Why a chat layer on top of your current system won't quote accurately
An artificial intelligence assistant built on top of a management system improves the speed of writing and lookups, not the quality of the quote. It works on the rates that are already loaded: if they are incomplete, expired or poorly modeled, what you get is a wrong proposal written much faster.
Why it matters
Practically every management system in the industry is adding a conversational assistant layer, and the promise being communicated is the same: quote by talking. It is a real improvement, but it solves the stage of the process that takes the least time.
According to Outtrip's survey of nearly 1,000 inbound travel agencies, conducted through August 2026, quote turnaround time is concentrated in finding the valid rate and confirming that it applies, not in writing the proposal. Speeding up the writing of a stage that was already the shortest does not move the final number.
What exactly does an assistant built on top of a system do?
It does three things, and does them well:
- Looks up information that is already in the system, in natural language instead of with filters.
- Writes text based on that information: descriptions, itineraries, emails.
- Assists the person quoting, by suggesting or filling in.
What all three have in common: they operate on existing data. The assistant does not control how that data got in, whether it is still valid, or whether the way it is stored correctly reflects how the supplier prices.
What is the real bottleneck?
Having the correct rate available at the moment of quoting. It sounds obvious, and it is where almost everything breaks.
An inbound tour operator's rates arrive however they arrive: a PDF from the hotel, an Excel file from the transport provider, an in-house spreadsheet with tailor-made services, an email with a rate negotiated for a specific group. In many markets in the region, the rate sheet is renewed every three months. If loading those rates is not done in bulk and is not connected to the way they actually arrive, the agency lives out of date — not because of sloppiness, but because of arithmetic: there simply is not enough time to go through record by record.
A conversational assistant does not touch that problem. It can read what is there faster, but it does not change how much of what is there is still valid.
Why doesn't this happen in other industries?
It is a comparison worth making, because it explains why AI felt immediate in other work tools and not here.
Adding an assistant to a note-taking or task management tool is relatively straightforward: the operations are few and simple, and the cost of an error is low. Adding a poorly written task has no financial consequence.
In quoting, the volume of data is much larger, each piece of data has to be handled with its own rules — a transfer rate is not calculated like a hotel rate — and errors have a price. The more complex the operation the AI has to act on, the less it is enough to just put a chat in front of it.
How to tell a copilot from a system that quotes
| Copilot on top of the system | System that quotes | |
|---|---|---|
| What it operates on | Data already loaded | Also how data gets in and is maintained |
| What it speeds up | Lookups and writing | The entire process |
| If the rate has expired | Quotes with the expired rate | Should stop and flag it |
| If the rate does not exist | May fill in the gap | Should state that it is missing |
| What improves as the agency grows | Convenience | Operational capacity |
Neither category is bad. They are different things at different prices, and it pays to know which one you are buying.
Three questions for a demo
- How do new rates get into the system? If the answer is "they get loaded," ask how many records per hour and multiply that by your catalog, for every rate sheet renewal.
- Where does each number in this proposal come from? Ask them to show it in the demo, on a specific service.
- What happens if a rate has expired or does not exist? The right answer is that the system says so. If instead it returns a number anyway, that number is an estimate dressed up as a price.
What we are not saying
That conversational assistants are useless. They are useful, and an agency that today searches its system with filters and menus will work better with one. Nor that a traditional management system is bad: it handles operations, which is a different and equally real problem.
What we are saying is narrower: speeding up the writing of a proposal does not speed up the quote, because that is not where the time was. And if the price shown in that proposal came from a rate sheet nobody has updated in four months, speed works against you — the wrong proposal reaches the client sooner.
Frequently asked questions
Is it worth adding an AI assistant to a travel management system?
It helps speed up writing and lookups on data that is already loaded and up to date. It does not solve the quoting problem if the bottleneck lies in how rates get loaded and maintained, because the assistant works with whatever is in the system.
What is the real bottleneck in a quote?
It is not writing the proposal but having the correct, valid rate available at the moment of quoting. In many markets in the region, rate sheets are renewed every three months and arrive in scattered formats.
What is the difference between a copilot and a system that quotes?
A copilot works on information that is already loaded and assists the person quoting. A system that quotes also controls how that information gets in and is maintained, which is where it is decided whether the price is correct.
What should you ask a vendor offering AI for quoting?
How new rates get into the system, where each number in the proposal comes from, and what happens when a rate does not exist or has expired.
Outtrip AI is the quoting, itinerary and booking platform for DMCs and inbound travel agencies.