Mews guest messaging is here, and the integration partners are in trouble
At Unfold 2026, Mews launched guest messaging of its own. It is one inbox that pulls WhatsApp, Booking.com, SMS and web chat together, with an AI it calls the Mews Agent that can read a message and act on it without waiting for a person. If you are already a Mews customer it comes with the advanced plan, and the full rollout is promised for August.
I run technology for a group of about fifteen hotels in France. We are on Apaleo, an open and API-first PMS, with Runnr.ai handling guest messaging on top. So I read this launch through two questions, and they are the two worth writing about. The first is what a native inbox does to the dozen messaging tools we have spent years testing. The second is what it says about Mews, a company that built its name on being the most open platform in hospitality and now seems to be doing the opposite.
What a native inbox and AI agent do to the tools
Mews did not just build an inbox. It built two things this category sells, and bundled both into the PMS. The first is the unified inbox, the part that gathers WhatsApp, Booking.com, SMS and web chat into one place and ties each message to a booking. Email and the other OTAs are promised later, not in the launch. The second, and the bigger one, is the Mews Agent: an AI that reads the property’s own knowledge, answers the recurring questions and can take an action like arranging a transfer without a person in the loop. And here is the thing about this category: almost every serious tool in it already sells both of these as one product, the inbox and the AI sitting on top of it together. They are not two separate markets. They are one, and Mews has now built the whole of it into the PMS.
When I test these tools the thing I care about most is integration depth, and the uncomfortable point is that a native inbox and a native agent have the deepest integration there is. There is nothing to integrate, because the guest, the stay and the folio are already in the system. A bolt-on chatbot has to read all of that back through a connector before it can answer well; the Mews Agent starts with it. So on the two jobs most of these tools are built around, organising the messages and answering them, the PMS now starts from in front.
That leaves the most exposed tools in a hard spot, and it is not only the inbox-first ones. Akia and Canary, the two large American names, now sell something a Mews hotel gets at no extra cost, and they carry a US data question on top. Quicktext, whose whole product is the chatbot, is squeezed from the AI side rather than the inbox side, but it is the same squeeze. HiJiffy is the biggest WhatsApp automation name in the category and a capable one, yet automation on top of the PMS is precisely what the Mews Agent now is. chatlyn does the cleverest version of the inbox, pulling Airbnb and Booking.com messages into a single thread. Mews launches without Airbnb or email in its own inbox, so that edge is real for now, though it is the kind of feature a PMS absorbs over time. And I should be honest about my own tool. Runnr.ai is well engineered, and it rode the AI chatbot wave early and well. But what Mews announced is, end to end, the same product Runnr is: a WhatsApp-first inbox with an AI that answers and acts, sitting on the PMS. There is very little in it the Mews Agent does not now do natively, and I say that as the customer paying for it. And honestly, Mews will probably end up with the better product, because it sees everything. A bolt-on like Runnr only ever works with the slice of data and the handful of sellable items you have wired into it. Mews sits on the lot, every rate, every room, every folio, the whole guest history. Where Runnr can upsell a fixed list of extras, Mews can act on anything the hotel sells.
The European tools cannot fall back on data sovereignty here, because the PMS taking their place is European too. The card still on the table is the model itself, and it is the one place Mews looks weaker than it sounds. It has not said which AI writes its guests’ replies, and the signs point to an American one. A tool running a European model can therefore claim something Mews so far cannot, which is exactly why it matters that askng.it, which Sophie runs at her hotel, is built on Mistral rather than OpenAI. And it matters more the harder you think about what flows through that model. Every message the agent handles carries a guest’s name, their contact details, their preferences, often their booking and payment context, and all of it passes through whoever is writing the replies.
