Product

The AI operations desk built for your inbox.

RentaraAI drafts guest replies, triages requests, tracks renewal deadlines, and prepares filing materials for review — always with you as the licensee of record and the final approver. Nothing goes out without your click.

One inbox, in your voice.

Guest messages route through the AI immediately. It reads the current conversation context and the property's operational house rules, drafts a reply that sounds like you wrote it, and drops the draft in your inbox quickly. You approve and send.

Compliance workflow + checklist tooling.

Renewal and filing reminders surface against the rules you have configured — RentaraAI is the workflow and checklist layer for the rules already in your account, not a legal-advice product. Every filing is reviewed and submitted by the named human operator — you stay the licensee of record in every city we cover. Verify the rules that apply to your address with the city and a licensed attorney.

Triage, escalation, and turnovers.

The AI triages tickets, drafts vendor briefs, and hands anything you flagged off-limits straight to the named human — with the full context already attached.

Capability snapshot

What's live today, in beta, or on the roadmap.

Same list as the /terms snapshot — live capabilities, beta capabilities behind documented gates, and the rows still on the roadmap. Adding or flipping a capability ripples here and to /terms at the same time so marketing and legal stay in lockstep.

Showing 10 capabilities. Reference: /terms.

Approval workflow

A human approves every outbound guest message.

The desk never sends on its own. Each guest message moves through four well-defined states — once an operator approves a draft, it sends and the action is logged against a named human. Edit a draft, hold a draft, or route it to escalation at any point.

  1. Step 101

    Arrived.

    A guest message arrives on a connected inbox. The desk reads the conversation context — property, reservation, prior turns — and decides whether to draft, defer, or escalate.

  2. Step 202

    Triaged.

    Routine questions are queued for an AI draft. Anything you marked off-limits bypasses the desk entirely and goes straight to the named operator with the full context already attached.

  3. Step 303

    Drafted.

    The AI prepares a reply in your voice, drawing on the current conversation context and the property's operational house rules. The draft lands in your inbox quickly, waiting on your click.

  4. Step 404

    Approved & sent.

    You review the draft, edit if you want, and click Approve & send. That single click closes the loop: the message goes out, the audit ledger gets a new row, and the property's per-action log is updated.

Guest-message lifecycle

Drafts across the full stay, not just the inbox.

The desk prepares, drafts, escalates, and logs messages across four states — pre-stay, in-stay, departing, and post-stay — so nothing falls into the gap between an arrival and a review.

01Stage 1

Pre-stay.

The reservation lands on your connected booking channel. Pricing, availability, cancellation, and pet-policy drafts are queued for approval before the guest ever has to ask. Door codes, wifi, and house-rule drafts prepare on a clean schedule.

02Stage 2

In-stay.

A wifi drop, a stuck lock, or a 2:00 a.m. question. The desk gathers context, drafts a reply that respects your house rules, and surfaces anything urgent to the named human — drafts still wait for your click.

03Stage 3

Departing.

Checkout windows, late-arrival replies, and deposit reconciliation drafts are prepared for review and queued behind your approval. Late-checkout drafts land before checkout-day itself.

04Stage 4

Post-stay.

Thank-you and review-request drafts are timed to checkout plus a configured grace window. Cleaner turnover briefs and damaged-unit notes are prepared for review and dispatched after you approve.

Escalation handling

Off-limits topics route to a named human, with context.

Configure the topics the AI must never reply to, the hours it routes without drafting, the dollar thresholds that demand an owner call, and the named human each escalation lands in front of.

  • 01Topic rules — the desk bypasses its own draft pipeline and routes the message to the named operator for any topic you flagged off-limits.
  • 02Hours rules — operations outside your typical hours route straight to the on-call human, even when the topic is routine.
  • 03Dollar rules — reservation-impacting asks above a per-stay threshold (refund, partial refund, compensation) escalate to you for approval before any draft is prepared.
  • 04Named-human routing — every escalation lands in front of the operator you configured for that property, with the conversation history, attachments, and AI summary attached.

Escalation rules are configurable per property. Updates take effect on the next inbound message — nothing is silently waved through.

Audit log

A per-property ledger you can export.

