Brian Chesky’s New AI Lab Bets on Richer Interfaces Over Chatbots

Brian Chesky, the chief executive of Airbnb, is in the early stages of funding a new AI lab focused on user interaction and design, Bloomberg reported Thursday. The venture will develop AI models; Chesky plans to remain at Airbnb and will not serve as the lab’s chief executive, according to several people familiar with the matter who spoke on condition of anonymity because the details have not been made public.

The setup rests on a specific argument Chesky has pressed publicly for years: that travel and e-commerce AI requires richer interfaces than the text-based chatbots OpenAI and Anthropic have popularized, and that no frontier lab has built one suited to the category.

Chesky’s Case Against Chat

The Design School Theory

Chesky co-founded Airbnb nearly 20 years ago after studying design in college, and that background runs through how he describes AI in travel. He has argued publicly in recent years that AI for travel and e-commerce requires a rich user interface rather than the text-based chat windows OpenAI and Anthropic have made central to consumer AI, per Bloomberg’s reporting. The argument is sensory before it is technical. Choosing where to spend a week involves pictures, atmosphere, the feel of a neighborhood, and the quality of light in a particular room. A chatbot that delivers a ranked list of accommodation options doesn’t replicate the experience of scrolling through a space and deciding you want to be there.

Travel is one of the few consumer categories where a purchase decision is also an aesthetic one. Chesky has argued, in interviews and on stage, that this quality means AI in travel has to be designed for the category from the start rather than layered onto a generic chat product. He is among what TechCrunch described, in its confirmation of the plans, as “many Silicon Valley machers who are unsatisfied with the models coming out of the frontier labs.”

His answer is a lab. Bloomberg described the venture as focused on developing AI models and exploring how people interact with AI-powered services, based on people familiar with the plans who said the details are still forming. Airbnb and a representative for Chesky declined to comment on the report.

Passing on OpenAI

Airbnb has not partnered with OpenAI to place a booking experience inside ChatGPT. Chesky said last year the AI startup’s tools weren’t quite ready for what Airbnb wants to deliver, according to TechCrunch. That’s a different call from both Expedia Group and Booking Holdings, which have each integrated their live inventory into ChatGPT’s travel ecosystem. Airbnb’s absence from that ecosystem has been deliberate, and consistent with Chesky’s stated view of what current AI can and can’t do for travel.

Internally, though, Airbnb has leaned hard into AI. Chesky said last month that product pilots which once took years now spin up in weeks, with employees using AI coding tools to accelerate development across the company. Those gains are feeding Airbnb’s push toward a more all-encompassing travel platform, adding car rentals, hotel bookings, and other services. Chesky has said those additions could eventually generate $1 billion or more in additional annual revenue.

A Different Path from Expedia and Booking.com

In October 2025, OpenAI launched in-chat travel apps inside ChatGPT, naming Expedia Group and Booking Holdings’ Booking.com as its first travel partners. Users can now search hotels and flights through natural language, with real-time pricing and availability delivered directly inside the conversation. The feature uses the Model Context Protocol (MCP, an open-source standard that lets AI models connect to external data sources and services), and it opens ChatGPT’s 800-million-user base to travel brands as a direct distribution channel. Tripadvisor was named as a next partner.

Airbnb passed. Booking Holdings described leading generative AI partnerships as a key milestone in its 2026 proxy statement to shareholders. Booking.com said the ChatGPT app gives “travelers a new avenue to explore hotels, homes and unique places to stay on our platform,” building on its ongoing collaboration with OpenAI. Expedia’s rationale has been framed around meeting travelers where they are, positioning ChatGPT as one more distribution surface rather than a structural threat to the company’s own platform. Chesky’s public comments frame exactly that approach as the experience he considers insufficient for travel’s demands.

The Interface-First School of Thought

• $700 million: Hark’s Series A funding round, closed May 2026
• $6 billion: Post-money valuation at time of raise
• $100 million: Brett Adcock’s personal seed investment, late 2025
• 70: Hark’s employee count at the time of the Series A

Chesky is not the first founder in the current AI cycle to treat the interaction layer as a product category worth building from scratch. Brett Adcock, who also founded robotics company Figure.AI and electric aircraft maker Archer Aviation, launched Hark in late 2025 with $100 million of his own capital. The lab describes itself as building AI models and purpose-built hardware in parallel, on the premise that every device people currently use to reach AI was designed before AI existed. TechCrunch noted the structural resemblance to what Chesky appears to be pursuing when it confirmed the Airbnb CEO’s plans this week.