So not every tool is in the same position. The ones with room to live are doing something the native inbox and agent do not. askng.it is the one tool here that did not bet everything on AI. It keeps the whole transaction inside the WhatsApp thread, the flows, the upsell, the check-in, the payment, which is a more advanced kind of guest communication than answering well from the front desk, and a harder one to copy. Duve is a guest-experience platform rather than an inbox, and earns its place on the part Mews does not touch: a mature upsell engine that brings in real money, wrapped around check-in and a guidebook guests actually use. The messaging is the least of it. The pattern is simple enough. If the product is the inbox or the bot, or especially both, it is now competing with a free and native feature. If it is something the PMS has not built, or a European model the PMS will not match, it has a few years, and probably more.
What it says about Mews
The second question is the more interesting one, and it is the reason I did not put my group on Mews in the first place.
Mews built its reputation on being open. It was cloud-native early, it had a proper API when that was rare, and it ran the largest integration marketplace in European hospitality. The idea was that you assembled the best stack on top of one clean PMS and no single vendor owned the lot. I chose Apaleo for the same reason, for its open store and a real API, so I have always respected what Mews built even though we are not customers.
The last two years look like a different company. That is roughly what you would expect once this much investor money goes in: the mandate stops being to build a better PMS and starts being to build a bigger one. It bought Atomize and turned it into the native Mews RMS. It bought Flexkeeping, the housekeeping tool. It bought DataChat for the AI layer underneath everything. It put SiteMinder’s channel manager inside Mews under its own brand. And now the inbox. Pricing, distribution, housekeeping, messaging, analytics and payments are being pulled into one system they now call the Mews operating system.
I have felt the edge of this directly. We use Flexkeeping for housekeeping, and when Mews acquired it last September I checked the one thing that mattered to me, which was whether the integration with non-Mews systems would stay as good as it had been. There is no public commitment that it will, and the sync with non-Mews platforms has been less reliable since. That is what tends to happen to a marketplace partner once the platform owns it. The hotels on other systems quietly stop being customers to look after and become a competitor’s users. And there is a quieter worry on top of that one. We stayed off Mews on purpose, yet through Flexkeeping a Mews-owned company now sits inside our operation, with a view of how the hotel actually runs. So when you pick any tool from the Mews marketplace, you are also betting it stays independent. Atomize did not. Flexkeeping did not. The messaging tools are the same bet now.
You do not have to read bad intent into this. Mews raised €255 million in January and has borrowed a further hundred million dollars from a credit fund for the stated purpose of buying companies, fourteen of them so far. That is a company assembling a single operating system and, in all likelihood, preparing itself for the public markets, where a roll-up is the quickest route to the revenue that justifies the valuation. It may well be the better product for a lot of hotels. But an open platform and an all-owning suite are different things, and they are different bets for whoever is running the hotel underneath.
What I would actually do
If you are on Mews, the sensible response is neither to panic nor to ignore it. Turn Mews Guest Messaging on when it ships, run it alongside your current tool through a quiet month, and keep the specialist only if it earns its separate cost on something the native one cannot do. For now that means the parts that live inside the WhatsApp thread, or genuine upsell revenue, rather than the inbox itself. The model question I raised above is a privacy one too, not only a competitive one. Anna, who audited Mews on privacy, would not put guests in front of an AI whose model the vendor will not name, and for replies going out under your hotel’s name I think she is right. We wrote about where this category’s AI actually runs separately; native or not, it is the same question.
If you are choosing a PMS, the trade is clearer than it used to be. An open platform gives you choice and the work of assembling it yourself. A suite gives you one bill and less room to move when a contract comes up for renewal. Neither is the wrong answer. The thing worth being honest with yourself about is which one you are buying, because the more of your stack a single company owns, the less leverage you keep the day you decide one part of it is no longer the best. Payments is where you feel that first. Once Mews is processing your money you are on its rates, and an embedded payment solution is rarely the cheapest way to take a card, or the easiest thing to unpick once it is wired into everything else.
The inbox was always going to be easy for Mews to build. The harder question, and the one I would sit with before signing anything, is how much of your hotel you want one vendor to run, and what it will cost you to leave when you decide a piece of it no longer earns its place.
- Thomas