Disclosures

AI involvement

UnitWell is run by a software agent, not a person. It drafts and sends routine messages, classifies maintenance requests, prepares renewal and payment-plan proposals, and writes the summaries owners and managers read.

A reply you receive from the agent may be produced automatically, at any hour, without a person reading your message first. Depending on how an account is configured, the agent runs on rule-based logic or on a large language model; either way, it is automated.

Automated work is labelled in the product: agent activity appears as tasks and as entries in the message trail, so a manager can see what was sent and why.

Human approval

Consequential actions are never taken by the agent on its own. Anything that moves money above $400, changes a lease, or penalizes a resident is filed as a task that waits for a person to approve it.

That includes late fees, returned-payment fees, payment plans, lease renewals, move-outs, applicant approvals and denials, tour scheduling, and vendor dispatch above the threshold.

Approving the task is what applies the effect. Each approval is recorded in the audit trail with who approved it and when, and a declined task records the decline the same way.

Not advice

Nothing produced by UnitWell is legal, tax, or financial advice, and using the product does not create a professional relationship of any kind.

Documents, templates, notices, and tax summaries generated here are starting points that a manager reviews. They are not a substitute for a licensed attorney or accountant, and they are not tailored to any particular jurisdiction or situation.

The agent is instructed not to draft notices, lease clauses, or other quasi-legal text and not to state statutory limits as fact; it is built to decline those requests and point back to a person. Notice and letter templates in the product are marked as pending counsel review until counsel has reviewed them.

Fair housing and equal opportunity

UnitWell supports equal housing opportunity. The product must not be used to advertise, screen, price, or communicate in a way that treats people differently on the basis of a protected characteristic.

Screening criteria are fixed per applicant segment in the product rather than set per applicant, so the same requirements apply to everyone in a segment. Applicant approvals and denials are approval tasks decided by a person, not by the agent.

If you believe the product has been used unfairly with respect to your housing, contact us at hello@unitwell.ai.

What is not live yet

Several external integrations are simulated in the current build. We would rather say so here than let the product imply otherwise:

• Payments — rent payments are recorded against the ledger; no card or bank rail is connected, so recording a payment does not move real money. Subscription billing and automated owner payouts are likewise not connected.
• E-signature — lease signing returns a simulated envelope by default. No signature is collected through a real e-signature provider unless one has been explicitly configured.
• Email delivery — outbound email is recorded in the product's message trail by default and is not delivered. A real delivery path (Amazon SES) exists and is switched on per deployment.
• SMS — no SMS provider is connected. Messages on the SMS channel are recorded in the message trail only; nothing is texted.
• Applicant screening — credit and background reports are simulated. No consumer reporting agency is queried.
• The large-language-model brain is off by default. Unless it has been enabled for a deployment, the agent's replies come from deterministic rules rather than a language model.