— Mirko Lalli
The MIT data point that landed this week deserves attention: roughly 95% of task-specific enterprise generative AI initiatives have yet to produce measurable business returns. The problem, researchers found, is not the models themselves but tools that fail to learn from feedback, fit existing workflows, or solve clearly defined operational problems. This lands at the same moment IBS Group spins out Naviq Technology as a dedicated AI travel venture, ITILITE announces voice AI for corporate booking, and Hospitality Net publishes yet another framework for hyper-personalization. The gap between what vendors are selling and what operators can actually use remains wide.
My read: we are entering a correction phase. The World Economic Forum's Global Risks Report 2026 flags technology fragmentation as a near-term concern, and I see that playing out in travel tech stacks right now. Hotels chasing AI-driven personalization in guestrooms, airlines retooling loyalty programs around elite status monetization, cruise lines like Four Seasons ordering a third yacht before their first completes a full season. Capital is moving, but integration lags behind ambition. Spain adding 3,000 ultra-luxury rooms from international brands signals confidence in demand, yet the operational backbone to deliver personalized service at scale is still patchy.
For the next twelve months, I would focus less on adopting new AI tools and more on auditing which existing ones actually connect to revenue or cost outcomes. The 95% failure rate is not a technology problem. It is a workflow problem, and that is something DMOs and hoteliers can fix without waiting for the next product launch.
According to a 2025 MIT report, roughly 95% of task-specific enterprise generative AI initiatives have yet to produce measurable business returns. Not because the models are broken. Because the tools fail to learn from feedback, fit existing workflows, or solve clearly defined operational problems.
This week's news is a masterclass in what happens when an industry oscillates between these two poles: AI theater versus AI that actually works.
Where the Money Actually Lands
The hospitality sector keeps announcing AI integrations. Shiji and FPG claim their Infrasys POS and CheckMax integration will boost incremental revenue per guest by 5-15% for hotel dining outlets. Lodging Interactive launches Commingle360, combining QR code feedback with human-crafted review responses across 140 platforms. ITILITE integrates Voice AI for corporate travel bookings between the US and India.
Notice the pattern. The initiatives gaining traction share three traits: they target specific revenue moments, they augment human decisions rather than replace them, and they connect to existing transactional workflows. The 95% that fail? They start with "let's add AI" instead of "let's solve this operational problem."
The Unmanaged Half
Here's the camera flip. While everyone obsesses over managed corporate travel programs, independent hotels are quietly capturing the other half of corporate spend — the unmanaged trips that skip the RFP entirely. According to Hospitality Net, these properties are loading competitive rates directly into GDS corporate booking tools, bypassing the procurement process altogether. No tender. No negotiation theater. Just presence where the decision happens.
This matters because it reveals how discovery is fragmenting. The formal channel controls less than you think. The informal channel — where travelers book without asking permission — is where technology actually determines who wins.
Wellness Moves to the Room
The biggest shift in hotel wellness may not be the spa renovation. According to eHotelier, smart room technology that monitors health metrics is embedding longevity into the guest experience. The room becomes the wellness product. This is hyper-personalization done right: not a segment strategy, but a sensor-driven response to the individual body in the bed.
Meanwhile, Spain adds 3,000 ultra-luxury hotel rooms from international brands. Four Seasons orders a third yacht. Radisson accelerates global expansion. The luxury segment is betting that high-touch service, not AI efficiency, justifies premium pricing.
My contrarian take: both bets are correct, but most operators will pick the wrong one for their asset. The middle will hollow out. Budget properties need ruthless automation. Luxury properties need AI invisible to the guest but omnipresent in operations. The undifferentiated three-star hotel running neither playbook is the real risk.
What This Means for You
Stop asking "how do we use AI?" Start asking "which operational problem costs us money every week?" The 5% that work start there. The 95% that fail start with a vendor demo.
That's your briefing for the week. More on the ground next time.