Speak any language.
In seconds.
No account.

Fast enough to feel like a real conversation.

Translation plays back in under 2 seconds — and common phrases are pre-loaded so the very first one feels instant. When you're offline, 8 languages keep working on-device with no signal at all (English, Spanish, French, German, Portuguese, Italian, Mandarin Chinese, Japanese). Hospital basement, construction site, middle of nowhere: it works.

Pull out a speaker. It just works.

Connect any Bluetooth speaker — JBL, AirPods, whatever you have — and Puente AI automatically routes the translation audio to it. No button to press, no setting to change. On Android it wakes up as soon as the speaker connects. On iOS it switches the moment audio routes. Just pull it out and start talking.

Trained on your field, not just the dictionary.

Generic apps translate words. Profession Packs teach Puente AI the vocabulary your job actually uses — clinical terms, legal language, OSHA safety words, restaurant service phrases. 9 packs available (Medical, Legal, Trades, Restaurant, Finance, Education, Childcare, Emergency, Faith) — included with Plus and up as part of the linguistic premium.

One app. Every situation.

Counter conversation, doctor's office, job site, video call, family dinner — Puente AI has a mode for each one. Auto-detect listens hands-free. Tabletop plays out loud for a group. Earbud mode gives each person private audio. Remote connects two phones from anywhere. Group mode handles up to 8 people at once, tracking every speaker separately.

The other person hears what you meant — not just what you said.

Word-for-word translation often sounds wrong or loses the meaning entirely. Puente AI builds a running picture of the whole conversation — what you're talking about, the tone, who's speaking — and uses that to make every translation sound natural on the other end. Idioms get localized. Slang gets adapted. The register matches the situation.

The right tone without thinking about it.

A doctor visit sounds different from a restaurant order. Puente AI knows that. Set it to Auto, Formal, or Casual — or just activate a Profession Pack and it sets the right register automatically. Medical conversations come out formal and precise. Service industry conversations come out natural and friendly. You focus on talking, not on settings.

Your conversations stay yours.

No account. No login. No email required — ever. Puente AI doesn't store your audio anywhere. Nothing is sold, nothing is tracked. When you're offline, everything stays on your device. Even in cloud mode, audio is processed and immediately discarded. Your history lives only on your phone and you can delete it anytime.

Smart glasses, bone conduction, lapel mics — just works.

Smart glasses (Ray-Ban Meta, Xreal, Engo) auto-connect via A2DP and route translation to your lenses speakers — eyes up, hands free. Bone conduction headphones (Shokz, AfterShokz) are detected automatically and switch to single-channel solo mode. Professional lapel mics (Røde, DJI, Hollyland, Deity, Sennheiser) are auto-selected as the highest-priority input with no configuration needed. Connect your Apple Watch or Wear OS watch to trigger recording with a button press using Push-to-Talk mode.

Empathy Engine — your tone, translated.

Voice matching handles the who. The Empathy Engine handles the how. It reads six emotional dimensions in your voice in real time — pause density, vocal tremor, onset sharpness, dynamic range, rhythm regularity, and sustained vowel ratio — to build a full emotional fingerprint. Then it shapes the translated voice to match: slowing down when you're being deliberate, adding weight when you're emotionally present, staying smooth when you're calm. The person on the other end doesn't just hear your words translated. They hear your tone translated. If you're telling someone something important, the translated voice carries that weight.

Live Captions — see it as it's spoken.

A real-time caption bar floats on screen as translations arrive — showing exactly what was said and what it means, simultaneously. The speaker's name appears if known. Works on your phone screen today, with AR glasses caption output coming next. Set it to phone-only, glasses-only, or both from Settings.

Voice Identity — know who said what.

Puente AI learns who's speaking using voice fingerprinting and directional audio cues. In group conversations, each speaker gets their own color-coded lane in the transcript — so you always know who said what, not just what was said. Speaker attribution is live and automatic, with no setup required.

Emotion-Aware Translation — tone travels too.

Beyond the six vocal dimensions of the Empathy Engine, you can now manually lock a translation register: Auto, Formal, Casual, or Domain-specific for medical, legal, and trades contexts. The output tone adapts to match — and a badge on each translation card confirms which register was used. The Worker echoes back exactly how your words were delivered.

Voice Passthrough — your voice, their language.

When enabled, translated audio sounds like the original speaker — not a generic voice. Pro users can clone their own voice with explicit two-step consent. If cloning fails or times out, it falls back to natural TTS automatically. Your voice, in their language.

Mesh Rooms — one QR code. Every language.

Start a live translation room and share a QR code. Anyone who scans puente.chat/r/[roomID] joins and sees captions in their own language — in real time. No accounts, no setup. The host's local translation runs unaffected even if the room fails. Rooms auto-expire after 4 hours. Perfect for meetings, classrooms, and events with mixed-language groups.