Reagent Travel

Travelby Reagent AI, LLC

Launched Mar 11, 2026 on ChatGPT

Search and compare flights just by chatting. Reagent Travel highlights what matters and suggests smart tweaks as you go. Instead of digging through endless results, quickly land on the right flight and know why it fits. When you’ve found the one, just hit Book.

2ChatGPT Tools

Reagent AI, LLCDeveloper

TravelCategory

Available Tools

Create Itinerary Request

CreateItineraryRequest

Full Description

If the user wants to book a flight, this tool can be used to create a flight itinerary request once the user has provided an origin, a destination, and a single outbound date. Ask clarifying questions only if required information is missing or ambiguous (e.g., missing location(s), missing date(s), user specified date range(s), or unclear one-way vs round-trip when only outbound date is given). Do not delay for preferences or confirmation; default to economy cabin unless specified otherwise and infer the traveller persona from knowledge about the user. Normalize dates safely (never past dates; infer the next future year if missing). Include time preferences only when explicitly stated. Flights only; do not use for lodging or ground transportation.

Parameters (4 required, 4 optional)

Required

from_airportstring

Origin IATA code (exactly 3 uppercase letters, pattern ^[A-Z]{3}$). If the user mentions a specific airport, use its airport code (e.g. LHR). If they mention only a city, use the city code (e.g. LON). Only use locations explicitly mentioned by the user.

from_datestring

Start date for the itinerary (YYYY-MM-DD). Must be >= today\’s date. Must be a single date, not a range. Only use date specified by the user (year may be inferred).

to_airportstring

Destination IATA code (exactly 3 uppercase letters, pattern ^[A-Z]{3}$). If the user mentions a specific airport, use its airport code (e.g. LHR). If they mention only a city, use the city code (e.g. LON). Only use locations explicitly mentioned by the user.

traveller_typestring

Traveller persona inferred from Model Set Context and User Knowledge Memories. Choose one of: budget, standard, road_warrior, vip. Use stable signals from memory (e.g., cost-minimizing behavior → budget; frequent or time-sensitive business travel → road_warrior; strong preference for comfort, status, or premium travel → vip). Default to standard if memory data is missing, conflicting, or unclear. Do not ask the user to choose unless they explicitly mention it.

Options:budgetstandardroad_warriorvip

Optional

cabin_classstring

Cabin class for the trip. Use economy unless specified or dictated by the traveller persona.

Options:economypremium_economybusinessfirst

parent_itinerary_request_idstring

Itinerary request ID of the related parent itinerary request, if any.

return_datestring

Return date for the itinerary (YYYY-MM-DD), leave blank if a one way trip. Must be >= from_date. Must be a single date, not a range. Only use date specified by the user (year may be inferred).

time_preferencesobject

Optional time preferences for outbound and return legs. Only include when the user explicitly specifies a time preference. Allow extra time if there are time constraints (e.g. use "arrive-before-1300" for a "meeting at 2").

Get Next Best Itinerary

GetNextBestItinerary

Full Description

Retrieve flight itineraries for an existing itinerary request. The initial call (when no prior results exist) returns 1–4 top results. Each subsequent call iteratively returns the next best available itinerary based on the requested feature improvement(s), refining results across calls and surfacing strong alternatives when tradeoffs are required. Use the relevant current itinerary’s position (as shown in the UI list) when requesting improvements.

Parameters (2 required, 6 optional)

Required

featurestring

The feature to improve. For the first (parent) search, use "price" unless otherwise specified. For all child searches, use the most important feature in the user's request. Use "times" only when referring to BOTH outbound and return.

Options:pricedurationoutbound durationreturn durationstopoverstimesoutbound timereturn timeairline preferredairline not preferredredeye preferredredeye not preferredredeye outbound not preferredredeye outbound preferredredeye return not preferredredeye return preferredfrom airport preferredfrom airport not preferredto airport preferredto airport not preferredoutbound stopover airport preferredoutbound stopover airport not preferredreturn stopover airport preferredreturn stopover airport not preferred

itinerary_request_idstring

The ID (itinerary_request_id) of the itinerary request.

Optional

current_result_idstring

The position/ID of the itinerary option being improved (as shown in the UI list). Use "0" for the initial call when no prior results exist.

feature_optionstring

Feature-specific option. For "outbound time"/"return time"/"times", a single string with format "leave|arrive-before|after|around-HHMM" (e.g., "leave-before-0800, arrive-around-1500, etc") or set to "0" to optimize for saved time request(s). For "airline preferred"/"airline not preferred", the airline code(s) (comma separated if >1) (e.g. "UA,AA" for United & American). For "from/to airport preferred"/"from/to airport not preferred", the airport code(s) (comma separated if >1) (e.g. "JFK,LGA"). For "outbound/return stopover airport preferred"/"outbound/return stopover airport not preferred", the airport code(s) (comma separated if >1) (e.g. "JFK,LGA"). For all other features, set to "0". Do not pass empty strings; use "0" instead.

feature_secondarystring

Secondary feature to improve, if any. Include only for improvement requests in the current query.

feature_secondary_optionstring

Secondary feature option, if any. Same usage as "feature_option".

feature_tertiarystring

Tertiary feature to improve, if any. Include only for improvement requests in the current query.

feature_tertiary_optionstring

Tertiary feature option, if any. Same usage as "feature_option".

Links

Website Privacy Policy Terms of Service