Seven Years In: Building Language AI Before the Boom, and What Comes Next

A look back at XL8’s first seven years, and the bets shaping our next chapter.

We founded XL8 in Silicon Valley in October 2019. ChatGPT was still three years away, and “AI translation” mostly meant stiff, word-for-word output. We bet on something bolder: that machines could help people understand each other in real time, with the nuance a story or a debate deserves.

Seven years later, our technology has subtitled films and series, put live AI captions on U.S. public broadcasts, and helped audiences follow prime ministers, scientists, and founders on stage. Here is how we got here, what we learned along the way, and where we are heading next.

Seven years at a glance

2019 — Where It All Began

XL8 was founded in Silicon Valley in October, marking the beginning of our journey to transform global communication through AI-powered language technology. We began commercializing our proprietary machine translation engine.

2020 — Expanding Our Technology

We launched our proprietary speech-to-text technology, expanding our capabilities beyond machine translation.

2021 — Going Global

We introduced live subtitling, joined the NVIDIA Inception Program, and established our first international engineering office in Seoul, South Korea.

2022 — Introducing MediaCAT

We launched MediaCAT, our AI-powered media localization platform, bringing advanced language technology to the media and entertainment industry.

2023 — Recognition and Momentum

We raised $7.5 million in Series A funding, bringing our total funding to $11.5 million. We also received two DEG EnTech Awards and published research at EMNLP 2023.

2024 — Innovation at Scale

MediaCAT was named Best Product in AI & Data at the Product Awards, while XL8 received the Next Innovator Award at NextRise. Meanwhile, EventCAT powered multilingual communication at the SK AI Summit, an event attended by more than 30,000 participants.

2025 — Breaking New Ground

We delivered live AI-powered Spanish captions to U.S. public broadcasters, expanding the reach of our real-time language technology. EventCAT also received the prestigious Hermes Award for Technical Achievement in Localization.

2026 — Connecting the World, Advancing AI

EventCAT supported major international events, including SusHi Tech Tokyo and the World Knowledge Forum. At IBC in Amsterdam, we introduced MediaCAT On-Prem and Ops Agent, marking another milestone in our AI-powered media localization solutions.

Some rooms we will not forget. At the SK AI Summit, Chairman Chey Tae-won and OpenAI co-founder Greg Brockman spoke to a global audience. At SusHi Tech Tokyo, about 40,000 attendees heard from Prime Minister Sanae Takaichi and Governor Yuriko Koike. At the World Knowledge Forum, David Cameron, Sebastian Thrun, and Richard Dawkins took the stage. In between came a policy seminar at Korea’s National Assembly, an AI Builder Salon in Seoul with Anthropic, the World Coffee Leaders Forum, and the 30th Asian Television Awards.

What seven years taught us

1. Start with the hard problem. Translating finished text was never the end goal. We went after live speech early, with speech-to-text in 2020 and live subtitles in 2021, years before real-time AI became a category. That head start is why our live products hold up in rooms where there is no second take.

2. Media made us precise. Subtitling films and series taught us discipline about timing, names, and tone. The same standards now run through every live session, from a director’s Q&A to a policy forum at the National Assembly.

3. Trust is earned in real rooms. A demo is easy. A summit with 30,000 people, a live broadcast, or a packed plenary hall is not. Every event taught us something no benchmark could.

4. Language access is human. Our technology has shown up at classrooms, research exchanges, sporting events, and even weddings. Whatever the setting, the goal is the same: no one should miss the moment because of the language it happens in.

What’s next

Our next chapter starts from the same belief we founded XL8 on: language should never decide who gets to take part in a conversation.

Make live multilingual the default. On a broadcast, on a conference stage, or in a meeting room, real-time translation should be as expected as a microphone. That is the standard we are building toward.

Thank you

None of this happens without people. To our team, who turned an ambitious idea into products that work under pressure. To our customers and partners, who trusted us with their biggest stages and most important stories. And to everyone who has used XL8 to follow a conversation they would otherwise have missed: thank you.

Seven years in, we are still just getting started.

Breaking language barriers since 2019.