Everything your agent learned about your life is calibrated to your home environment: your grocery store, your neighborhood, your routines, your kitchen. Travel strips all of that context away and forces your preferences to operate in unfamiliar territory, which is where most AI travel tools fail because they optimize for popularity rather than for you.
Michael Tiffany
Every article in this series so far has been teaching your agent about your life at home. Travel is what happens when that entire knowledge base collides with an unfamiliar environment, and it's the domain where the collision reveals exactly how much your agent has actually learned. If you've taught your agent about your food restrictions, does it know to research restaurant options in a foreign city where the staff may not speak your language? If you've taught it about your health, does it know that your medication needs to be refrigerated and that you'll need a hotel with a mini-fridge? If you've taught it about your fitness routine, does it know that you'll want a running route near whatever hotel you're in? MIT researchers recently demonstrated an AI travel planning system that treats itinerary generation as a constraint satisfaction problem, and I think that's the right mental model: travel planning is the act of satisfying your personal constraints in an unfamiliar constraint space.
For families with food allergies or intolerances, travel is especially fraught. A scoping review in Tropical Diseases, Travel Medicine and Vaccines found that food-allergic travelers face elevated risk during transportation and at unfamiliar restaurants, and a separate survey found that 14% of people with food allergies avoid traveling entirely because of it, while 67% of those who do travel report difficulty eating at restaurants and 30% experience an allergic reaction during their trip. Your agent should be the system that prevents those reactions by holding your dietary constraints, translating them for unfamiliar food environments, and helping you find safe options before you're standing hungry on a foreign street corner.
I think the useful way to frame travel preferences is to separate the knowledge your agent already has into things that are portable and things that are location-dependent.
Your dietary restrictions travel with you everywhere and become more important, not less, when you leave the environment where you've already solved the food problem. Your medication schedule travels with you and acquires new logistical requirements (time zone adjustments, refrigeration needs, carrying documentation through customs). Your sleep preferences travel with you and collide with hotel rooms that have the wrong pillows, the wrong blackout situation, and a thermostat you can't control. Your fitness habits travel with you and need to adapt to whatever equipment and terrain are available.
Your grocery store, your kitchen, and your neighborhood knowledge are all useless in a new city. Your service providers can't help you when you're a thousand miles away. Your chore routines are suspended. The entire infrastructure of your daily life stays behind, and your agent needs to help you rebuild a temporary version of it at your destination using only the portable preferences plus whatever it can learn about the local environment.
This is why teaching your agent about travel is less about stating new preferences and more about testing whether the preferences you've already taught are robust enough to survive displacement. A well-taught agent shouldn't need a separate "travel profile"; it should be able to infer your travel needs from what it already knows about you.
Before any trip, ask your agent to generate a travel brief that draws on everything it knows. The brief should cover at minimum the domains that are most affected by displacement.
For food: "Based on your dietary restrictions, here are the constraints that apply at your destination. I've identified three restaurants near your hotel that accommodate gluten-free diets and have English-speaking staff. The hotel offers breakfast but the continental option is primarily bread and pastry; you may want to plan around that. I've drafted an allergy card in the local language that you can show to restaurant staff."
For health: "Your lisinopril dosage stays the same but you'll cross two time zones, so consider shifting your dose gradually. Your EpiPen should be in your carry-on, not checked luggage. The nearest hospital to your hotel is [enter your name], about eight minutes by taxi."
For wardrobe: "The forecast shows daytime highs around 28°C and evening lows around 15°C. Based on your wardrobe preferences, you'll want layers. You mentioned you avoid synthetic fabrics, so the hiking base layer should be merino."
For fitness: "There's a running path along the waterfront about a ten-minute walk from the hotel. The hotel gym has free weights but no rowing machine, which you've mentioned preferring. Your knee has been bothering you on long runs, so you might consider shorter routes with less pavement."
The quality of this brief is a direct test of how well you've taught your agent across all the previous articles. If the food section is generic ("look for gluten-free restaurants"), your food article needs more detail. If the health section is missing ("I didn't know about the EpiPen"), your health information is incomplete. The travel brief exposes gaps in your agent's knowledge that wouldn't surface at home, where your existing infrastructure compensates for what the agent doesn't know.
