Itinerary drafts
A first draft built only from places worth recommending.
Generic AI trip planners write from memory, which means whatever the internet has said about a destination. Cordovan drafts from a curated catalog of properties and operators vetted by working advisors. It then sizes, sequences and explains every choice against your client's profile. You get a draft worth refining instead of one you have to fact-check.
Selection, not recall.
Ask a general-purpose AI for ten days in New Zealand and it will write something fluent from what it absorbed in training. Some of it is out of date. Some of it is the internet's top ten. Occasionally it is a lodge that never existed, assembled from the names of two that do.
Cordovan works the other way around. Every recommendation comes from a catalog of hubs, lodges and experiences that advisors have visited, or whose operators they know. The AI's job is to select, sequence and explain from that catalog. It never adds a place that isn't in it.
That one decision removes the worst failure, invented or stale places, before a draft is ever written.
It never starts from a blank page.
A draft begins with two things: your client's travel profile and the vetted catalog. Everything after that is selection and sequencing, and it ends with you. Because the route is chosen first and each stop is filled from a short list for that place, the drafts stay focused as the catalog grows. Today it holds around 300 experiences and properties across 50 hubs in Australia, New Zealand and the South Pacific.
-
Start from the client
Travel months, nights, budget, experience tier, travel style, party, lodging and guiding preferences, and the modes of transport they will and won't take.
-
Choose the route
Hubs are picked from a gateway and joined only along connections that work in practice, then sized to the client's night budget.
-
Fill each stop
A base lodge and roughly one experience per night, drawn from that hub's shortlist after anything closed, out of season or out of preference is filtered away.
-
Explain, flag and hand over
Every pick carries the reasons it was chosen. Anything missing is flagged for you to fill from experience, never papered over.
Distance is a poor guide to travel time.
Wellington and the top of the South Island look like neighbours on a map. Between them is Cook Strait, so getting across means a ferry or a flight, never a quick drive. A planner that routes by distance pairs them anyway. Cordovan routes on rules about how places actually connect, written by people who have made the journeys rather than estimated by a map service.
Every hub knows its island
A New Zealand hub without North or South Island recorded is rejected before it can reach a draft.
Stops connect only where travel works
Each hub lists the places it pairs well with, and how: a gateway, before, after or an extension. Routes move only along those links.
Remote places go through their gateway
Lord Howe and Kangaroo Island carry their own airport and the mainland hub that serves them. Road spokes like the West Coast route via Christchurch.
Getting there costs nights
A hub that is slow to reach spends part of the trip's night budget, so a short trip isn't stretched across too many stops.
Alternatives aren't stops
Places marked as substitutes for each other are offered as options, never as consecutive days in the same route.
Nothing in a draft is generic.
Generic planners work from a one-line prompt. Cordovan works from the structured travel profile you already keep for each client, with more than twenty fields across twelve sections. The draft reads it automatically, so every hub, lodge and experience is chosen for this traveler, not the internet's average one.
| From the profile | What it captures | How it shapes the draft |
|---|---|---|
| Travel months | Each month the client would travel | Removes anything closed for the season and favours places at their best |
| Next trip | Nights, budget, experience tier (Upscale, Luxury, Ultra Luxury), areas of interest | Sets the night budget the route is sized to, and the countries in scope |
| Trip focus | Adventure, relaxation, wellness, wildlife, culinary, culture, romance; lively or quiet | Becomes the travel styles every hub and experience is scored on |
| Activities | Walking, hiking, biking, wine, food, iconic sites, sailing, snorkelling | Reinforces which experiences rise to the top |
| Traveling party | Solo, couple, family, friends, multi-generational | Family trips leave out adults-oriented experiences |
| Lodging | Preferred and acceptable star ratings, lodging type, beds and rooms | Filters lodges by rating and ranks the preferred rating first |
| Touring style | Small groups welcome, private guides only, or no preference | Filters experiences by how they are guided |
| Transportation | Yes or no to helicopters, small planes, boats, trains, self-driving | A no removes everything that depends on it: no scenic flight for a client who won't fly small planes |
| Flights and bucket list | Departure airports, airlines, cabin, dream experiences | Sits beside the draft for you when booking and reviewing |
Where AI itineraries go wrong, and what Cordovan does instead.
Each common failure of AI-written trips maps to a specific safeguard.
| Failure | Generic AI planner | Cordovan |
|---|---|---|
| Invented places | Can blend real names into places that don't exist | Recommends only what is in the vetted catalog. Nothing comes from the model's memory. |
| Closed or stale venues | Training data lags by months or years | Every entry has a status, and anything inactive is filtered out. Catalog changes are versioned and reviewed. |
| Seasonal closures | Writes for a generic good day | Seasonal closures are filtered against the client's travel months |
| Overpacked days | Five highlights a day | Roughly one experience per night, around a base lodge |
| Illogical routing | Zig-zags to fit a list | Walks a hand-built map of real connections from a gateway, sized by recommended nights |
| Impossible transfers | Guesses connections | Transfers resolve from each hub's real gateway airport. A hub without one is flagged, not invented. |
| Can't be booked | No supply behind the plan | Every entry has a supplier or vendor behind it, or it never enters the catalog |
| Generic picks | The internet's top ten | Filtered and scored against this client's profile |
| Unexplained choices | "Trust me" | Each pick carries the reasons behind it, which become the rationale your client sees |
| Silent gaps | Fills holes with plausible filler | Missing lodging, thin supplier coverage and route gaps are raised as flags for you |
A draft for you, not a final plan for your client.
First-hand knowledge lives in the catalog
Which room faces the view, which guide is best, which places pair well and for how many nights: it is written down once and used in every draft.
Your own records sit alongside it
Add the properties and partners your agency knows. Your entries are private to your agency and drafted from just like the shared catalog.
Every change is reviewable
A bad entry is corrected in one place and gone from the next draft. There is no model to retrain.
Flags, not filler
Where the catalog is thin, you are told, and you fill the gap from experience instead of letting a model guess.
Honest about the edges.
A draft is only as good as the catalog behind it, and a few things are still yours to check.
Live availability
Drafts know seasons and status, not whether a lodge has rooms on your client's dates or a venue's daily hours.
Exact leg times
The connection rules decide which legs are sensible. Hour-by-hour drive and flight times between hubs are on the way.
Coverage
The catalog is deepest in Australia and New Zealand and still growing across the South Pacific.
Want to see a draft built for one of your own clients? Request a private demo, or read more in the Journal.