From equipment arrival to an expert-reviewed assessment package.

AI that shows its evidence

• Multimodal intake: images, nameplates, inspection photos, PDFs and structured measurements.
• RAG grounded only in approved manuals, historical jobs and configured sources — citations shown in-app.
• Never invents tolerances, specs, parts or test results. Unknown values stay UNKNOWN.
• Equipment-specific reasoning templates for pumps, motors, gearboxes, compressors and blowers.
• Historical-job similarity search for scopes, hours, parts, pricing and actual outcomes.
• AI suggestions are editable drafts — technical release requires an authorised human.
• Deterministic unit conversion and money arithmetic performed outside the language model.
• Evaluation dashboard: suggestion acceptance, estimate error, unsupported-output flags and model version.

Security, governance & guardrails

• Tenant isolation, role/site scoped RBAC, encryption in transit and at rest, full audit logging.
• Technical approval and commercial approval are separate permissions.
• No autonomous declaration that equipment is safe, compliant or fit-for-service.
• NDT, tolerance and OEM requirements remain subject to qualified human review and applicable standards.
• No ISO/SOC/GDPR certification is claimed; the architecture is designed toward enterprise expectations.
• Configurable retention, export and deletion; privacy-by-design and customer data segregation.

What this is not

• Not a generic CMMS / EAM / ERP.
• Not predictive maintenance or sensor monitoring.
• Not full shop-management software.
• Not a generic CPQ tool.
• No auto-approval of engineering decisions.
• No sensors, IoT, RFID or integration required for first value.