Last Updated: October 5, 2026
AI Use Disclosure
BehindTheDeed uses AI to summarize, interpret, and explain property records. AI assists — evidence decides. This page explains exactly where AI is involved and the hard limits we enforce.
Where AI is used
- Report drafting: narrative summaries, red-flag explanations, and stakeholder questions are drafted from retrieved records.
- Interpretation: AI explains what a permit, lien, or assessment change may mean for your due diligence.
- Classification: findings are grouped and prioritized (e.g., open permits vs. closed, funded liens vs. released).
- Report copilot: for purchased reports, an assistant answers questions grounded in that report's findings and citations.
- Internal QA: automated checks flag unsupported claims, missing citations, and legal-overreach language before a report is released.
What AI is allowed to do
- Summarize and explain findings that come from retrieved sources
- Suggest due-diligence questions for your agent, inspector, title company, seller, or lender
- Explain confidence levels and what they mean
- Compare properties across the dimensions in your purchased reports
- Recommend verification steps with official record custodians
What AI is never allowed to do
- Invent records, sources, or findings that were not retrieved
- Certify title or guarantee a clear chain of title
- Make legal conclusions or provide legal advice
- Guarantee property value, condition, safety, insurability, or financing
- Replace inspections, appraisals, title review, surveys, insurance review, or environmental assessments
- Hide uncertainty — low-confidence and unavailable findings stay visible
How we keep AI grounded
- Every material finding must trace to a source with a retrieval date and confidence level.
- High-impact claims without citations are flagged by QA and routed to review instead of being published.
- The report copilot answers only from your purchased report, its evidence, and our published educational content.
- AI-generated insights are based on available data and should be treated as due-diligence guidance, not final determinations.
- Reports that fail QA checks are marked Needs Review and handled by a human before release.
Questions or concerns about an AI explanation?
If something in a report reads as unsupported, confusing, or wrong, tell us through contact/support — include the report ID. We take AI-explanation concerns seriously and will review the underlying evidence.