What problem does it solve?
Generic AI-generated construction safety paperwork (RAMS) is often vague, lacks site-specificity, ignores the hierarchy of controls, leaks personal data, and fails to meet regulatory requirements like UK CDM 2015 or India BOCW, leading to regulator rejection and worksite safety gaps.
Core Features & Use Cases
- Structured Intake Enforcement: Forces explicit input for construction activity, site context, and ordered work sequence, refusing to proceed on vague requests to eliminate generic, copy-paste safety paperwork.
- Bidirectional RA-MS Cross-Referencing: Ties every method statement step to its corresponding risk assessment hazards, and flags any unaddressed risk assessment hazards to eliminate gaps between planned work and risk controls.
- Hierarchy of Controls Compliance: Enforces ranking of controls from elimination to PPE, blocking unjustified PPE-only treatments that fail to adequately mitigate construction risks.
- Jurisdiction-Aware Regulatory Grounding: Automatically cites applicable construction law (UK CDM 2015, India BOCW, US OSHA, EU OSH) and ISO 45001 standards, with mandatory state detection for Indian sites to avoid incorrect form citations.
- Mandatory De-identification: Scrubs all personal and health data from inputs before drafting, with a controlled exception for user-supplied competent persons required for sign-off records.
- SME Review & Branded Output: Runs a mandatory domain expert review before delivery, and produces branded DOCX/PDF reports via a shared report engine.
Use case: A UK principal contractor erecting a mobile access tower to replace cladding on an occupied building with an adjacent public footpath can use this skill to generate a compliant, sequenced RAMS with activity-specific rescue plans, named competent persons, and empty crew sign-off rows in minutes.
Quick Start
Use the rams-builder skill to generate a compliant, site-specific RAMS for your construction works by providing the exact activity, site details, and ordered work sequence when prompted.