AI import, drop in the PDF

A supplier sends you a 14-page programme as a PDF. Day by day, hotel by hotel, with a price table at the back. Someone on your team now spends the afternoon typing it into your system: day one, the hotel, the price per room, the transfer. Then day two.

How much of your week goes into copying?

I'm Eric, from TripCreator. In episode one I promised that moving into TripCreator wouldn't be a typing project, and I asked you to hold me to it in episode nine. This is episode nine. Today is about AI import: how TripCreator reads a document and turns it into a trip.

Over the next twenty minutes I'll explain how it works in four steps, walk through a real-world style import, spend some time on the design decisions that make it safe to use with prices, give you a checklist for reviewing an import well, and cover how operators use it to bring years of old itineraries into TripCreator.

Travel businesses are surrounded by documents that are almost a trip. A supplier's quote. A partner's programme. A ground handler's proposal. Your own old itineraries, sitting in Word files from before you had a proper system.

All the information is there: the days, the hotels, the transfers, the activities and the prices. But it's in a document, not in your system. So someone has to copy it across, line by line. It's slow, it's dull, and it's exactly the kind of work where mistakes creep in.

AI import does that copying for you, in four steps.

Step one is to upload what you have. That might be a supplier quote, a partner's programme or an old itinerary from Word. PDFs work, and so do documents in other languages. A programme written in French or German doesn't need translating first.

You don't need to clean the document up or put it in a special format. The point is to work with the documents you actually receive.

Step two is the AI reading the document. It finds the structure of the trip: which day is which, and which hotels, transfers, activities and flights belong to each day.

It also reads the prices, and that's the hard part. A price on its own means nothing. It matters whether it's per person, per room or per night, and what it includes. A hotel at 180 euros per room per night is a very different number from 180 euros per person for the stay. The AI reads the price together with what it applies to, because a misread price unit is the kind of error that costs real money.

Step three: you get a draft trip, and it's not a summary or a list. It's an actual itinerary, laid out day by day in your planner, just as if someone on your team had built it.

Where the document mentions a hotel or a supplier you already have in your product library, the draft links to your existing product. So the trip connects to your own data from the start.

It also brings in images: from the document itself where it has good ones, or from the map where it doesn't. So the draft already looks like a trip, not a spreadsheet.

Step four matters most: you review it and save it.

Nothing goes live until you've checked it. The draft sits in your planner, and you go through it like you'd check a colleague's work. If a line is wrong, you fix it and carry on. If a day is missing, you add it. Then you save.

The AI does the copying. You keep the judgement. That's deliberate.

Let's walk through an example. A partner in France sends you a nine-day programme for a small group, written in French, as a PDF. Fourteen pages: a cover, a day-by-day section, hotel descriptions, and a price table at the back that gives a per-person price in a twin room, a single supplement, and a list of what's included.

You upload it. While the AI reads it, you get on with something else.

When you come back, there's a draft trip in your planner. Nine days. Each day has its hotel, its transfers and its activities, with the descriptions taken from the programme. The two hotels you already work with are linked to your existing products. The others are new, with their descriptions and photos from the document.

The prices are on the lines. The per-person twin price and the single supplement have been read as what they are, not mixed up. What's included is noted.

Now you review. You notice the farewell dinner on day nine has no price, because the document said "included" in one place and didn't price it anywhere. The draft has left it empty rather than guessing. You check with your partner, fill it in, add your markup, and save.

That afternoon you send your customer a proper itinerary page, under your own brand, built from a document that arrived that morning in another language.

Now, the design decision I want to spend real time on, because it's the one I'd ask about if I were you.

AI tools are known for making things up confidently. In most settings that's an annoyance. In travel pricing it's a disaster. An invented price becomes a quote, a quote becomes a booking, and you've sold something at a loss.

So TripCreator's AI import reads, and it doesn't invent. Prices come from your document. If a price isn't in the document, the line is left empty for you to fill in, like the farewell dinner in our example. It isn't guessed, estimated or rounded. An empty line is easy to spot and fix. A plausible wrong number is not.

That principle runs through all of TripCreator's AI, and it'll come up again next episode.

A question I hear a lot: "Couldn't I just paste the PDF into a general AI chatbot?"

You could, and you'd get text back. Maybe a nice summary. But three things would be missing.

First, it would be text, not a trip. You'd still have to build the trip in your system, line by line. The copying problem wouldn't be solved, just moved.

Second, it wouldn't know your library. It couldn't tell that the hotel on day three is a hotel you already work with, at rates you've already negotiated.

Third, and most important, a general chatbot is happy to fill gaps. Ask it for a price that isn't in the document, and it may well give you one. For travel pricing, that's exactly what you don't want.

AI import is built for one job: turning the documents travel businesses receive into trips in their own system, without inventing anything.

The second design decision is about what happens after the import.

The products found in a document, like a hotel, a transfer or an excursion, can be saved to your product library. The next time you build a trip with that hotel, it isn't another import. It's your own product, with your own data.

So every import makes your library a bit richer. Over time, the trips you build rely less on documents and more on your own catalogue, which is exactly where you want to be.

Since the review is your part of the process, here's a checklist for doing it well. It takes a few minutes, and it's worth doing every time.

One. Count the days. Does the draft have the same number of days and nights as the document? Are the dates right?

Two. Check the price units. For each priced line, is it per person, per room, per night or per group, as the document says? This is the single most valuable thing to check.

