Lease Compliance Checker

A lease compliance checker helps leasing counsel review a draft residential or commercial lease before it goes out for signature. The model reads the draft against a jurisdiction-specific rule set and returns structured findings: missing or incomplete disclosures, clauses that conflict with local or state requirements, and language that may create fair housing exposure.

The output is an audit memo for humans, not a signed-off lease. Counsel decides what to change, what to accept with a documented rationale, and what to escalate to outside counsel or a compliance lead. The model does not execute the lease, send it to the tenant, or mark the file as cleared.

Related work often sits upstream and downstream of this review. Heads of terms may already exist from a Lease Heads of Terms Generator. Rent and unit packaging may have been set during Dynamic Rent Pricing at Lease-Up. Screening outcomes from AI Tenant Screening & Scoring should never be mixed into fair housing–sensitive lease language without counsel’s explicit control of what is disclosed and how.

The checker needs two inputs before it produces findings:

1. Draft lease text (or a structured clause pack that maps cleanly to the form in use).
2. Jurisdiction rule set covering the property’s governing law, including required disclosures, prohibited or restricted terms, notice and cure rules that must appear in the form, and fair housing–sensitive phrasing constraints for that market.

If either input is missing, incomplete, or not confidently matched (for example, the draft is blank, the file cannot be parsed, or the property’s city/state/county rule pack is not loaded), the system returns empty output: no findings list, no “likely compliant” summary, and no partial score. Empty output is preferred over a silent pass. Counsel should treat an empty result as “review not run,” not as clearance.

Optional context that improves precision without replacing counsel judgment includes the property type (multifamily, single-family rental, mixed-use), lease product (standard renewals vs. first-generation forms), and which form version or playbook the portfolio already uses. Those fields refine which rules fire; they do not authorize the model to invent requirements that are not in the loaded rule set.

Counsel (or a paralegal under counsel’s process) submits the draft and confirms the jurisdiction pack. The model walks the document clause by clause and against a disclosure checklist for that jurisdiction.

Typical finding categories include:

• Missing disclosures: lead-based paint, mold, bedbug history, flooding, utility billing method, smoke detector or carbon monoxide notices, rent control or just-cause addenda where required, and any locality-specific pamphlets or statutory inserts the rule set lists.
• Non-compliant or high-risk clauses: unlawful early termination fees, waiver of habitability or statutory rights that cannot be waived, notice periods shorter than law allows, lockout or self-help language, attorney-fee shifts that conflict with local rules, and security deposit handling that does not match statutory timelines or itemization rules in the pack.
• Fair housing risk language: preferences or restrictions that could be read as discriminatory (familial status, disability accommodations framed as exceptions rather than obligations, source-of-income language where protected, and steering-adjacent marketing copy that leaked into the lease body).

Each finding should cite the draft location (section or clause ID), the rule or checklist item that triggered it, and a short plain-language explanation of why it matters. Severity tags (blocker vs. advisory) help triage, but severity is a workflow aid. Counsel still owns the final call.

Where the draft is silent and the rule set requires an affirmative disclosure or addendum, the finding should say “required content absent,” not invent substitute language as if it were already approved. Suggested remediation can be offered as draft options for counsel to edit; those suggestions remain proposals until counsel accepts them into the form.

The compliance checker is a pre-execution gate inside the leasing legal workflow. A practical sequence looks like this:

1. Draft or form assembly completes (often from a controlled template plus deal-specific riders).
2. Checker runs against the confirmed jurisdiction rule set.
3. Findings are reviewed by leasing counsel; blockers are fixed or waived with written rationale.
4. Only after counsel clearance does the lease proceed to countersignature and execution systems.

Human-in-the-loop means the model flags issues; counsel clears the draft. Do not auto-approve on “zero findings.” Zero findings means the model did not match a loaded rule against the text it could read. It does not mean the lease is lawful in every respect, especially for novel deal structures, new ordinances not yet in the pack, or commercial terms outside the residential checklist.

Access control matters. Fair housing–sensitive findings and screening-adjacent context should stay with legal and compliance roles. Property staff can see remediation tasks counsel assigns without receiving raw model commentary that overstates legal conclusions.

When the portfolio uses multiple form families, pin the rule set and form version in the run record so later audits can explain what the model was asked to check. That audit trail supports internal QA and outside counsel review when a dispute later turns on whether a disclosure was in the executed set.

Jurisdiction packs go stale. A new disclosure ordinance, rent regulation update, or published guidance can change what “compliant” means faster than a static checklist. Empty or outdated packs should fail closed (empty output or an explicit “rule set unavailable” state), not guess from general training knowledge presented as local law.

Ambiguous clauses are another failure mode. The model may flag language that is commercially aggressive but lawful, or miss a cleverly worded workaround that still conflicts with statute. Counsel should treat borderline hits as prompts to read the clause, not as automatic redlines.

Parsing failures (scanned PDFs, nested exhibits, handwritten riders) can hide required addenda. If the draft cannot be fully ingested, return empty output or a hard parse-error state rather than auditing only the cover pages and implying the rest was reviewed.

Fair housing review is language and pattern risk detection, not a substitute for a fair housing compliance program, training, or counsel opinion on a specific applicant decision. Keep screening scores and protected-class inferences out of the lease text path unless counsel has defined exactly what may appear.

Finally, do not use the checker as a substitute for licensed legal advice in jurisdictions where that distinction matters for how the tool is described to internal users. Position it as a draft audit assistant that accelerates counsel’s checklist, with clearance remaining a human legal act.