Former Google Maps and Uber Exec Raises $12M for Atlas, an AI Travel Agent
Manik Gupta, a former Google Maps director of product and a senior Uber executive, has raised $12M in a seed round for Atlas, an AI agent for travel. Atlas runs a trip over weeks: it remembers preferences, monitors prices, and books flights and hotels on its own. The bet is that travel is a multi-week project, and a narrow agent can do it better than a general-purpose chatbot. For founders, it is a test of whether a vertical agent can survive alongside the tech giants.
AI-processed from TechCrunch; edited by Hamidun News
Voyage Labs, a startup founded by Manik Gupta, former director of product management at Google Maps and former director of product at Uber, has raised $12M in a seed round, the first round of investment. The money is meant for Atlas, an AI travel agent. The idea is simple and familiar to anyone who has planned a vacation: on Monday you say you want a warm place in February, on Thursday you refine the budget, and two weeks later the assistant tells you a suitable flight has dropped in price and can book it. For developers and founders, this is a reason to look at how a narrow agent can differ from a general-purpose chatbot.
Kleiner Perkins led the round, with SignalFire and SV Angel also participating. The company announced it on Tuesday. The funds will go toward expanding partnerships with suppliers (that is, those who provide access to tickets and hotels) and growing the engineering team.
What Atlas does
Trip planning has long been a favorite demo for AI assistants: you can try it in Muse, Gemini, and about half a dozen other services. But usually it is just one feature of a general-purpose agent. Voyage Labs is betting that a specialized travel assistant will beat its competitors.
Atlas is currently available in a closed beta. According to Gupta, it can be used to plan, book, and run an entire trip, from the first spark of inspiration to getting home. The assistant works as a persistent helper: it takes in ideas that come to the user during planning and looks into each one. Because the horizon is long, it relies on conversation history that spans weeks and tailors recommendations to the person's tastes. Atlas also tracks price changes in real time and ultimately books flights and hotels.
Chief technology officer Yogesh Jain explains the choice this way: trip planning is a project with its own lifecycle, stretching over days and weeks, so long-lived agent technology suits it very well.
Why bet on a narrow vertical
The market for AI travel assistants is already crowded, but the team has experience in exactly this area. Gupta was director of product management at Google Maps, then spent four years as chief product officer (CPO) at Uber. He says that for eight years his only job was building the map, and that topic truly captured him.
The main question is how Atlas will compete with the general-purpose assistants of the big labs. Gupta acknowledges that the market is changing quickly and no one can say with confidence which product formats will survive in the coming years. His position is that the "travel and active leisure" segment is large enough to need a focused solution. The team plans to build an end-to-end experience where everything the user needs is in one place.
What this means for those building agents
A general-purpose assistant answers questions well. A long-lifecycle project requires memory, background price monitoring, and access to real booking systems. That is why Voyage Labs' plans include supplier partnerships: without them, the agent remains an advisor and cannot buy a ticket.
There are also honest limitations. The TechCrunch report contains no data on the number of users, the accuracy of recommendations, or the economics of bookings: the product is still in closed beta. It is also unclear how the company will make money. The big labs can build long-lived agents too, so Voyage's advantage has to come from depth of integrations and domain knowledge, not from the idea itself.
The practical takeaway is this. If your domain is structured as a multi-week project with many steps, prices, and deadlines, a vertical agent with memory and direct supplier access may prove stronger than a general chatbot. You can only test that in production, and Atlas will be one of those experiments.
- Nov 21, 2026A swimming team captain handed ChatGPT three scheduled tasks and cut her routine work by a multiple
- Nov 21, 2026MIRA Neural Network Renders Rocket League at 20 FPS, but Ping Hits 460 ms
- Nov 21, 2026Google built a closed copy of the internet with fake banks to train Chrome's AI agents
- Nov 21, 2026MLPerf Training Adds Nvidia Rubin and TPU Ironwood in First Appearance: 153 Systems
Key Questions Answered
What Atlas does?
Trip planning has long been a favorite demo for AI assistants: you can try it in Muse, Gemini, and about half a dozen other services. But usually it is just one feature of a general-purpose agent. Voyage Labs is betting that a specialized travel assistant will beat its competitors. Atlas is currently available in a closed beta. According to Gupta, it can be used to plan, book, and run an entire trip, from the first spark of inspiration to getting home.
Why bet on a narrow vertical?
The market for AI travel assistants is already crowded, but the team has experience in exactly this area. Gupta was director of product management at Google Maps, then spent four years as chief product officer (CPO) at Uber. He says that for eight years his only job was building the map, and that topic truly captured him. The main question is how Atlas will compete with the general-purpose assistants of the big labs.
What this means for those building agents?
A general-purpose assistant answers questions well. A long-lifecycle project requires memory, background price monitoring, and access to real booking systems. That is why Voyage Labs' plans include supplier partnerships: without them, the agent remains an advisor and cannot buy a ticket. There are also honest limitations. The TechCrunch report contains no data on the number of users, the accuracy of recommendations, or the economics of bookings: the product is still in closed beta.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.