Inside Waydone's AI: planning trips without training on your prompts

When you ask Waydone to plan five days in Lisbon, a small ballet runs behind the scenes. We think it's worth being transparent about that ballet — because trust starts with knowing what your tools are doing with your words.

What actually happens when you send a prompt

1. Your prompt arrives at Waydone with your auth token. We attach the relevant trip context — destinations, dates, party size, preferences — that you have already entered.
2. We forward that bundle to one of our AI providers (the underlying large-language-model). The provider returns suggestions.
3. Waydone reshapes the suggestions into the day-by-day structure you see in the trip view, runs basic safety checks, and saves the result against your trip.
4. The conversation context is kept so follow-up questions ("make day three more relaxed") can be answered without re-asking the whole brief.

The no-training commitment

Our contracts with the AI providers contain an explicit ban on using your prompts or our outputs to train, fine-tune or improve their general models. Inputs and outputs are deleted within short retention windows we have negotiated. We don't sell prompts and we don't share them with advertising networks.

What Waydone does with the data

• Provide the response you asked for and keep the trip history you save.
• Run abuse and safety monitoring (short-lived logs, then deleted).
• Debug crashes and bugs via error monitoring (no full prompts, only stack traces and request metadata).
• Aggregate anonymous, statistical metrics so we can tell whether new features actually help.

What we don't do

• Train AI models on your prompts.
• Sell prompts or trip data to anyone.
• Share prompts with advertising networks.
• Reuse a customer's trip plans to serve another customer.

Our Privacy Policy spells out every detail with retention windows and your GDPR/RGPD rights. The short version: the AI is a tool we lease to help you plan; your trip remains yours.