No. It explains plain-language meaning of clauses and contrasts them with statutory norms you have stored, but every reply includes a 'this is not legal advice' line for ambiguous situations and the bot is configured to recommend a tenant attorney for evictions, security deposit lawsuits, discrimination cases, and habitability disputes. Storing the disclaimer text as a setting keeps it consistent across answers.

The pasted lease is included in the model prompt for that turn and stored in the conversation row in wp_sleek_ai_chat_log on your WordPress database. The model API call goes to your provider (OpenAI, Anthropic, Google, OpenRouter) under whatever retention policy your account has. No third-party chatbot SaaS sees the lease. If you do not want the lease persisted in the log, set the chatbot to log only the user-bot exchanges and not the pasted document.

Write a short reference post per jurisdiction with the things that matter: notice periods, security deposit caps, retaliation rules, habitability standards, late-fee limits. Tag the posts with a region taxonomy. The system prompt instructs the bot to ask for the user's location once and load the right reference. For multi-region sites this scales to dozens of jurisdictions without inflating any single prompt.

Leases run 8,000 to 25,000 tokens depending on length and clause count. GPT-4o, Claude Sonnet 4.5, and Gemini 2 Pro all handle that in a single prompt without retrieval. For huge ground-lease or commercial documents over 50 pages, pair SleekAI with an OpenAI Files vector store and have the bot retrieve only the relevant clause section per turn. Most residential leases do not need that.

Yes. Add a preset like 'Summarise the key terms' that triggers a structured summary: rent and increases, term and renewal, notice required, deposit and what can be withheld, pet rules, sublet rules, fees, and any unusual clauses. The bot pulls each from the pasted text and produces a one-paragraph each. Useful as the first turn before the user starts asking specific questions.

If the lease was pasted with its numbering intact, the bot reads clause numbers directly from the text and quotes them correctly. If the user pasted a poorly formatted version where numbering is broken, the bot will quote the line itself rather than invent a number. Tell users in the initial message to paste the lease with formatting preserved (PDF text-extract usually works, screenshots OCR less reliably).

Yes. After explaining the notice clause, the bot can produce a one-paragraph notice letter referencing the specific clause and date, formatted for email or a printed copy. The user reviews, fills in the landlord's address, and sends. This is the highest-utility output for renters approaching end of term and turns the chatbot from explainer to practical tool.

Yes with caveats. Commercial leases have very different structure (CAM charges, percentage rent, holdover, options to extend, exclusivity), so the system prompt and jurisdiction notes should be separate. Most sites split the bot into 'residential lease' and 'commercial lease' via SleekAI multibot, each scoped to a different URL pattern with its own reference content. Commercial users also expect a much stronger 'see a real attorney' nudge.