How Travel Agents Can Use AI to Write Itineraries Faster (Without Sounding Generic)

The writing problem at scale

An agency handling 30 active proposals has a writing problem. Every proposal needs a destination overview, hotel descriptions, day-by-day activity write-ups, and context-setting about why this particular itinerary suits this particular client. That's hours of writing per proposal — largely for content you've written before, in slightly different forms.

AI writing tools solve this problem well — when used correctly. Used carelessly, AI produces output that reads like a travel brochure from 2005: generic, hollow, full of "breathtaking" sunsets and "unforgettable" experiences that mean nothing to anyone.

Used well, AI produces a strong first draft that you personalise in 5 minutes rather than writing from scratch in 30.

The right mental model: AI writes structure, you add expertise

The mistake most agents make when using AI is asking for the finished product: "Write me a description of the Raffles Hotel Singapore." What comes back is technically accurate but reads like Wikipedia. Anyone could have written it. It has no personality, no recommendation, no reason a client should trust your specific choice.

A better mental model: AI writes the skeleton, you add the flesh. Use AI to:

• Get the factual structure right (property highlights, location, key amenities)
• Identify the right emotional beats (romance, heritage, exclusivity, calm)
• Produce the first paragraph you can then sharpen

You then add: your personal experience with the property (or a colleague's), the specific room category you recommend for this client, the detail only an agent who's done a FAM trip knows, and the sentence that makes the copy feel like advice from a trusted friend rather than a directory listing.

Prompts that produce better output

The quality of AI output is almost entirely determined by the specificity of the prompt. These patterns produce more usable results:

"Write a 3-sentence hotel description for [property name] in [destination]. The client is a couple celebrating their 25th wedding anniversary. Tone: warm and evocative. Focus on privacy and romance, not facilities."

"Summarise Day 3 of an Italian Riviera itinerary: the clients leave Portofino after breakfast, take a ferry to the Cinque Terre, spend the afternoon in Vernazza, and return to Santa Margherita for the night. Write as a day narrative, 100 words, from the traveller's perspective."

"Write a destination intro for Kyoto for a proposal being sent to a family of four with two teenagers aged 14 and 17. Open with something that makes the teenagers want to go, not just the parents."

The specificity — client type, emotional goal, word count, perspective — is everything. Vague prompts produce generic output. Specific prompts produce first drafts you can actually use.

Travesy's built-in AI writer

Travesy includes a Claude-powered AI writer built directly into the proposal editor. Rather than switching between your proposal builder and a separate AI tool, you can generate hotel descriptions, destination intros, and day narratives in context — and save approved copy directly to your Knowledge Library for reuse.

The tone controls (formal, friendly, luxury, adventure) let you match the output to the style appropriate for the specific client, without rewriting the prompt each time.

The review step you cannot skip

AI tools — including Claude, ChatGPT, and others — occasionally invent details. This is the most important thing to understand about AI-generated travel content:

• AI will describe amenities that don't exist. "A rooftop infinity pool" at a property that doesn't have one.
• AI will state a restaurant is Michelin-starred when it isn't.
• AI will describe a train route as daily when it's seasonal.

These errors are stated with complete confidence — the same tone as accurate information. Before sending any AI-assisted copy to a client, verify:

• Property details (amenities, room categories, star rating)
• Activity descriptions (opening times, access restrictions, seasonal availability)
• Transport claims (route existence, frequency, duration)

This isn't onerous if you know the destination. You're not fact-checking from scratch — you're catching confident inaccuracies that could embarrass you in front of a client.

Building a content library that compounds over time

The highest-leverage use of AI writing is not generating individual descriptions — it's building a reusable content library. Once you've produced and verified good copy for a destination or property, save it to your knowledge base. Don't write the same Bali temple description 15 times.

After 3 months of saving consistently, most agents find that 80% of what they need for a new proposal is already in the library. AI is only needed for the personalisation layer and the occasional new destination.

This is the compound effect of good content practice: each proposal you write makes the next one faster, until building a new proposal takes 15 minutes instead of 3 hours.