Quicktext review: good chatbot, integration gap persists

I first reviewed Quicktext and came away with a specific conclusion: the chatbot is competent, the booking conversion is real, and the integrations are not what the marketing claims they are. As director of technology for a hotel group in France (about fifteen properties), I’ve spent another few months with the product, tested their new WhatsApp channel, and tried to make sense of their Quinta branding. The score stays at 6. Here’s why.

One thing I will say upfront: Quicktext has grown to over 3,500 hotels across 76 countries with roughly 125 employees, and they’ve done it on barely $1M in outside funding. That’s a company growing on revenue, not on VC promises. I respect that model. It tells you the product generates enough value that hotels keep paying for it. But it also means development resources are finite, and you can feel the constraints in the areas where Quicktext still falls short.

Velma on WhatsApp

Quicktext now offers their chatbot (still called Velma, still a name I find slightly precious for a piece of business software) through WhatsApp. The idea is sound. Guests can message your hotel on WhatsApp and get AI-powered responses about availability, room types, facilities, check-in times, and local information. If the conversation moves toward a booking, Velma can walk the guest through that process.

Quicktext claims Velma has been trained on 35 million real hotel guest requests across 38 languages. That’s a large training corpus, and in testing, it shows. The WhatsApp responses were accurate and reasonably fast. Velma pulled the right information from our knowledge base, handled French and English without difficulty, and managed basic German queries adequately. A guest asking “Do you have rooms for 14-16 March?” got a sensible answer with pricing and a booking link. That part works.

Where Velma stumbles is on queries it cannot classify. When the AI doesn’t understand what a guest is asking, it falls back to a generic response that reads like a polite dead end. The problem is that many guests simply stop responding after receiving one of these blocks. I watched it happen repeatedly in our conversation logs: guest asks something slightly unusual, Velma returns a vague “I’d be happy to help, please contact the front desk” type response, and the conversation dies. Quicktext claims an 85% automation rate, and that may be accurate across their full hotel base, but the 15% that fails can fail badly enough to lose the guest entirely.

What doesn’t work as well is the feel. The WhatsApp experience feels bolted on rather than native. Response formatting is sometimes awkward, with text blocks that clearly weren’t designed for a chat interface. The rich message elements (carousels, buttons, quick replies) are limited compared to what WhatsApp Business API actually supports. It’s functional, but it reads like a website chatbot forced into a messaging app, not a messaging-first experience built for the channel.

Compare this to tools where WhatsApp was a primary design consideration from the start, and you can sense the difference. It’s not broken. It just isn’t polished.

The Quinta question

Quicktext has introduced a sub-brand called Quinta, positioned as their hotel-specific AI platform. After spending time trying to understand where Quicktext ends and Quinta begins, I’m not sure anyone at the company has a clean answer either.

From what I can tell, Quinta is the label for their newer AI capabilities, particularly the parts built on large language models rather than their older intent-based chatbot architecture. The website uses both names. Sales conversations reference both names. Some documentation says Quicktext, some says Quinta, and at least one page managed to use both in the same paragraph.

For a company trying to establish credibility in the hotel AI space, this brand confusion is an unnecessary obstacle. Pick a name. Commit to it. Explain the transition clearly. What they’ve done instead is create ambiguity that makes it harder for a buyer like me to understand exactly what I’m evaluating.

Direct booking conversion, still

The strongest argument for Quicktext remains the same one it was a year ago: it converts website visitors into direct bookings. We tracked this again over the past quarter. Roughly 9% of chat conversations that began with an availability question ended with a completed booking. The number is consistent with what I measured before.

That’s not a transformational figure, but it’s concrete revenue. For a hotel paying €5 per room per month, the chatbot needs to generate only a handful of direct bookings to pay for itself. On that maths alone, Quicktext works. The chatbot understands hotel-specific queries well. It knows our room categories, our breakfast service, our cancellation policy. Updating the knowledge base remains quick, usually two or three minutes when something changes.

Integrations: still the weak point

I had hoped for progress here. The integration list on their website has grown. More PMS names, more channel managers, more third-party tools. But length of list and depth of connection are different things.

I tested the integration with our group’s PMS again. The data flow is still unidirectional. Quicktext reads reservation data from the PMS (dates, room types, guest names) and uses it in chatbot conversations. Fine. But nothing flows back. Chat interactions don’t appear in guest profiles. Preferences expressed during conversations don’t update guest records. No events trigger PMS workflows.

Their API documentation has improved marginally. Some of the “coming soon” endpoints from last year are now live. But the webhook payloads remain thin. I spent an afternoon trying to build a simple automation: when a guest asks about late checkout via the chatbot, create a task in our operations system. The webhook data didn’t include enough context to make this work reliably. I gave up after three hours and built a workaround using email parsing instead, which is the kind of thing you resort to when an API isn’t ready. At least I have the time to spend three hours on this. An independent hotelier wouldn’t.

For a company founded in 2017, now nine years old, the technical infrastructure should be further along. When I’m evaluating tools for fifteen properties, I need integrations that work at group level, not just for a single hotel. The chatbot AI is mature. The integration layer is not. That disconnect is the core of why this stays at a 6.

European credentials

Quicktext remains headquartered in Paris with EU-hosted servers. Guest data stays within European jurisdiction. Their GDPR documentation is specific and detailed rather than the vague hand-waving you see from some American competitors. The support team speaks native French, which continues to matter when troubleshooting something technical at the end of a long day.

But there’s a catch. Quicktext’s privacy policy mentions that personal data is transmitted to “LLC located in the United States” as a subcontractor, without naming which company. Given their Q-Brain+ AI and the rest of the market, this is almost certainly OpenAI. The privacy policy mentions standard contractual clauses to cover the transfer, but the vagueness itself is a problem. If you’re going to promote European credentials, name your US sub-processors. Don’t hide them behind “LLC.” For a French company selling data sovereignty as a feature, this is an uncomfortable gap between the marketing and the fine print.

These are still advantages relative to the American alternatives. For hotels in France or anywhere in the EU that care about data sovereignty (and you should care), Quicktext’s European roots are a differentiator, just not as clean a one as their marketing suggests.

Pricing, again

Still custom. Still requires a conversation with sales to get a number. Based on more recent research, pricing starts from approximately €250 per month rather than the per-room figures I’d previously heard from colleagues. That’s a meaningful minimum, particularly for smaller independent hotels. There’s no free trial either, which adds friction: you’re committing to a sales process and a contract before you’ve had any hands-on time with the product. I maintain my position from last year: publish the pricing. If your product is good (and parts of it are), you shouldn’t need to hide what it costs behind a contact form.

I should also mention something that came up in conversation at a conference earlier this year. A contact at Edgar Suites, the apart-hotel group in Paris, described Quicktext as “the worst service provider we have worked with in nearly 10 years” and alleged the company “deliberately abused the SEPA mandate” by debiting significant sums when they tried to terminate their contract. I can’t verify the specifics of that dispute, and one bad experience doesn’t define a company. But allegations about billing practices and contract termination are the kind of thing any buyer should ask about directly before signing.