Generic AI and leases
A fluent answer is not the same as a checked lease answer.
The problem is not that general AI is useless. It is that a lease question needs structure, current sources and a fast route back to the signed document.
Source, party, hierarchy and current-law check may remain hidden.
The signed document stays open for the person making the decision.
The confident answer is the dangerous one
Fluency does not prove the lease was read correctly.
The useful difference is not a louder AI claim. It is the route back to the signed source and the visible boundary around the answer.
May treat the upload as one long block of text
Maps pages, clauses, definitions and parties
May recall a rule without checking its current status
Opens the current source when law affects the answer
Can blend the lease, law and assumptions
Labels the lease, law, supplied evidence and open point
A polished answer can hide a weak source route
Keeps disagreement and uncertainty visible
Human check still required. A visible source route makes error easier to find. It does not make error impossible.
Where confidence outruns evidence
Four things a plausible paragraph can hide.
Each problem is manageable when it is visible. The risk comes from receiving one polished answer with no way to see which layer failed.
The text may not have been extracted cleanly
Scans, schedules, plans, handwritten changes and broken page structure can alter what the model receives before it answers.
Definitions and cross-references can be flattened
A clause may depend on a definition, another schedule, a later deed or a headlease. Reading each paragraph as isolated prose can change the result.
The right clause can be applied to the wrong party
Landlord, tenant, management company, headlessee and underlessee are not interchangeable. The answer needs the parties and governing interest made explicit.
Recall is not an as-at legal check
A named Act or section can sound convincing while being prospective, amended, outside scope or unrelated to the question actually asked.
The LEASE-iQ route
Make the source part of the answer.
LEASE-iQ is designed around the steps a person needs in order to check the work, not around presenting the longest explanation.
Structure the document
Identify clauses, schedules, definitions, parties and cross-references before answering the live question.
Trace the question
Find the covenant, right, obligation or payment route that actually controls the issue.
Connect current sources
Bring in the relevant current law where it adds to, qualifies or overrides the document.
Compare candidate answers
Run more than one reasoning path and leave disagreement or missing evidence visible instead of smoothing it away.
Open the signed clause
Give the user the clause, page and source PDF so the answer can be checked before it becomes a letter or decision.
The boundary
LEASE-iQ does not make judgment disappear.
Use the source-backed first pass
Routine questions can move faster when the document, role and current source align and the user can open the clause.
Show both readings
Where wording, document hierarchy or facts conflict, the useful output is the competing case and the evidence that would resolve it.
Escalate a focused question
A disputed, high-value, safety-sensitive or reliance-heavy issue still needs the appropriate professional judgment.
Sources for this page
Open the law and guidance yourself.
Research on how language models use long contexts. It supports caution about retrieving information from long documents, not a claim that every model always fails.
↗Building Trust methodEvidence and verification standardHow Building Trust separates the document, current source, inference, uncertainty and human decision.
→Links and current-law position checked 18 August 2026. Apply the governing documents, building facts and current law to the case in front of you.
Take the next step in LEASE-iQ
Build a source map before asking for a conclusion.
Use this when the issue is unclear or several clauses may interact.
Keep checking: A well-structured question makes uncertainty easier to see. It does not turn incomplete sources into certainty.
