Coconut Travel
A tour-matching system: prices from a hundred suppliers in one place and an AI-written proposal in five minutes
Task
To prepare an offer for a client, a manager went through a hundred tour operator and supplier websites by hand: looking for suitable options, comparing prices and terms, then turning it all into a document. One tour took 5–6 hours, so the client waited until the evening or the next day, and a manager could only handle a few requests. The goal was to bring every source into one system and leave people with nothing but checking the finished result.
Solution
- Integrated every source the company works with: via API where there is one, by parsing websites where there isn’t, and by reading spreadsheet exports from suppliers who send price lists.
- Brought the data to one format: destinations, hotels, rooms, meal plans, dates and prices from different sources are compared side by side, and the same tour from different suppliers is merged.
- A manager enters the request (destination, dates, travellers, budget and preferences), and the system queries every source in parallel, in minutes.
- An AI agent analyses what was collected: it drops options that don’t fit the client’s terms, compares price and what each tour includes, and picks the best ones with a reason for each.
- From the chosen options the AI writes a commercial proposal with hotels, terms and prices in the company’s branding. The manager checks it and sends it to the client.
- If a supplier’s website changes or a source doesn’t respond, the search doesn’t stop: the system collects data from the rest and shows which sources were missing.
Before and after
The main change: managers no longer search and compare by hand. They enter a request and check the finished proposal.
Before
5–6 hours
Client request
The manager clarifies destination, dates, travellers and budget
Searching a hundred websites
Opens operator and supplier websites one by one
Comparing by hand
Puts prices and terms into a spreadsheet and rechecks availability
Building the proposal
Copies descriptions and prices into a document
Reply to the client
Hours later, or the next day
After
5 minutes
Client request
The manager enters the search parameters in the system
Data collection
The system queries every source: APIs, websites and spreadsheets
AI analysis
AI picks the best options by price and terms
Ready proposal
AI writes the proposal; the manager checks and sends it
Result
-
5 MIN per proposal
What it meansMatching and writing up one tour used to take 5–6 hours. Now clients get a proposal while they’re still in the conversation, and a manager can handle many times more requests.
-
100 sources in one place
What it meansAbout a hundred operator and supplier websites that managers used to check one by one are now queried automatically and compared.
-
3 integration methods
What it meansAPIs where available, website parsing and spreadsheet exports: we connected every partner the company works with, however they share their data.
Screens
Comparing options and the AI’s picks 
The finished commercial proposal
What shall we build together?
Tell us what you want to launch or improve. We’ll discuss the task and suggest the next step.