allocation Suggestion: Evaluating geographic proximity, trade competencies, and active SSIP accreditations to recommend vetted contractors.
• Document Data Extraction: Parsing technical certificates, EICRs, and service sheets to extract expiry dates and remedial actions for compliance lead verification.
• Predictive Telemetry Monitoring: Flagging anomalies in sensor and energy data indicating potential equipment deterioration.

To protect our clients, contractors, and building occupants, the following actions are strictly prohibited from sole automated execution:

• No Automated Contract Cancellation or Penalties: AI cannot terminate contracts or assess financial damages.
• No Unsupervised Safety Risk Acceptance: AI cannot override safety exceptions, asbestos warnings, or statutory inspection deficits.
• No Sole Automated Contractor Suspension: Contractor onboarding and disciplinary actions require human compliance review.
• No Binding Quotations: Commercial proposals are formally reviewed and issued exclusively by authorised human surveyors.

Every AI agent in our platform operates under strict autonomy policies (OFF, MANUAL, ASSIST, or CONTROLLED_AUTO). Confidence thresholds and spend guardrails automatically trigger escalation to human supervisors whenever ambiguity arises.

AI outputs are probabilistic recommendations. We mitigate algorithmic bias by evaluating models against objective technical standards (qualifications, geographic radius, verified certifications) rather than subjective profiling.

Under UK GDPR Article 22 and the Data (Use and Access) Act 2025, if an AI recommendation affects your service request, quote, or contractor allocation, you have the right to request an explanation, express your point of view, and demand a full manual review by an EntireFM manager.