Hark closed a $700 million Series A at a $6 billion valuation in May, with Nvidia, AMD, Intel Capital, and Qualcomm Ventures among the backers. Its design director is Abidur Chowdhury, who led the design of the iPhone Air at Apple before joining Hark to build AI interfaces that feel native rather than retrofitted onto existing hardware. “Today’s AI models aren’t nearly intelligent enough, they feel quite dumb, and the devices we use to access them are fundamentally pre-AI,” Adcock wrote in a January internal memo shared with TechCrunch. Hark’s first multimodal models are expected this summer.

The size of the Series A, raised before Hark shipped a public product, signals how much institutional capital is treating the interface layer as a structural bet worth paying for early. Nvidia, AMD, and Intel put chip-company money into a lab whose founding argument is that the screens between people and AI need to be redesigned from scratch. Chesky’s lab, in early funding with no named chief executive, is entering a field where the comparable already has a nine-figure raise and a hardware-plus-model strategy being built simultaneously.

Morgan Stanley and the Agentic Counter

What Morgan Stanley Is Building

The interface-design school has a structural challenge arriving from a different part of the market. Morgan Stanley will soon open its stock administration platforms to external AI agents, CNBC reported last week in an exclusive. The bank’s ShareWorks and Equity Edge platforms serve stock compensation plans for nearly half of S&P 500 companies and eight of the ten largest unicorn startups. Autonomous AI agents will be able to pull data and administer those plans directly, bypassing the human-facing login screens both platforms currently require. In April, Morgan Stanley executives attributed $1.2 trillion in assets gathered to its workplace strategy. Mark Mitchell, chief product officer of Morgan Stanley at Work, told CNBC what the new direction implies:

In a future state, our corporate clients will not be logging into ShareWorks or Equity Edge.

Mitchell said those clients will instead use agentic AI tools running within their own organizations, interacting with Morgan Stanley’s systems “in a purely agentic way.” The bank has already granted early access to a small group of clients and plans to extend it to all 3,400 of its administration clients by next year. Morgan Stanley, which began its OpenAI partnership in 2022, is using the same MCP standard that powers Expedia and Booking.com’s ChatGPT apps as the integration layer. JPMorgan Chase and Goldman Sachs are deploying AI agents internally for tasks including writing code, per CNBC, but neither has yet publicly announced steps to open its systems to external agents.

Agents That Skip the Screen

The Morgan Stanley move is a financial-services story with an architecture that extends well beyond finance. A platform designed for AI agents to query directly is a platform that has moved past the interface question. The agent accesses the data; the quality of the human-facing design becomes irrelevant to the transaction.

In travel, Booking.com has already said its autonomous agents handle thousands of customer inquiries daily. If accommodation platforms build MCP hooks for AI agents the way Morgan Stanley is building them for corporate stock plans, a traveler’s AI assistant may eventually book a trip by querying a platform’s data pipeline directly. No booking interface loads on any device. The decision about where to stay gets made upstream, inside the agent’s optimization logic, before any human opens an app.

The business logic compounds from there. If a traveler’s agent selects accommodation based on price, availability, and past preferences pulled from the platform’s API, the brand whose name appears on that platform becomes less relevant than whether the platform responds cleanly to an automated query. That is the scenario the interface-first school is betting remains further away, or slower-moving, than the pace Morgan Stanley is setting. Chesky’s lab, still without a named chief executive, is making that bet from the earliest funding stage.

The Lab’s Open Questions

Several facts about the new venture have not been disclosed:

• The chief executive who will lead it
• The total capital Chesky is committing
• Any formal relationship between the lab and Airbnb’s product roadmap

Chesky’s standing in AI circles predates the lab by nearly two decades. He met Sam Altman, OpenAI’s chief executive, in 2006 through Y Combinator, the accelerator that incubated Airbnb. As OpenAI scaled rapidly, the two met regularly, with Chesky offering advice on managing a hypergrowth tech company. He was reportedly considered as a potential OpenAI board member, according to TechCrunch, and he helped broker Altman’s return after OpenAI’s board fired him in late 2023. His connection to OpenAI’s founding network puts him in an unusual position: a travel CEO who knows more about the frontier AI industry’s internal dynamics than most of its direct participants.

TechCrunch, in confirming the plans, noted that Chesky will not lead the lab and will remain at Airbnb’s helm. Whoever takes the chief executive seat will work alongside, as TechCrunch put it, “a founding chair known as a micromanager.”

Paul Graham, Y Combinator’s co-founder, coined the term “founder mode” in 2024 partly in response to Chesky’s hands-on management approach at Airbnb. The AI lab is the first direct financial bet Chesky has placed on AI outside Airbnb itself.

Bloomberg’s sources said the lab remains in early stages of funding, with no chief executive yet named. The details, they said, could change.