We are happy to announce our latest TripBuilder release. Please see these release notes for the most significant changes. Let us know if you have any questions, feedback, or ideas for features or improvements!

A new major update to Planner Copilot brings a set of new capabilities to design trips with the help of AI:

Prompts can contain links to TikTok, YouTube, and Instagram videos and photo slide shows. Planner Copilot will analyze it and use it to generate an itinerary suggestion. Also, Planner Copilot will show a preview of the video itself to entertain the users while they wait for the response.

You can pass a link to a website, e.g., a travel blog or a Wikivoyage page. Planner Copilot will analyze the website and use it to generate an itinerary suggestion.

Planner Copilot will fetch itinerary templates, and pick up to 6 itinerary templates that are related to the user’s prompt. For these 6 templates, small tiles will be shown directly in the chat. Template suggestions will only be shown for the initial itinerary suggestions, not anymore when the user is refining the itinerary with follow-up prompts.

This feature is behind a feature flag, see Settings > Discovery > Planner Copilot. By default, it is disabled for existing TripBuilder instances.

Planner Copilot better leverages past prompts/responses of the same session when the user refers to them for further refinements, improving the conversational experience for users.

Planner Copilot has been upgraded to the latest OpenAI models with reasoning support. This will lead to even better proposals and conversational chats.

The performance has been improved through caching of remote calls for hotels, activities, video analysis, and template suggestions.

Search results in the Itineraries and Components sections now sort by relevance when the text search is used. This fix ensures that your text queries prioritize the most accurate matches rather than showing unfiltered or poorly ordered results.

We improved the Global Recommended Hotels CSV upload to handle large files reliably. Uploads of up to 9,000 hotels which previously timed out now take a maximum of 30 seconds to complete.

Users will now receive clear success or error notifications when the upload of large hotel lists finishes or fails, avoiding confusion and frustration.

Travel documents no longer incorrectly show “0 Nacht” for Point-to-Point (P2P) template accommodations that include overnight stays. The night count is now displayed correctly.

Together with this change, the inventory warns the user upon saving of an activity if activity overnight segments start/end on the same day.

We added room allocation information to the ConnectAPI activities offer search endpoint. Previously, this data was only available on the details call, limiting what suppliers could offer during search.

Suppliers can return more accurate pricing and availability results during the initial search, adapting the response with the room needs.

We implemented passenger selection for ConnectAPI activities. Previously, the endpoint assumed all itinerary passengers were participating in every activity, causing incorrect pricing and making it impossible to answer per-passenger questions or assign extras to specific guests.

Operators and users can now specify which passengers participate in each activity, get accurate per-participant pricing, and answer activity questions on a per-passenger basis.

We've added dynamic checkout field requirements for accommodation bookings, ensuring you only see the traveler information fields that are actually needed for each specific accommodation. When a supplier requires additional documentation like first-passenger contact information, those fields now automatically appear as mandatory during checkout. The system intelligently adjusts what information it requests based on the specific requirements of each accommodation offer, rather than showing the same fields for every booking.

This enhancement eliminates the frustration of incomplete bookings due to missing required information. Now the checkout process clearly indicates all required fields upfront, ensuring you can complete accommodation bookings successfully on the first attempt without surprises.

We updated the SunnyCars rental car search logic to only return results from the searched airport. Previously, a wide search radius caused the supplier to return best-priced offers from nearby airports and locations instead of the requested one, resulting in fewer relevant results for users.

When a search originates from an airport, we now filter out non-airport pickup locations, giving users more accurate and relevant car rental options.

We automated the process of updating Alamo rental car location data, eliminating the need for manual monthly uploads. The system now automatically retrieves the latest location information from Alamo's data feed on a weekly schedule, ensuring rental car pickup and dropoff locations remain current without manual intervention.

Previously, location data required manual download from Alamo's server and manual integration every one to two months, which occasionally resulted in outdated location information. The automated workflow ensures Alamo rental car searches always reflect the most current available locations while reducing operational overhead.

Added a cancellation_policy_label column to the core TripBuilder components data export. The column holds the interpretation of the cancellation policy at the moment the component was booked and can take the following values: non-refundable, partially refundable, free cancellation. This can help tour operators to run investigations on risk exposure.

The data is only available for accommodation components. Also note that for certain hotel providers, the data will only be available for accommodation components booked after summer 2025.

Added airline_* columns to cover the (operating) carrier code and flight number. The new fields are:

The TripBuilder itineraries table in Analytics now includes per-component-group sales price sum columns, giving tour operators a more granular breakdown of revenue across component categories. All prices are in the sales currency of the itinerary.

Important remark: the conversion calculation of the component-group sales prices is done in the data warehouse from the net prices and might show slight differences compared to the ones shown in TripBuilder’s Customer Care. This can happen due to interpolated FX rates and different rounding patterns.

The new columns are: