The principle
Propertybot charges for compute we run on your behalf — LLM tokens, voice synthesis, grounded search. Once a credit is consumed and a script is delivered, that compute has been spent. A refund only makes sense when what came back is genuinely broken — not when the result isn't what you imagined, didn't close a buyer, or doesn't match a script you would have written.
What qualifies for a refund
We'll refund a credit or a Refresh charge only when one of these is true and demonstrable:
- Hallucinated content— the script references neighborhoods, businesses, events, or property details that don't exist. Our verifier catches most of this, but if something slips through, that's on us.
- Illegible or broken output— the script is empty, corrupt, missing sections, or rendered in a way that's unreadable.
- A clear pipeline failure we caused— server error, validation crash, fallback to a mock script. (When this happens, the system usually auto-refunds the credit; this is the manual path if it didn't.)
What does NOT qualify
- The script wasn't what you expected, or you would have written it differently.
- The showing didn't close, or the buyer wasn't engaged.
- The tone isn't to your taste — that's what the Style Profile is for.
- You changed your mind after generating.
- You used the credit and then decided you wanted the money back.
- Free promo-code redemptions ($0 transactions are not refundable; nothing was charged).
In short: taste isn't a defect.If you want to influence how scripts read to you, retake the questionnaire in Settings — that's what shapes the output.
How to request a refund (required evidence)
Email hello@propertybot.ai with all three of the following:
- Stripe transaction ID — starts with
pi_orch_; found in your Stripe receipt email under “Payment details.” - The prep URL — the
/prep/[id]link to the script you're flagging. - A screenshot of the issue — the specific hallucinated reference, the empty section, or whatever is broken. The screenshot must clearly show the problem.
Requests without all three will be returned with a request for the missing piece. This isn't bureaucratic — it's how we cross-reference each claim against our pipeline logs and verify the issue is real before we issue money back. It also keeps the system fair for users who actually had something go wrong.
Verification + processing
We cross-reference the transaction ID against our Stripe ledger and the prep ID against our pipeline logs. If the claim checks out, the refund is processed to the original payment method within 5 business days via Stripe. We aim to respond to all requests within 2 business days.
If a claim doesn't check out — e.g., the script is grounded and the references are real — we'll respond with the specific evidence (anchor sources, verifier report) so you can see what happened. We're not adversarial about this; we just need real signal to act.
Refresh charges ($1.99)
Refresh purchases follow the same rules. If the refreshed script hallucinates or breaks, same email, same three pieces of evidence, same verification.
Chargebacks
If you have a problem with a charge, please email us first — we can almost always resolve it faster than a chargeback. Filing a chargeback without contacting us may result in your account being suspended pending resolution, since chargebacks carry meaningful fees from Stripe and put our payment processor relationship at risk. If our investigation shows the underlying claim was unfounded, the chargeback dispute will be defended with our pipeline logs and the original prep output.
Abuse policy
Accounts that file repeated unfounded refund requests, attempt to extract refunds for consumed credits without evidence of defect, or chargeback charges they made no effort to resolve directly will be suspended. Beta-period access codes (CLOSER1, etc.) carry the same usage standards — abuse will result in code revocation across the cohort.
Contact
Refund questions: hello@propertybot.ai. Please include the three required items so we can act on your request without going back and forth.