Lettings Compliance Checker

01 / Framing the compliance problem

Landlord compliance is a difficult product domain because the answer to a seemingly simple question often depends on several things at once: the property, the letting unit, tenancy status, dates, documents, evidence and the route by which an obligation applies.

A useful checker therefore cannot simply present a generic checklist. It has to collect enough structured information to work out what is relevant, retain uncertainty where the evidence is incomplete, and explain the result in a form somebody can act on.

The central design problem was how to turn a large body of regulated, date-sensitive requirements into a usable assessment without pretending that an AI model should decide legal truth.

The product needed to distinguish:

• what the landlord said;
• what could be resolved deterministically from those answers;
• which requirements applied;
• what appeared covered;
• what needed checking;
• what needed action;
• what evidence or dates were missing;
• what should appear in practical reports and workbooks.

It also needed to preserve the boundary between structured compliance guidance and legal advice or certification.

02 / Designing the deterministic core

The application is built around explicit objects and transitions rather than generated prose.

Its practical model includes:

• cases;
• properties;
• letting units;
• raw wizard answers;
• resolved facts;
• requirement results;
• action items;
• report/workbook artefacts;
• payment/access state;
• editable guidance/content around, but not inside, the deterministic legal truth chain.

The wizard is schema-driven and applicability changes according to the property/tenancy context. Results are generated from deterministic logic and then carried into review and paid-output surfaces.

03 / Building the product surface

The live application includes:

• structured case setup;
• a schema-driven question flow;
• deterministic result generation;
• free pre-payment findings/review;
• paid results and detailed guidance;
• downloadable workbook/reporting material;
• practical logbook/worksheet outputs;
• a compliance calendar export;
• Stripe-backed payment/access flows;
• scenario and output-testing infrastructure;
• local administrative tools for content, scenarios, diagnostics and release support.

The broader development process has also involved extensive output-parity checking, scenario seeding, customer-surface review, security hardening work and explicit distinction between local test evidence and hosted/live acceptance.

04 / Reaching live production

Lettings Compliance Checker reached live/deployed production status in September 2026.

That is a product maturity statement, not a legal one. The service provides structured guidance to help landlords identify what may need attention. It does not certify legal compliance or replace professional legal advice.

05 / What I did

Through Niche Clever I have been designing and developing the product across the problem model, user journey, compliance logic, reporting and implementation process.

A major architectural decision was to keep AI out of the authoritative legal evaluation path. Coding agents can assist development, analysis, testing and documentation, but the rules that determine durable requirement state remain inspectable and reproducible.

That separation is central to the product rather than an after-the-fact safety statement.

06 / What this work demonstrates

The project is a strong example of using a deterministic core where the consequences of inconsistent model output would be unacceptable.

It also demonstrates a wider method: turn ambiguous real-world material into explicit intermediate structures, preserve the source/evidence distinction, make state transitions inspectable, and use AI where it helps without allowing it to become the source of truth.