Traini has built a high-quality multi-breed, multi-scenario pet audio-video dataset, continues to optimize its PPI (Pet Perception & Interaction) system with fine-grained annotation, and pursues lightweight, edge-deployed solutions to protect privacy and improve responsiveness in homes.

Traini’s core technology, PEBI (Pet Empathic Behavior Interface), is a multimodal system that analyzes audio, visual, and behavioral signals to translate pet emotions into conversational form, currently covering nearly 120 dog breeds with reported emotion translation accuracy up to 94%.

Traini reports serving over 2 million dogs, with YouTube videos surpassing 70 million views, and has launched PetGPT and T-Agent to boost engagement and autonomous product recommendations while collaborating with around 40,000 local U.S. pet stores.

The company announced a $7.5 million funding round to support research and development, product iteration, and market expansion.

Traini previously raised funds in May 2023 through an angel round with investors such as FutureX Capital, BlueSea Partners, Tao Foundation, Mint Capital, Valkyrie Fund, and Hong Feng.

Post-funding, Traini plans to launch the Cognitive Smart Collar, the world’s first cognitive pet wearable and first human–dog language translation device built on generative AI, available for pre-order via traini.app.

Founder Arvin Sun emphasizes the mission to strengthen the bond between pets and humans by reigniting natural instincts through smarter pet intelligence.

The core team includes veterans from OpenAI, Chewy, and ByteDance, underscoring the company’s focus on leveraging advanced AI to enhance pet emotion understanding.

The latest funding round was led by Banyan Tree, Silver Capital, ZhaoTai Group, and NYX Ventures, with participation from Starting Gate Fund, Jade Capital, Tao Foundation, Feng Hong, and other individual investors.

Traini emphasizes a data-driven, multimodal approach to overcome data scarcity and environmental complexity, using a train-as-you-use mechanism where anonymized pet interactions feed back into model training to build a large, high-quality pet audio-video dataset and enable continuous personalization.

Data strategy centers on multimodal fusion and a train-as-you-use loop to expand data, improve personalization, and address data-scarcity challenges.

The company reaches over 200 million dogs globally, with more than 60 million YouTube playback views; PetGPT (launched in 2023) has high adoption, and T-Agent (2024) supports product searches, recommendations, and automatic orders.