FAQ
Good questions.
Quick answers on grounding, safety, providers, data and getting started.
Is HostOS a PMS or a channel manager?
No. HostOS is an AI workforce that operates for you. The Inquiry Agent is live today — answering guest questions grounded in your property — with revenue, trust-and-safety and operations roles arriving as each one earns trust.
How does it know my property on day one?
Add your property and import your own Airbnb data export — HostOS turns listings, conversations, reviews and house rules into structured facts, a knowledge base and your communication style. Day one feels like day one-thousand.
Will the AI invent answers to guests?
Never. Answers come only from grounded, host-approved knowledge, gated by a confidence threshold. When it’s unsure it escalates to you with a reason — it never fabricates.
How does autonomy stay safe?
Agents are advisory-first and earn autonomy rung by rung within your limits. Nothing is sent automatically until it has been proven in shadow and cleared the send gate — a trusted source, sufficient confidence and a safe intent — and every decision is audited. Consequential calls stay human-owned.
Is my data safe and private?
Yes. Postgres row-level security isolates every record, secrets are encrypted, models receive only the question and governed context — never your full guest history — and every action is audited and reversible. See the Security and Compliance pages.
Which providers do you support?
Add your property directly today. Provider connectivity — Airbnb via an authorized aggregator, Hospitable, Guesty and more — is in development on a provider-agnostic architecture, so new integrations don’t change how HostOS works.
What does it cost, and is there a free plan?
There’s no permanent free plan — HostOS is a premium AI workforce. Today you can start free in early access; when billing launches, every plan begins with a 14-day free trial. Pricing is in ₹ (INR) with GST invoicing. See Pricing.
How long does setup take?
Minutes. Add your property (or connect once), teach it in plain language, and it starts drafting grounded replies — no manual training.