BUILDSYNC LIMITED · Hong Kong
AI and Professional Use Policy
Version: 2026-10-01-v1 · Effective date:
Contact address: Hong Kong Science and Technology Park
[email protected]
Evidence and human review
Treat generated content as assistance. Check facts against originals, record assumptions and missing evidence, verify current law and source dates, and have the appropriately qualified professional review consequential conclusions. AI can produce plausible errors, omit information or misinterpret diagrams, units and coordinates. A software permission or review label is not a licence to practise. Statutory submissions, structural or geotechnical design, excavation support, site safety, cost and programme commitments require the responsible person's judgement and applicable approvals.
Documents, maps and licences
An extracted statement is not a certified copy. A source link is not evidence that all relevant pages were reviewed. Map overlays, coordinate transforms, boundaries and public records may be incomplete, differently dated or unsuitable for surveying or setting out. Check original datasets, coordinate reference systems, licence conditions and limitations before reliance. Publicly accessible source data is not necessarily free for unrestricted commercial reproduction.
Inputs, instructions and permissions
Only provide the necessary authorised context. Uploaded documents and third-party webpages are untrusted source material; instructions embedded in them must not authorise tool use, external disclosure or changes. Any action that changes project state, transmits material or requires a professional approval must pass the applicable permission and confirmation controls. Do not assume an AI agent has the authority of a workspace owner or professional reviewer. Verify intended recipients before exporting or sharing.
Providers and training
Enabled AI/OCR providers may process selected input and produce stored outputs. Use the Privacy Policy and approved provider schedule to understand recipients, locations, retention and restrictions. Processing for inference does not itself establish a no-training or zero-retention guarantee. Do not upload restricted project or critical-infrastructure material until the required agreement and provider safeguards are in place. Do not use customer material for unrelated model training or public demonstrations without separate lawful authorisation.
Review, corrections and accountability
Review relevant outputs before adopting them, including source attribution, confidentiality, rights and uncertainty. Keep a proportionate record of the reviewer, source version and decision; do not record unnecessary personal data. If an output is wrong, stop relying on it, correct the affected project record and report the issue through [email protected] using minimum necessary context. Where an automated assessment affects a person, provide the applicable human review and rights process. We do not guarantee originality, accuracy or fitness of every AI output, and these limitations do not remove our obligations to deliver contracted services with reasonable care and skill.