Open-source · Self-hosted · Encrypted

Talk to anyone.
In any language.Without compromise.

LinguaChat is the world's first bilingual messenger that combines context-aware AI translation, Signal-grade encryption, and a fully open-source self-hosted stack — built for a world with 7,000 languages.

200+
Languages
E2E
Encrypted
100%
Open-source
A
Ahmad Hassan
Arabic ↔ English
Hey! Are we still on for the meeting tomorrow?
هل ما زلنا نلتقي غداً؟
نعم، الساعة الثالثة عصراً مناسبة تماماً!
Yes, 3 PM works perfectly!
Great, I'll send the agenda shortly.
رائع، سأرسل جدول الأعمال قريباً

Everything you need.
Nothing you don't.

Built on open-source foundations so you stay in control of your data, your translations, and your infrastructure.

🌐
200-Language Translation
Powered by Meta's NLLB-200 model, supporting 200 languages with 94%+ accuracy. Ollama + Mistral for nuanced context-aware conversations.
🔐
Signal-Grade Encryption
X3DH key exchange and Double Ratchet algorithm. Every message encrypted with AES-256-GCM before it leaves your device.
🧠
Context-Aware AI
The last 10–20 messages are passed to the translation model as context. Slang, names, and ongoing topics stay consistent throughout.
🖥️
Fully Self-Hosted
Run LibreTranslate and Ollama on your own server via Docker. No third-party API — your data never touches external servers.
📱
Cross-Platform App
React Native + Expo delivers native iOS and Android apps from a single codebase, with offline language detection via on-device models.
Real-Time Delivery
Socket.IO and Redis Pub/Sub power sub-100ms message delivery. Supports group chats, broadcast channels, and presence indicators.

Four steps to
borderless conversation

1
You send a message
Your message is encrypted client-side using the Signal Protocol before leaving your device.
2
Server translates
The server briefly decrypts, passes chat history + new message to NLLB-200 or Ollama, gets the translation.
3
Re-encrypted delivery
Both original and translated versions are re-encrypted and delivered to the recipient in milliseconds.
4
Dual-bubble display
Recipient sees their language. Both parties can tap to reveal the original. Full transparency, always.

Exclusive features
no competitor offers

💬
Dual-bubble messaging
Every bubble shows both languages. Tap to reveal the original. Builds trust and enables passive language learning.
🎯
Tone-aware translation
Formal business or casual friends — the LLM adapts translation style to the conversation context automatically.
📖
Per-contact glossary
The app learns names, nicknames, and recurring terms per contact and applies them consistently forever.
🟢
Confidence indicator
Color-coded confidence on every translated bubble. Tap amber or red to see alternative translations.
🎤
Voice note translation
Send a voice note, recipient gets it transcribed and translated via Whisper + NLLB. Original audio still playable.
🌍
Cultural context notes
When an idiom doesn't translate cleanly, a footnote explains the cultural context — respectfully, not patronizingly.
👥
Multilingual group chat
5 people, 4 languages — each member sees every message in their own language. Translated once per target language.
"Say it better" assistant
Before sending, tap to ask the LLM for a more natural phrasing in the recipient's language and culture.

Built on proven
open-source foundations

Translation engine
NLLB-200 + Ollama
Meta's 200-language model for fast phrase translation. Ollama + Mistral for context-aware conversation translation.
NLLB-200OllamaLibreTranslate
Encryption
Signal Protocol
X3DH key exchange + Double Ratchet algorithm. AES-256-GCM for all stored messages. Forward secrecy by default.
libsignalX3DHAES-256-GCM
Real-time
Socket.IO + Redis
WebSocket-based delivery with Redis Pub/Sub fan-out for multi-server scalability. Sub-100ms latency.
Socket.IORedisNode.js
Mobile client
React Native + Expo
Single codebase for iOS and Android. On-device language detection via franc. Secure enclave key storage.
React NativeExpoKeychain
Database
PostgreSQL + Prisma
Encrypted message storage with Prisma ORM. Redis for sessions, presence, and translation caching.
PostgreSQLPrismaRedis
Infrastructure
Docker + Self-hosted
One-command Docker Compose deployment. Runs on any VPS. No vendor lock-in, no third-party API costs.
DockerFastifyRailway

Simple, transparent pricing

Start free forever. Scale when you're ready. No per-message fees, ever.

Community
Free / forever
Self-hosted open-source edition. Full control, full privacy.
  • Unlimited messages
  • 200-language translation
  • E2E encryption
  • Self-hosted on your server
  • Community support
Enterprise
Custom
On-premise deployment, SSO, audit logs, white-label.
  • Unlimited users
  • All Business features
  • On-premise deployment
  • SSO + audit logs
  • White-label option
  • Dedicated support

We believe language
should never be a wall

LinguaChat is built by Tutorns — a team that has experienced language barriers firsthand, from business negotiations lost in translation to personal connections that never formed because no common language existed.

We set out to build the messenger we always wanted: one that speaks your language automatically, respects your privacy completely, and runs on infrastructure you control.

By combining open-source AI translation with Signal-grade encryption, we're creating a world where the language you were born into doesn't determine who you can communicate with.

Privacy firstYour conversations belong to you. We design for zero-knowledge by default.
Open by defaultEvery line of code is open-source. Inspect it, fork it, contribute to it.
Accessibility for allThe free tier is genuinely free. Language access shouldn't cost money.
The team
S
Shoaib
Founder & CEO · Vision & Product
?
CTO · Open role
Signal Protocol · React Native · ML infra
?
AI/ML Lead · Open role
NLLB-200 · Ollama · NLP engineering
?
GTM Advisor · Open role
B2B SaaS · European enterprise sales
📍 Based in Europe · Backed by Tutorns · Seed round open · Reach out →

Latest from LinguaChat

🔐
Engineering
How we combine Signal Protocol with real-time AI translation without breaking E2E encryption
A deep dive into our server-side decryption approach, why it's a pragmatic tradeoff, and the client-side alternative we're building next.
April 20268 min read
🌍
Product
Why context matters: how 10 messages of history makes translation 3× more accurate
We ran tests comparing single-message translation vs. history-injected prompts across 12 language pairs. The results were striking.
March 20266 min read
📊
Market
5–6 billion people are locked out of natural communication. Here's how we're changing that.
The data behind language barriers in global messaging, and why now is the right moment to solve it with open-source AI.
February 20265 min read

Let's build
together

Whether you're an investor, a potential team member, an enterprise customer, or just someone who believes language shouldn't be a barrier — we want to hear from you.

Ready to break the language barrier?

Join thousands on the waitlist. Free forever for self-hosted. No API costs. No lock-in.

Get early access →