The clean version of Andy Tsen's story ends in January 2022. A 15-person studio spends years building the game it wants to play, launches it into a market that had repeatedly been declared too small, and watches Zenith: The Last City reach No. 1 in sales on Steam, Meta Quest and Oculus Rift. Development costs come back quickly. Hundreds of thousands of players arrive. A $35 million financing follows.
But games do not end when the launch trailer does. A massively multiplayer world makes a recurring promise: come back, and there will be something worth doing. Zenith's launch proved that people wanted the promise. The months afterward exposed how difficult it was for a small team to renew it.
That gap - between making a world and continuously feeding it - is the useful center of Tsen's career. It connects the engineer who left Cisco after roughly six months, the product manager who moved through social and mobile games, the VR founder who talked his way back into Y Combinator, and the CEO now building Aura, an AI assistant that lives inside Unreal Engine and Unity.
A recruiter named Hillary
Tsen did not begin with a tidy identity as a game developer. He studied computer science and economics at the University of California, Davis, then started as a software engineer at Cisco. The company was fine; the fit was not. Around six months in, a recruiter named Hillary called about Metamoki, the studio behind the Facebook game Mob Wars.
The timing mattered. A social game could look a lot like a website, so the skills Tsen already had were suddenly game-development skills. The job opened a path through engineering and product roles at companies including GREE, Mobee and Jana. He watched social games grow, then mobile games. The lesson he carried was less about genre than about timing: when a new platform changes who can build and who can play, old skills acquire new value.
In Boston, he helped organize a large VR community from 2014 to 2016. The meetups gave him access to early Vive and HoloLens hardware and to people trying to understand what immersive computing might become. In 2016, an Oculus DK2 settled the matter. He quit his job, moved home and committed the next stretch of his career to VR.
“I got to see the future before the future had arrived.”Andy Tsen, on his years in Boston's VR community
Through the Oculus Launch Pad orbit, Tsen met engineer Lauren Frazier. Their partnership was tested in work before it became a company. They built Conjure Strike, a multiplayer VR title, and collaborated for a couple of years. By the time they founded Ramen VR in 2019, the ambition was specific: a large online world influenced by JRPGs, MMOs and anime, built for headsets that were only beginning to become accessible.
The rejection that did not stick
Starting a VR studio in 2019 looked unfashionable. Rift and Vive had not created a mass market, and companies were moving away from the category. Ramen applied to accelerators anyway. Y Combinator took the founders to a final interview and rejected them.
Tsen wrote back. His message, as he later summarized it, said the decision hurt, he disagreed, and here was why. The partners invited the pair in again. Around the same time, Techstars accepted them too. A week that began with rejection ended with two offers. Ramen chose YC's Summer 2019 batch.
The better detail is what happened before the fundraising machinery. The founders put a website online with concept art and a Stripe checkout. There was no finished game, only an idea and a promise. People paid roughly $2,000 in a day. That small test became evidence for a Kickstarter, which raised $280,075. Ramen posted on Reddit, built a Discord, ran playtests while the product was still rough, and let an audience form around the work.
“Hey, what's the worst that can happen, right?”Andy Tsen, on asking Y Combinator to reconsider
This is the repeatable part of the story. The founders did not ask the market whether VR MMOs were sensible in the abstract. They asked whether someone would pay for this one. Then they asked whether backers would fund it. Then whether alpha players would return. Each step converted a grand thesis into a smaller behavior that could be observed.
Winning launch week, meeting month two
Zenith launched on January 27, 2022. It reached No. 1 across several storefronts, including the full Steam chart, not only its VR section. Players spent close to two hours a day in the game during the launch period. The studio made back its development costs and announced a Series B led by Anthos Capital and Dune Ventures.
Tsen's favorite outcomes were social rather than financial. He heard from guildmates who crossed the country to meet, from players who felt more comfortable connecting with other people, and even from couples who met in Zenith and married. The world had done what its founders hoped a social VR world could do.
The operating model was less forgiving. Years of production had filled the launch with content, but the team did not enter launch with another six to nine months ready to go. Live service required a different cadence, roadmap and measurement system. The release was not a finish line. It was the moment the content treadmill switched on.
By 2024, the company said the vast majority of Zenith's hundreds of thousands of players had stopped after about a month. A free-to-play mode did not repair the economics. New content development ended, while the servers remained available. The decision was painful and public. It was also precise: the studio had not solved retention, and continuing to operate the game at a loss was no longer feasible.
The tool hidden inside the game
Aura grew from the work Ramen wished had been easier. Game production contains many jobs that matter without feeling creative: update the same field across 80 items, connect Blueprint nodes, search a project for the relevant class, run a verification pass, find why a system broke. General coding assistants can help outside the engine. Ramen's bet is that a game tool needs to understand the project and work where designers, artists and engineers already spend their day.
Tsen calls the product a mech suit. The person inside still decides where to go. The machine supplies leverage. Aura uses specialist agents for different forms of engine work because a studio itself is specialized. A designer, engineer and artist do not approach the same problem with interchangeable expertise; the software should not pretend they do.
Ramen introduced Aura for Unreal Engine, then acquired Coplay in March 2026 to bring Unity support into the same product. By June, Aura 15 included a sandbox, reusable skills, verification, level-design and batch-editing capabilities, plus subscription plans with unlimited-use modes. The company had moved from asking players to inhabit one elaborate world to helping teams assemble many of them.
There is a tension inside any AI product for creative work. Faster production can expand what a small team attempts; it can also disrupt jobs and raise questions about copyright and authorship. Tsen has spoken about those risks directly. His stated goal is to move game development toward AI assistance ethically and consciously, with tools that give people more time for mechanics, judgment and ideas.
“You can compete on ideas.”Andy Tsen, on what AI could change for smaller game teams
The next promise
Tsen once described work as a form of meditation - a place to improve himself, care for the people around him and test what he truly values. It is a revealing frame for a founder whose public record contains both a chart-topping launch and an admission that the product could not hold enough players.
The common founder story treats persistence as refusing to change. Tsen's version is more mobile. He stayed with games while moving from social to mobile to VR to AI. He stayed with the goal of expanding creative possibility while changing the product that might deliver it. He challenged one rejection, accepted another kind of verdict from retention data, and carried the unfinished work into a new tool.
Aura will face its own version of month two. A fast demo can show a Blueprint appearing from a sentence. A durable product has to remain accurate inside messy commercial projects, earn trust across different crafts, and fit into the schedules of teams under pressure. Ramen knows what it feels like when opening-week demand and long-term use tell different stories. This time, that knowledge is part of the product.