An advisor forwarded me one of these on Monday. Ten days in Spain, built by her client over a couple of evenings with a chatbot, and I want to say something before I say anything else: it was a good document. Seville, Cordoba, Granada, Barcelona. The Alcazar, the Mezquita, the Nasrid Palaces, the Sagrada Familia. Somebody had sat down after dinner and thought hard about this.
It also could not be built. The reason why is the most interesting thing I have read all month.
Let me be clear about where the failure actually sits, because I think our industry gets this wrong in a way that costs it clients. The problem is not that the model does not know Spain. It knows Spain roughly as well as a good guidebook knows Spain, and I have made a living alongside good guidebooks for years without resenting a single one of them. The problem is that knowing Spain and building a day in Spain are two different skills, and only one of them requires having stood there.
An itinerary has bones and it has sinew. The named places are bones. Everything that holds them together is the other thing, and the other thing is judgment.
Granada to Barcelona, in an afternoon
The day I watch fail most often is the one that cannot exist. Morning at the Alhambra, evening tapas on Carrer de Blai. On a map it looks like a fair ask. Two Spanish cities, a fast train, plenty of daylight.
The fastest direct AVE from Granada to Barcelona runs six hours and nineteen minutes, and there are usually about two direct departures a day. Add the transfer to the station, the boarding window, the walk out of Barcelona Sants with three suitcases. The afternoon has eaten the whole day, and the morning with it. Nobody really sees the Alhambra on a morning like that. They are watching the clock from the Court of the Lions onward, which is a shame, because the Court of the Lions deserves more attention than anyone gives it while checking a phone.
I do not blame the machine for this. It had the distance and a true sentence about Spanish rail and it put them together, which is more or less what I would have done at twenty-five. What it did not have is any feeling for what a day costs the people living inside it. That is not a data problem. That is a having-been-tired-in-a-train-station problem.
The fix takes about ten seconds once you know it. Granada and Barcelona are not neighbors. A program that wants both needs a night in between, or a flight, or a different second city. I tend to reach for the flight.
The hours nobody publishes in a form you can look up
Opening times are published. The shape of the local week is not, at least not anywhere a planning tool can go and fetch it.
Plenty of municipal museums close on Mondays. The Mezquita restricts general visits during Sunday morning Mass, as it should. In a smaller Andalusian town the one kitchen worth sitting down in does not open until half past two, and by four the street has gone quiet again. In Madrid the dining room is dark until nine, and a generated plan will put dinner at seven thirty with total confidence.
None of that is dramatic. Each one is just the difference between a day that flows and a day your client spends standing outside something, being patient, doing arithmetic about what they paid.
Here is the part I actually enjoy, though. The local week is not an obstacle. It is the good news. Lunch starts at half past two because lunch is a real meal here and people take it seriously, and a client who learns that on day two is eating like a Spaniard by day five. That is a better trip than the one on the spreadsheet. You just have to build for it on purpose.
August is a shape, not a verdict
Advice about August in Spain comes in two flavors online. Go, or do not go. I would like to retire both.
Heat does not usually change where I send people. It changes the order of the hours. The Alhambra early, the inside of the Mezquita through the worst of the afternoon, the Albaicin after seven when the stone starts giving light back instead of heat. Run an August day in that order and it is a good day. Run it in the order a spreadsheet produces and it is a slog.
Then there is the question of who is doing the walking, which no document has ever answered for me. A couple who nap after lunch will do beautifully in that shape. Three generations moving as one unit, with a stroller at one end and an eighty-year-old at the other, will not, however cleverly I arrange the hours. That call is about the people, not the weather, and I cannot make it until somebody tells me about the people. There is more on this in how Spain's calendar should bend an itinerary.
The room that matches the list and not the family
Hotels are where the missing layer shows most clearly, and I say this with sympathy, because the recommendations are usually excellent. Famous, well reviewed, correctly located, often lovely. And frequently wrong for the five people in question.
Picture a restored sixteenth-century house in central Seville. Fourteen rooms, no two alike, several of them up a flight of worn stone with no elevator. For a couple on their second trip to Spain it is close to perfect and I would put them there myself. For a family of five who need connecting rooms, one of whom had a knee replaced in March, it is a beautiful building and a week of small daily frustrations.
The question was never which hotel is best. It is which hotel is right for these people, this week, at this point in their lives. Change one fact about the party and the answer changes with it. I find that reassuring, honestly. It means the job is not going anywhere.
Inventory that is not there yet
A generated itinerary will hand you a nine o'clock Nasrid Palaces entry as though it were settled. It is not settled. It is not yet a fact. The Alhambra releases tickets ninety days out and the best slots close long before that window runs down, and Sagrada Familia opens at sixty with its own pressure on top. I went through the mechanics in how timed-entry ticketing actually works.
The subtler issue is not the sell-out risk. It is that the document sounds equally certain whether a line was verified, guessed, or invented. Nothing on the page flags which one still needs a human to go and check. Every line reads settled, and settled is exactly the quality that makes it risky in a client's hands.
What I can do that the document cannot
None of this is an argument against the tools. I use them daily and they have made parts of my week better. And the client who turns up with a generated draft has handed their advisor something valuable: an unusually detailed account of what they want this trip to feel like. Which cities carry weight. How fast they think they want to move. What kind of mornings they are picturing.
That draft is a brief. It was never an itinerary. Read it that way and it becomes the best client questionnaire you have ever been handed, filled out voluntarily, at length, for free.
Which leaves the advisor as the verification layer and puts me in the interpretation seat, where I am glad to sit. What I can do from here is stand in the place, in the week, and know that the early Alhambra slot is worth the alarm in October and not optional in July. That this particular family will spend two hours in a Tuesday market and be finished with cathedrals after the second one. That the driver who suits a couple's slow day in the Sierra is not the one I send with a group of sixteen. Look through our sample programs and the same logic sits under every routing. The sequence exists for a reason, and the reason is almost always something that happened to somebody.
Spain rewards this kind of attention more than most places I have worked. The best part of a day here is usually the part nobody scheduled, and getting a client to that part takes someone who has stood in it and understood why it mattered.
So send me the draft. Forward the whole thing, dates included. I will tell you which days hold, which ones do not, and what I would put in the gaps. If you are still working out how you want to partner with a ground operator in Spain, the advisor page lays out the model plainly.