TWO is redefining Human-AI Interaction and building immersive AI experiences powered by SUTRA, the next generation of multilingual and cost-efficient Gen-AI engines.

Building AI
for next billions

in 50+ languages

Most LLMs are trained on data in English and struggle on multilingual tasks, leaving a massive gap of numerous languages — and potentially limiting access to the powerful technology for billions of people.

We are building AI for all languages, beyond English, beyond text and voice. We are building AI for all.

Introducing
SUTRA

SUTRA models are trained with an innovative LLM architecture that learns new languages independently, making it multilingual, highly-scalable, and cost-efficient. SUTRA excels in 50+ languages with conversation, search, and visual capabilities.

SUTRA surpasses other leading LLMs by 20-30% on the MMLU performance benchmark for non-English languages with high-accuracy and cost-efficiency.

SUTRA support 50+ languages including Hindi, Gujarati, Korean, Arabic, Japanese, and more. SUTRA's unique LLM architecture decouples concept learning from language learning, enabling faster training cycles and cost-efficient inference.

SUTRA API is available now.

Immersive
Experiences

beyond text and voice

We believe AI will become an integral part of all aspects of our lives, transforming communication, information, entertainment, and commerce. With the power of SUTRA, we are bringing AI directly to consumers with next-generation experiences.

ZAPPY - an AI-native messaging and social app powered by SUTRA that lets users befriend AI personas and interact with them in real-time voice and video calls.

Geniya - a mobile-first AI search experience that makes finding information fun and like talking to a friend.

Geniya is in public Beta and powered by SUTRA-Online.

Bridging the language gap in language models

SUTRA is an innovative multilingual LLM architecture that powers text generation, summarization, and instruction-following in 50+ languages.

Building multilingual AI for the next billions

Most LLMs are trained on data in English, leaving a massive gap of capabilities in most other languages. We are building AI for the 81% of the world that doesn’t speak English.

AI-native consumer experiences beyond text and voice

SUTRA models are the technology backbone of TWO’s products and services used by hundreds of thousands of people, growing every day.