Every action — draft, approve, send, escalate — is logged against a property, a timestamp, and a named actor. The audit log is yours to export and share with your compliance team.

TimestampActionStatus
Wi-Fi + access draft preparedAI desk · awaiting review
Awaiting approval
Locked-out guest escalatedRouted to Tomás · ADA-411
Routed to operator
Approved by Alex · sentGuest reply delivered after approval
Approved & sent
Kitchen-light triage queuedVendor shortlist ready
Awaiting approval

Connected-channel behavior

Bookings, PMS, comms, and locks — one desk.

The desk reads from and writes to the systems you already operate. Each channel category behaves with the operations desk as the single source of truth for what left your account and who approved it.

Bookings

Reservations, guest details, and arrival metadata flow in from your connected booking channels. Drafts go back through the same channel after your approval.

PMS

Property management system rows — unit availability, guest profiles, internal notes — are read on demand and updated only after an explicit approve event.

Comms

Outbound guest messages are sent through the Polsia email proxy after your approval. Inbound messages route into the desk and are triaged like every other arrival.

Locks

Connected smart-lock providers deliver lock-state and access events. The desk prepares access-code drafts; nothing unlocks or recodes until you approve.

Read the integrations directory for the supported vendor list and connection states.

Security & privacy

You stay the licensee of record. We stay in the audit trail.

Guest and reservation data is encrypted in transit and at rest. Access is scoped per account with the principle of least privilege — the desk reads only what it needs to draft a reply.

  • 01Data scope — the AI is narrowed, per call, to the data that call needs: guest-reply drafts see the latest 5 turns of the guest thread + the property's operational house rules; pilot demo drafts see the same fields on the host's own first property; nightly-rate suggestions see address / city / bedrooms; the narrow chat relay is auth-gated streaming chat — the server derives a per-call-site system prompt, accepts only enumerated `{role, content}` messages (≤ 20 turns, 4,000 chars each, 64 KiB total), rejects unknown fields, and excludes payment, credentials, session tokens, payout details, uploaded IDs, and cross-tenant portfolio data. It does not see data tied to properties you do not operate. Operational house rules are always sent when a property is in scope; the per-host style block (tone / signature) is always gated by your personalization toggle.
  • 02Retention — drafts and ledger rows are kept while your account is active. You export or delete on request; nothing is held past the documented retention window without your written instruction.
  • 03Subprocessor posture — every subprocessor with access to guest or reservation data is listed on the privacy page, with category, retention, and hosting region, in the same form as the security posture on /security.

See the security & data-protection page for the full control-surface description.

Onboarding

Get started in the first days of your 14-day pilot.

These three setup steps happen during your 14-day pilot. Work through them at your own pace — the pilot stays open the full 14 days, and each draft still requires your approval before it is sent.

  1. 01Day 1

    Connect one property.

    Connect one booking channel and one messaging channel for a single property. The desk operates end-to-end on that one property before any other unit is added.

  2. 02Day 2

    Configure rules.

    Set house rules, escalation topics, hours-of-operation, and named-human routing for the property. Approve a small batch of pre-stay drafts so the voice calibration lands before live traffic.

  3. 03Day 3

    Go live.

    Put the desk in front of real reservations. Each draft still waits for your click; you start the second-property rollout only after the first property has logged enough approvals to confirm the calibration.

Illustrative UI

What the operations desk looks like, mid-stay.

A quick visual — labelled sample — of the per-property audit log and review-inbox pattern. The desk is the single pane of glass for every draft, every approval, and every send.

Illustrative sample — not real customer data.Sample ledger rows that show what a property's audit log looks like once a few approvals have passed through.
Wi-Fi + access draft preparedAI desk · awaiting review
Awaiting approval
Lockout escalated to named managerEscalated to Tomás · ADA-411
Routed to operator
Approved by Alex · guest reply sentGuest reply delivered after approval
Approved & sent
Turnover brief queued for reviewMaria · 9:00 entry
Awaiting approval

Sample rows shown above — they are illustrative. The look-and-feel reflects the production desk; the data does not come from any active customer account.

See it in the flow.

How the AI works step-by-step, the guest-message journey, and the security posture — all live on the home page.

See the how-it-works walkthroughStart Pilot