Three. Look for the gaps. Any line without a price is a line the document didn't price. Decide what to do with each one.

Four. Check the matches. Where the draft has linked a line to one of your existing products, is it the right product? Same hotel, right room type?

Five. Read what's included. Breakfasts, transfers, guides, entry fees. Make sure what the document includes is what your customer will be told is included.

Then add your markup and save. If you do those five things, you can send the result with confidence.

Now, the onboarding story. When you move to TripCreator, you probably have years of old itineraries in Word files and PDFs. Those are some of your most valuable assets: your best routes, your best descriptions, your best supplier combinations.

With AI import, those come in as trips rather than staying as files. Here's an approach that works well.

Start with your best sellers. Pick the ten or twenty itineraries you sell most often and import them first. Review each one carefully, save the products to your library, and turn the best ones into templates.

Then let the rest come in as you need them. When a customer asks for something you did three years ago, import that old itinerary, review it, and use it as the starting point.

After a few weeks, your library holds the products you actually use, your templates hold your best routes, and the old files have quietly become a working system.

The other way operators use AI import is day-to-day quoting.

A supplier sends a programme in the morning. You import it, check it, add your markup, and send the customer a proper itinerary page the same day, under your own brand and with your own pricing.

For a DMC, that means a partner's programme becomes your proposal in an hour. For an agency, it means a supplier's PDF becomes something you can actually sell. For anyone, it means less copying.

To be clear about expectations, AI import isn't magic. A badly scanned document with half the prices cut off will give you a draft with gaps. That's the right result, because the gaps are real. And it doesn't replace your review. It replaces the typing that comes before the review.

That's a good trade. The hours go into checking and improving the trip, not transcribing it.

Here are the questions operators usually ask about AI import.

"What kinds of documents work?" Supplier quotes, partners' programmes and old itineraries, as PDFs or Word documents. Documents in other languages work too.

"Do I have to check everything?" You should review every import before you save it, using the checklist from earlier. It takes minutes, not hours.

"Will it get the prices right?" It reads the prices that are in the document, together with what they apply to. Where a price isn't there, it leaves the line empty. That's why checking the price units is the first thing on the review checklist.

"Can I import the same supplier's documents every year?" Yes, and each time the products it finds can be matched to the ones already in your library, so your data gets better rather than more cluttered.

To set realistic expectations, let me compare two imports: one from a clean document, and one from a messy one. Most of what you receive will sit somewhere between the two.

The clean document is a well-made supplier programme. Each day has a heading. Each hotel is named, with its room type. The price table at the back says clearly what each price covers: per person sharing a twin room, the single supplement, the child reduction, and a list of inclusions. When you import a document like that, the draft trip is usually very close to complete. Your review is quick: count the days, check the units, confirm the matches, add your markup.

The messy document is the kind every operator has received. It's a scanned printout of an old programme, slightly crooked. Some days run into each other. The prices are scattered through the text: one hotel priced per room, another per person, a lump sum at the end for "all transfers", and a few items described as "on request". There's a handwritten note in the margin.

When you import a document like that, the AI still does the heavy lifting. You get the days, the hotels and most of the activities laid out in your planner. But you'll see more gaps. The "on request" items have no price, because the document doesn't give one. The lump sum for transfers may need splitting the way you want it. The handwritten note may not have been read at all.

Here's the important point. The messy import still saves you most of the typing, and the gaps it leaves are real gaps in the document, not mistakes made by the AI. They're exactly the things you'd have had to chase anyway. The difference is that now they're clearly marked, rather than hidden in a page of text.

So the rule of thumb is simple. The cleaner the document, the less there is to review. But even the messiest document comes in faster than you could type it, and it never comes in with invented prices.

I mentioned that every import makes your product library richer. I want to come back to that, because over a year it changes how your business works.

In the first month, most of your trips start from documents. You import a supplier's programme, review it, and save the hotels, transfers and activities to your library.

By the third month, something shifts. When you import a new programme, more and more of its hotels and activities match products you already have. The draft links to them, with your descriptions, your photos and your negotiated rates. Your review gets faster, because the matched products are ones you've already checked.

By the sixth month, many trips don't need an import at all. You build them directly from your library, or copy a template, because the products are already there. Imports become the exception: a new supplier, a new region, a partner's unusual programme.

That's what I mean by a library that compounds. Each import isn't just one trip. It's an investment in every future trip that uses the same products. And because the products in your library are the ones the AI chooses from when it builds a trip from a brief, which is next episode's topic, a richer library means better drafts too.

So think of AI import as two things at once: a way to stop typing today, and a way to build your catalogue for tomorrow.

Today's takeaway: the documents you already receive are almost trips. Upload them as they are, let the AI turn them into a draft itinerary with days, products and prices, then review it and save it. It reads and doesn't invent, so a missing price stays missing until you fill it. Review with the five-point checklist. And every import makes your product library richer.

So, as promised in episode one: moving in is no longer a typing project.

Want to try it on a real document? You can start a trial and import one of your own supplier documents. The link is in the show notes. Or send us a supplier PDF you're tired of re-typing, with the prices blanked out if you like. We'll import it and send you back the itinerary link.

Next time is the final episode. There's no document at all this time, just a sentence from a customer, and TripCreator builds the trip from it.

I'm Eric. Thanks for listening.