Lexrent

AI-Powered Legal Help for Rent Reduction

Industry

PropTech / Legal Tech

A guided path instead of isolated tools

Commissioned by a client, we built LexRent. The application supports tenants in Switzerland in bringing together the details of their rental contract for a computational rent review: AI-powered recognition reads the available contract information from the uploaded document, a guided dialog adds missing details and captures what is actually paid today.

The application then shows a computational reduction potential with all calculation details. With a positive result, a template-based letter can be prepared and downloaded as a PDF. The details and assumptions are reviewed, and tenants handle the sending themselves.

Starting point

Ein Mietvertrag dokumentiert die Situation bei Vertragsabschluss, aber nicht die ganze Geschichte: Er sagt nicht zwingend, welche Miete heute bezahlt wird, wann sie zuletzt angepasst wurde oder ob seither bauliche Verbesserungen stattgefunden haben. Für eine Einschätzung müssen historische Vertragsangaben und heutige Informationen zusammengeführt werden.

Dazu sind die Begriffe erklärungsbedürftig: Der monatliche Gesamtbetrag enthält Nebenkosten, während die Berechnung mit Nettomieten arbeitet. Informationen liegen auf mehrere Quellen verteilt, und der Weg von einer Einschätzung zu einem fertigen Schreiben erfordert weitere Arbeit.

Solution

LexRent führt von der Dateiauswahl über die Ergänzung fehlender Angaben bis zur Berechnung und Dokumentausgabe. Der Vertrag liefert die historischen Informationen, die heutige Zahlung und weiterer Kontext werden im Dialog ergänzt.

Dabei macht die Anwendung die Unterschiede sichtbar: Anfangsmiete, heutige Bruttomiete, Nebenkosten und die daraus berechnete heutige Nettomiete bleiben klar getrennt. Die Person behält die Aufgabe, Angaben und Annahmen zu prüfen und über die Verwendung des vorbereiteten Schreibens zu entscheiden.

Not just an amount, but the path of the calculation

LexRent weighs the reference interest rate effect against the cost and inflation assumptions used. On that basis, a comparison rent is calculated from the initial net rent and compared with today's net rent. The results view shows a possible monthly and annual potential along with the individual calculation steps.

The honest classification matters: the result is an orientation based on assumptions, not a legally binding determination. Whether this gives rise to a claim remains a separate legal question.

Potential and calculation path visible

AI for contract recognition, rules for the other steps

Four assistance roles interlock in the flow: contract recognition turns the uploaded document into a structured starting point, the dialog helps complete the data record step by step, the calculation breaks the total amount into individual influencing factors, and the document assistance inserts the data into a letter template.

Generative AI is only used for contract recognition. The dialog, calculation and letter are rule- and template-based, with dedicated TypeScript functions in the browser. The web application is built with React, TypeScript and an Express backend, and the PDF output is created directly in the browser.

Only the recognition is generative AI

From the computational result to a reviewable draft

With a positive calculation result, a letter to the landlord can be prepared. In the preview, sender, recipient and subject can be adjusted, and the salutation responds to the recipient's name. The letter body deliberately follows a fixed template.

Letter and calculation appendix are available as separate PDF downloads. Tenants review the content and handle the sending themselves: there is no automatic sending and no delivery confirmation, but full control over the final step.

Download the PDF, send it yourself
ARCHITECTURE & TECHNOLOGIES

From lease to a checked letter

An AI agent reads the lease, then the platform works with fixed logic. It guides the tenant through missing details, calculates the reduction potential transparently and builds the letter from a template. The tenant checks every step and sends the letter personally.

PROCESS MODEL · BPMN 2.0RENT REDUCTION
Landlord AI MODEL GPT · OpenAI LEXRENT Tenant reviews, decides Platform rules, workflows AI agents prepare the work letter Lease uploaded Read lease initial rent, reference rate Add details guided dialog Calculate potential reference rate, inflation, costs Create letter fixed template Review and adjust sender, recipient, subject Generate PDF letter and calculation annex Send letter personally
AI agentRule or automationHumanApprovalAI service or modelExchange with services and third parties
TECHNOLOGIES USED
AI MODEL
GPT by OpenAI, only for contract recognition
AI SERVICES
None. Dialog, calculation and letter run on fixed logic.
DATA AND CONTROL
Hosted in Switzerland and the EU. Every calculation step visible, no automatic sending.