After the trip, spend five minutes telling your agent what worked and what you'd do differently. This is the feedback loop that makes the next trip better, and travel debriefs tend to be especially rich because displacement surfaces preferences you didn't know you had.
"The hotel was fine for sleeping but the shower pressure was terrible, and I realized I care about that more than I thought. The restaurant the agent found was genuinely safe for my allergies, which was a relief, but the portions were small and I was hungry by 9pm, so next time I need to plan for a late snack. The running route was beautiful but hilly, and my knee felt it the next day. I packed too many pants and not enough lightweight tops; the weather was warmer than expected and I changed shirts twice a day."
That debrief teaches the agent several new things: a hotel preference you'd never articulated (shower pressure), a meal-timing issue specific to travel (small restaurant portions require supplementary food), a fitness adjustment (reduce hill running when traveling), and a packing refinement (pack for warmer-than-forecast conditions). These observations are all specific to travel and would never surface during normal life at home.
The first trip with an informed agent will be decent. The third trip will be noticeably better. By the fifth trip, your agent has accumulated enough destination-specific knowledge and enough understanding of your travel-specific quirks that the pre-trip brief starts to feel like it was written by someone who knows you well.
Your agent will learn that you prefer aisle seats on flights under four hours but window seats on overnight flights. It will learn that you always forget your phone charger in hotel rooms and should be reminded at checkout. It will learn that you sleep poorly the first night in a new place and shouldn't schedule early meetings on arrival day. It will learn that "vacation you" has different food rules than "Tuesday night you" and that you'll eat pasta in Italy even though you're gluten-free at home, and that the consequences are worth it to you in that context.
These are exactly the kind of context-dependent, contradiction-rich preferences that the food article described at the very beginning of this series. Travel is where they're most visible, because travel is where your routines are most disrupted and your preferences are most tested.
How is this different from AI travel planning tools? Most AI travel tools optimize for destinations and activities based on aggregate popularity data. Your agent optimizes for your constraints and preferences, which are personal and often invisible to platform-level algorithms. The MIT constraint satisfaction approach is closer to what your agent does: treating your preferences as hard and soft constraints and finding solutions that satisfy them in an unfamiliar environment.
Should I teach my agent about business travel separately from personal travel? The underlying preferences are mostly the same (dietary needs, health requirements, fitness habits), but the contexts differ: business travel has schedule constraints, dress code requirements, and client-facing obligations that personal travel doesn't. Tell your agent about both, and it will learn to generate different briefs depending on the trip type.
What about traveling with kids? Multiply every constraint by the number of children and add the kids' article requirements to the travel brief. Your agent should know each child's dietary restrictions, medication needs, comfort items, and behavioral patterns, and all of those become harder to manage on the road. The packing list alone benefits enormously from an agent that knows what each child actually needs versus what you'd pack out of anxiety.
How do I handle destinations I've visited before? If you debriefed after the previous visit, your agent already has destination-specific knowledge: which restaurants worked, which hotel to avoid, what the transit system is like. Returning to a familiar destination is where the compounding effect is strongest, because you're building on observed experience rather than starting from research.
Copy and paste the prompt below into your AI agent to get started.
If your agent produces a detailed, personalized brief, the series is working. If it produces something generic, the gaps it flags are your roadmap for which earlier articles to revisit.
Based on everything you know about me (my food restrictions, my health needs, my fitness routine, my wardrobe preferences, and my travel history), generate a pre-trip brief for my next upcoming trip. Include food safety considerations at the destination, health and medication logistics, a packing suggestion based on the weather forecast and my wardrobe preferences, and fitness options near where I'm staying. Flag anything you don't know that you'd need to make the brief more useful, so I can fill in the gaps.
Fulcra was designed by people who get privacy and know the importance of an infrastructure solution that can be the secure private datastore for the rest of your life. Here data is yours, under your control, and only shared with the people and tools you choose to share it with.
