An AI translator sat in on our business meeting. Here's what happened.
We ran international business meetings through AI voice translation for three months. Some things worked better than expected. Others need honest discussion.
Three months ago we started running every international meeting through AI voice translation. Not as a test. As the default. If the meeting involved two languages, translation was on. No interpreter, no captions, no "let's just speak English and hope for the best."
Some things went better than I expected. A few things went wrong. Here's the honest version.
The English problem
Before translation, our default was English. Everyone on the team spoke it at varying levels, and the unspoken rule was that meetings happened in English because it was the only common language.
The problem is that "speaks English" covers a wide range. The person who lived in London for three years and the person who learned English from YouTube videos are having fundamentally different experiences in the same meeting. One is thinking in English. The other is translating in their head, which means they're always half a step behind.
This shows up in subtle ways. The non-native speakers talk less. They make shorter, safer points. They don't push back as often because formulating a counterargument in a second language, on the spot, under time pressure, is exhausting. The meeting runs in English but the best ideas stay in other languages.
When we switched to voice translation, the change was immediate. People who had been quiet for months started contributing. Not because they didn't have opinions before. Because they could finally express them without the cognitive overhead of performing in a second language.
What went right
Meetings got shorter. No more "can you repeat that?" or "what I mean is..." loops. People said what they meant in their language, and the translation carried it. Average meeting time dropped by about 20%.
Client calls improved. Our clients in Germany and Japan had always been polite about doing calls in English. When we offered to run the calls with voice translation so they could speak German or Japanese, the dynamic changed. They were more direct, asked better questions, gave more detailed feedback. One client told us it was the first time a vendor had ever adapted to their language instead of the other way around.
Onboarding got easier. New hires in non-English-speaking offices could participate in company-wide meetings from day one instead of waiting until their English was strong enough. That alone compressed the onboarding timeline.
What went wrong
Jargon still trips things up sometimes. We use internal shorthand — project names, made-up acronyms, product terms that don't exist in any dictionary. The AI handles most of it from context, but occasionally it guesses wrong. We learned to spell out abbreviations in important meetings.
Some people felt self-conscious at first. Hearing your voice come out in another language is a strange experience. A couple of team members said it felt weird for the first few calls. That went away after a week.
Background noise in open offices caused issues. This isn't specific to translation — any voice system struggles with noise. We bought a few cheap headsets and the problem went away.
The security question
This came up immediately when we proposed the switch. Our legal team wanted to know where the audio goes, who can access it, and what happens to it after the call.
We went with Hitoo specifically because it uses end-to-end encryption and processes everything without sending audio to external services. For our client calls — which regularly involve pricing, contract terms, and product roadmap discussions — this was non-negotiable.
If you're evaluating AI translation for business meetings, ask your vendor exactly how they process audio. "We use AI" is not an answer. "We process audio on our own infrastructure with end-to-end encryption and no third-party API calls" is an answer.
Three months later
We're not going back. The team communicates better, meetings are shorter, and we stopped losing information to the English tax. Voice translation isn't perfect — nothing that processes natural language at speed ever will be — but it's good enough that the alternative feels like a step backward.
The biggest surprise wasn't the technology. It was discovering how much we'd been leaving on the table by forcing everyone into one language. The ideas were always there. They just needed a way out.