“Map”. “Deals”. “Market”. So many of the questions Enea receives look like this: one or two quick words, straight out of a search box. There is nothing wrong with writing this way — it’s just a 25-year-old habit people carry with them. Except that this box isn’t a search engine.
Many conversations end right there, at the very first question: you arrive, you ask, you take, you leave. And that’s a pity, because anyone who stops there is using a concierge like a vending machine — and missing out on the best part.
A habit that isn’t anyone’s fault
Twenty-five years of search engines have taught us to write using keywords: “gaeta parking”, “beaches”, “august events”. Then ChatGPT and Gemini arrived and taught us commands: “write me”, “create for me”, “list for me” — and if you don’t like the answer, you try again, because sooner or later a different one comes out.
Two consistent habits that work brilliantly with the tools that created them. The thing is, Enea is doing a completely different kind of job, and no one has ever explained how to talk to a digital concierge. This article aims to do just that.
Enea is not Google. And it’s not ChatGPT either
Let’s look at this both ways, because honesty here benefits everyone. Large generalist models know the world: they write an email for you, summarize a contract, give you coordinates for anything on demand. Enea does none of this, and has no desire to.
On the other hand, it knows just one single city, but knows it inside out: it knows which attractions are closed on Mondays, which parking lots don’t exist in winter, and what time the shuttle to Serapo runs. Information that no generalist model possesses — not because it’s stupid, but because it isn’t on the Internet: it lives in an archive we compile by hand, entry by entry.
So the rule is simple: for the wider world, feel free to use large models, they are fantastic. For Gaeta, they know what Wikipedia knows. Enea knows what the locals know.
The question that is worth twice as much
Here is the one thing almost nobody does, and it’s the thing that changes everything: share something about yourself. That you’re traveling as a family. That you have a dog. That you don’t have a car. That you arrive on Monday and leave on Friday. A search engine has no use for these details; a concierge works with them. It’s the whole difference between asking “parking” and asking “where do I park for Montagna Spaccata if I arrive around 10:00 with a stroller?”. The first gives you a list. The second gives you an answer.
«Parking»This gives you a list
«Where do I park for Montagna Spaccata? I arrive around 10:00 with a stroller»This gives you an answer
«Make me an itinerary»This gives you a list
«I have two days, I’m staying downtown and don’t have a car: beach and historic center»This gives you an answer
«Beaches»This gives you a list
«A day combining the beach and history, with a small dog»This gives you an answer
«Restaurants»This gives you a list
«Tonight I’d like to eat somewhere traditional, there are four of us, in the Serapo area»This gives you an answer
«Events»This gives you a list
«What is there to do tomorrow evening with 15-year-olds?»This gives you an answer
We didn’t write the questions on the right: they are real questions received by Enea, paraphrased here. The people who asked them received tailored days planned just for them.
Why the map doesn’t always show up
With generalist models, the contract is simple: you command, it delivers. Ask for a map, you get a map. Enea works under a different contract: it answers your question, it doesn’t just execute commands. If answering requires a map, the map appears automatically — along with the route, stops, and schedules. If it doesn’t appear, that’s not a flaw: it just means the answer called for something else.
A concierge that brings you whatever you ask for, in the exact order you requested it, isn’t a concierge — it’s a vending machine. Enea decides what to show you based on what you need, not based on what keywords you mentioned: that is why a context-rich question works much better than a command.
Repeating the question won’t make it change its mind
It often happens that you see the exact same question repeated two, three, five times in a row. This, too, is an imported habit: with generalist models, trying again “works” because with every attempt a slightly different answer comes out, and sooner or later one comes up that you like. Not with Enea, and the reason is structural: the source is always the same — the archive, the real schedules, the actual listings. The wording may change, the answer does not change. Because it is the right answer the first time.
And when Enea says that something isn’t there — an opening hour it doesn’t know, a venue that doesn’t exist, an event that isn’t listed — that isn’t a fault to work around by rephrasing: it’s information. Perhaps the most valuable information an assistant can give you, because out there, some will invent an opening hour just to avoid telling you “I don’t know”.
How to talk to Enea, in five lines
- Write as you would speak to a local, not as you would type into Google.
- Tell it who you are and what you need: children, a dog, no car, how long you’re staying. That is where it becomes a concierge.
- Ask about Gaeta, not the world: generalist models are there for the world.
- Let it choose the tool: maps and itineraries arrive when they serve the answer.
- Do not repeat the question: if it says something isn’t there, that is how it is. And you are better off knowing it.
The extended version, with all the examples, lives on a dedicated page: How to talk to Enea.
Frequently asked questions
What is the difference between Enea and ChatGPT or Gemini?
They do different jobs. Generalist models know the world and execute commands: they write, summarize, generate. Enea knows just one city — real opening hours, closures, shuttles, parking options, mapped by hand — and works as a concierge: it responds to your needs, rather than executing orders.
Why does Enea sometimes not show the map or itinerary I asked for?
Because it decides what to show you based on what you need, not based on what you named. If the map serves the answer, it appears on its own. It is a design choice: a concierge chooses the right tool, a vending machine just executes.
If I repeat the same question, does Enea answer differently?
No: the wording may change, but not the substance. The source is always the database with real city data, so the answer remains the same — because it is the right one. If Enea says something isn’t there, repeating the question won’t make it appear.
How do you ask Enea a good question?
By giving it context, just as you would with a human concierge: who you are, how long you’re staying, how you get around, what you like. “Two days, no car, with a 3-year-old child” produces a tailored day; “what to see” produces a list.
Now you know how it’s done. Ask Enea a concierge question — tell it about yourself, and see the difference.
