There's a certain irony in hospitality technology: an industry built on service has some of the worst software in business. Front-desk systems that predate the iPhone. Reservation platforms held together with digital duct tape. And yet hotels—drowning in guest calls, group sales inquiries, and operational fires—desperately need automation.
Most AI vendors looking to crack this market hit the same wall. Legacy property management systems and central reservation platforms weren't designed to talk to modern software. API integrations, when they exist at all, require months of work and the kind of enterprise sales cycles that can kill a startup's momentum.
Lance, a Y Combinator W26 graduate with a team you could fit in a hotel elevator, thinks it has a workaround. Instead of building integrations, its AI agents simply watch the screen—like a human employee would—and operate the old software directly. No APIs. No middleware. Just computer vision and what the company calls "autonomous agents" that can see what's happening in legacy desktop applications and take action.
It's technically audacious. Whether it actually works at scale remains an open question.
The Vision Play
The company's approach centers on what it describes as "vision-based computer-use agents." These AI systems don't need permission from Oracle or Amadeus or whatever ancient vendor powers a hotel's core systems. They run on-premises, on virtual machines, or on Lance's own tablets, operating the same clunky interfaces that front-desk staff navigate every day.
According to Lance's public materials from earlier this year, the company has deployed across more than 50 properties—hotels operated by groups working with brands including Marriott, Hilton, and Hyatt. The startup hasn't named specific customers or published case studies, and the claim itself dates to the company's initial launch announcement. Still, for a team of three, that's either impressive traction or aggressive marketing. Perhaps both.
The product itself comes in three parts. A "Receptionist AI Agent" handles inbound calls, texts, emails, and messages from hotel platforms—fielding questions, logging requests, escalating the tricky stuff. All in multiple languages, naturally. A "Sales AI Agent" tackles group inquiries and RFPs, qualifying leads and drafting proposals. And Lance Pad, a tablet-based operations hub, aims to kill off the walkie-talkies that hotel staff still carry around like it's 1997.
That last bit feels telling. Hospitality operations haven't changed as much as the guest experience pretends they have.
Who's Behind It
CEO Caleb Chan comes from enterprise product roles—Google, Salesforce, and as Former Head of Product and Founding Engineer at Silicon Valley Commerce, which had what the company describes as a nine-figure exit. COO Gavin Brennen has deeper hospitality roots. He grew up around Marriott operations (family connections to the industry) and worked on AI agents at Domo before Lance. CTO Gatik Trivedi brings Stanford speech and language research credentials, a Berkeley M.E.T. degree, and experience building agent infrastructure at GI Partners.
It's the kind of founding team that looks good on paper: technical depth, enterprise software experience, industry familiarity. Whether that's enough to navigate hotel tech's notorious complexity is another matter.
Y Combinator announced Lance publicly in early February, highlighting the 50-plus hotel footprint in social media posts and the company's Launch YC profile. The startup has since partnered with Quore, an operations management platform—announced via LinkedIn, though Quore itself hasn't issued any formal statement. Lance has also been hiring founding engineers and, in a move that raised eyebrows for a three-person team, put up a billboard in Times Square.
Bold. Maybe too bold, depending on how the product performs in the wild.
A Market That Moves Fast Now
Lance isn't operating in a vacuum. The hotel AI space has gotten crowded, fast.
Canary Technologies—already established in guest engagement—launched its own "Hospitality AI Agent Studio" in early March, a low-code platform that lets hoteliers build their own agents. PolyAI powers voice assistants like Travel Outlook's "Annette" agent, which got an upgrade in February. Asksuite, HiJiffy, and Quicktext dominate the guest messaging and reservation AI segments. Sabre, with its massive installed base, brought SynXis Concierge.AI to market backed by enterprise CRS vendor scale.
The incumbents have real advantages: existing relationships, brand trust, years of integration work already done. But they also move like large companies tend to—cautiously, incrementally. If Lance's computer-vision approach works reliably, it could be the wedge a startup needs to slip past the establishment.
The question is reliability. Hotels can't afford agents that hallucinate room rates or mishandle group bookings. And the security questions around AI systems watching and manipulating production software—inside systems that hold guest payment data and personal information—aren't trivial. Lance will need to answer those concerns convincingly if it wants to move upmarket.
The Bigger Picture
Hospitality is having its agentic AI moment, to use the industry's preferred buzzword. Wyndham has reportedly deployed 250 agents in production, handling hundreds of thousands of guest calls. (That figure comes from an October 2025 announcement, though whether it's grown since then is unclear.) Marriott has talked publicly about an "agentic mesh" strategy and committed $1.1 billion to technology spending in 2026, with more than one-third allocated to digital and tech enhancements. BCG published research in March declaring the rise of "AI-first hotel companies." IDC predicted that agentic AI would "redefine travel and hospitality in 2026."
There's a certain amount of hype-cycle noise in all that. But the underlying driver is real: labor costs keep climbing, guest expectations keep rising, and the old model of humans answering every phone call doesn't pencil anymore.
For Lance, the path forward depends on turning early deployments—whatever their actual scale—into the kind of repeatable, referenceable success that vertical SaaS companies need to survive. The startup hasn't disclosed funding beyond its YC batch, which raises the obvious question: how long can it operate in land-grab mode before needing a serious war chest to compete with better-capitalized rivals?
The company also faces a classic startup dilemma. Move too slowly, and incumbents will catch up or copy the approach. Move too fast, and you risk breaking things in an industry where broken software means angry guests and lost revenue.
Through the Screen, Not the API
In the end, Lance is making a bet that the fastest way into hotel tech isn't through the front door—negotiating with vendors, building integrations, navigating procurement committees. It's through the screen itself, using AI that doesn't care how old the underlying software is because it just needs to see what's there.
It's an elegant solution to an ugly problem. Whether it's a sustainable business model is a different question entirely. Hotels need automation, yes. But they also need software they can trust not to screw up at 2 a.m. when a guest arrives to find no record of their reservation.
That's the test Lance hasn't taken yet—the one that comes after the billboard and the LinkedIn announcements. The test where the AI has to work, reliably, at scale, with real consequences when it doesn't.
For now, the company is betting that being small and audacious beats being big and slow. In tech, that bet sometimes pays off spectacularly.
Sometimes it just means you run out of money before you figure out the hard parts.
