How we use AI
Bacland Limited uses AI extensively to work faster and more accurately: in surveys and measurement, design and drawing preparation, planning research, cost appraisal, document drafting, software and day-to-day project administration.
Our AI tools are provided by Anthropic, OpenAI and Google, used only on business accounts whose terms keep our data out of model training. AI supports accountable people; every output that matters is checked and signed off by a qualified person.
Human review
A competent person checks every material AI-assisted output against source evidence before any advice, instruction or deliverable is issued, including planning points, quantities, costs, references and dates. Important assumptions and uncertainties are recorded.
AI is never used to sign certificates, approve safety-critical design or declare statutory compliance.
Data and confidentiality
Company, client and resident information is used only within our business AI accounts, never in free consumer tools. We use only the information a task needs, and never enter passwords, keys, entry codes, identity documents or bank details. Our privacy and information security policies and any client restrictions apply.
Fairness and decisions about people
An AI result is never the sole basis for tenant selection, employment decisions, safeguarding conclusions or any other decision that significantly affects a person. We check relevance, bias and accuracy, ensure meaningful human review and provide a way to challenge errors. We do not use AI to infer sensitive characteristics from someone's appearance or behaviour.
Transparency and records
We tell clients about significant AI assistance where it affects their understanding of a deliverable or risk, and keep proportionate records of sources, tool use, reviewer and corrections for material work.
We never present generated images as photographs of completed Baclands work, or invent project outcomes, qualifications or sources.
Incidents and review
Any use that creates an unmanaged safety, privacy, accuracy or discrimination risk is stopped. Concerns are reported to the Director, affected outputs are corrected and relevant parties are told where required.
Approved by the Director on 11 October 2026. Approved uses are reviewed at least annually (next review October 2027) and whenever tools, data or regulation change materially.