Three people, one table, three first languages: German, French and Italian, the way meetings often go when a project spans Zürich, Geneva and Lugano. Someone always asks the same question about the glasses: "Can they just translate for everyone, all at once?"
Short answer: not the way people picture it. Even G2 live translation works as a language pair — you set one source language and one target language, and the text of what's said appears on your lenses in your chosen target. That's true of essentially every consumer live-translation product on the market today, glasses or earbuds. A genuine three-way (or four-way) roundtable, where each participant simultaneously sees the conversation rendered into their own language, is not something the G2 — or any comparable device we know of — actually does. Plan the meeting around that limit rather than around a promise nobody is making.
Why translation is pairwise, not multilateral
Under the hood, live translation on the G2 takes speech, sends it to a cloud translation service, and returns text in one target language onto your display. It's built for the situation of one person listening in a language they don't speak well, with a counterpart speaking a different one. Add a third language to the room and the app still has to pick a single target for your lenses — it can't split the output three ways for three different sets of eyes. Each pair of glasses in the room is having its own private translation into its own owner's target language; the glasses aren't coordinating with each other or with a shared "meeting channel."
That's not a flaw specific to Even Realities. It reflects how live-translation pipelines are generally built: detect source, translate to one target, display or speak it. A true simultaneous multi-language interpreter (the kind a conference books) resolves N languages into N outputs at once, with a human — or, on the far frontier, a much heavier multi-channel system — doing the work per listener. Consumer AI glasses aren't there yet, and we'd rather say so than let a demo video imply otherwise.
What actually happens in a three-language meeting
| Situation | What the glasses do | Works? |
|---|---|---|
| One speaker (German) talking to you, target set to French | Live text in French on your lenses | Yes |
| Three people speaking three languages, all in one live session | You still see only one chosen target; the app doesn't auto-detect and switch per speaker mid-conversation | Not reliably |
| Each participant wears their own glasses, each sets their own target | Each person gets their own private translation stream — no shared "everyone sees everyone" view | Partially, per person |
| Fast back-and-forth switching between languages mid-sentence (code-switching) | Harder for any speech-translation pipeline to track; expect lag and occasional mistranslation | Weak spot |
Based on how Even Realities documents the Translate feature (source language → target language) and how cloud speech-translation pipelines generally work; not an Even-specific multi-speaker benchmark.
Two workflows that actually work
Rotate the target language
If the meeting genuinely needs three languages, treat translation the way a human interpreter would: one language pair active at a time. Agree on a speaking order, or pause between segments, and reset your target language before each block. It's not seamless, but it's honest — and for a structured meeting (status updates, round-the-table input) it works well.
Use a shared bridge language where you can
In a lot of Swiss business settings, everyone at the table already has a working level of one common language — often English or the locally dominant one. Let the glasses handle the one genuine gap (the participant who doesn't share it) rather than trying to translate every pairing. That turns a three-language problem into a single-pair problem, which is exactly what the feature is built for.
What we wouldn't rely on
Don't plan an important negotiation or a legal or financial discussion around simultaneous three-way translation working smoothly in real time — it isn't the feature's design point, and a misheard word in a fast, crosstalk-heavy room compounds the risk. For that kind of meeting, a professional interpreter is still the right tool, glasses or not.
Buy, skip, or test first?
Buy with confidence if your real need is one-to-one or small-group translation between two languages at a time — that's what the G2 is built for, and it does it well.
Skip the assumption that a single pair of glasses will silently interpret a genuinely trilingual roundtable for everyone at once. No current consumer device does that.
Test first with your actual language combination and a realistic amount of background noise in the Zürich showroom — bring a colleague who speaks the other language and try the rotate-the-target workflow before you rely on it in a real meeting. If translation is the main reason you're considering the G2, our translation glasses overview covers the single-pair feature in more depth, and the AI glasses database lets you compare it against other models.
Disclosure: AI-Eyewear is the authorised Swiss reseller of Even Realities. We'd rather set realistic expectations for a multilingual meeting than let a feature list imply more than the product delivers.