KYIV, Ukraine (AP) — The United States is offering Ukraine security guarantees for a period of 15 years as part of a proposed peace plan, Ukrainian President Volodymyr Zelenskyy said Monday, though he said he would prefer an American commitment of up to 50 years to deter Russia from further attempts to seize its neighbor’s land by force.
U.S. President Donald Trump hosted Zelenskyy at his Florida resort on Sunday and insisted that Ukraine and Russia are “closer than ever before” to a peace settlement.
Negotiators are still searching for a breakthrough on key issues, however, including whose forces withdraw from where and the fate of the Russian-occupied Zaporizhzhia nuclear power plant, one of the 10 biggest in the world. Trump noted that the monthslong U.S.-led negotiations could still collapse.
“Without security guarantees, realistically, this war will not end,” Zelenskyy told reporters in voice messages responding to questions sent via a Whatsapp chat.
Ukraine has been fighting Russia since 2014, when it illegally annexed Crimea and Moscow-backed separatists took up arms in the Donbas, a vital industrial region in eastern Ukraine.
Details of the security guarantees have not become public but Zelenskyy said Monday that they include how a peace deal would be monitored as well as the “presence” of partners. He didn’t elaborate, but Russia has said it won’t accept the deployment in Ukraine of troops from NATO countries.
Kremlin spokesman Dmitry Peskov said Monday that Russian President Vladimir Putin and Trump were expected to speak in the near future but there was no indication the Russian leader would speak to Zelenskyy.
French President Emmanuel Macron said Kyiv’s allies will meet in Paris in early January to “finalize each country’s concrete contributions” to the security guarantees.
Trump said he would consider extending U.S. security guarantees for Ukraine beyond 15 years, according to Zelenskyy. The guarantees would be approved by the U.S. Congress as well as by parliaments in other countries involved in overseeing any settlement, he said.
Zelenskyy said he wants the 20-point peace plan under discussion to be approved by Ukrainians in a national referendum.
However, holding a ballot requires a ceasefire of at least 60 days, and Moscow has shown no willingness for a truce without a full settlement.
Follow AP’s coverage of the war in Ukraine at https://apnews.com/hub/russia-ukraine
President Donald Trump and Ukraine's President Volodymyr Zelenskyy shake hands at the start of a joint news conference following a meeting at Trump's Mar-a-Lago club, Sunday, Dec. 28, 2025, in Palm Beach, Fla. (AP Photo/Alex Brandon)
President Donald Trump and Ukraine's President Volodymyr Zelenskyy shake hands at the start of a joint news conference following a meeting at Trump's Mar-a-Lago club, Sunday, Dec. 28, 2025, in Palm Beach, Fla. (AP Photo/Alex Brandon)
Ukraine's President Volodymyr Zelenskyy speaks during a joint news conference with President Donald Trump following a meeting at Trump's Mar-a-Lago club, Sunday, Dec. 28, 2025, in Palm Beach, Fla. (AP Photo/Alex Brandon)
PALO ALTO, Calif.--(BUSINESS WIRE)--Dec 29, 2025--
Palo Alto–based pet emotional intelligence startup Traini has announced the completion of a $7.5 million funding round, aiming to bridge the communication gap between humans and pets by developing intelligent systems capable of understanding pet emotions and intentions. The round was led by Banyan Tree, Silver Capital, ZhaoTai Group, and NYX Ventures, with participation from Starting Gate Fund, Jade Capital, and renowned tech investors including VPs from NVIDIA, Julian Qian (Member of technical staff at Anthropic), Weihe Zheng (Founding Partner, Cowin Capital), Jia He (Founding Partner, Nanshan Capital), Peter Xu (CEO, Plug and Play China), and Zach Zhang (Founding Partner, Edgewater Investments). Existing investors Tao Foundation and Xiaomi Co-founder FengHong also participated. The proceeds will be used primarily for technology R&D, product iteration, and market expansion.
This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20251229406590/en/
Prior to this, Traini also received investments from senior management at Google, Meta, and Palo Alto Networks and institutional investment from FutureX Capital, BlueSea Partners, Tao Foundation, Mint Capital, and Valkyrie Fund.
Following this round of financing, Traini will launch its first smart hardware product—the world’s first cognitive pet wearable, the Cognitive Smart Collar. This device is also the world’s first human–dog “language” translation device built on generative AI. It is now available for pre-order through the Traini App and the website, traini.app.
Traini founder Arvin Sun stated, “Our mission is to develop an intelligence that reignites the natural instincts of our furry companions and builds a true spiritual bond between pets and their humans.” Traini’s founding team brings extensive experience from companies like OpenAI, Chewy, and ByteDance.
Traini is the world’s first—and so far the only—technology company to translate pet behavior using multimodal generative AI. Its breakthrough technology has pioneered a new way for humans to communicate with their pets.
The defining feature of Traini’s product lies in its ability to enable real-time voice conversations through its proprietary multimodal models. Traini’s core product, PEBI (Pet Empathic Behavior Interface), supports multimodal interaction across text, images, video, and audio. By analyzing pets’ vocalizations, facial expressions, and behaviors, PEBI delivers a precise understanding of pets’ psychological and emotional states and expresses them in a human-like conversational form—achieving true, two-way communication between humans and pets.
Traini’s models cover nearly 120 dog breeds, with emotion translation accuracy reaching up to 94%. PEBI can be applied to behavior translation, emotion recognition, personalized service matching, and pet-assisted healthcare diagnostics. In addition, Traini provides APIs to veterinary clinics and hardware companies, empowering medical services and intelligent device interactions. Traini has established deep partnerships with leading smartphone and electric vehicle brands, and soon users will be able to communicate with their pets in real time directly through mobile operating systems and in-car infotainment systems.
In product quality validation, beyond collaborating with pet behavior experts for extensive data annotation, Traini has innovatively adopted a “human–pet vocal spectrogram comparison” methodology. By using the spectrogram of human speech—captured when expressing corresponding emotions and intentions—as a baseline, Traini compares it against pet vocal patterns to enhance emotional mapping accuracy.
Traini‘s smart dog collar integrates advanced AI with real-time emotion tracking. It is powered by Traini’s proprietary Valence–Arousal model, 3D Pet Emotion Model and Instant Emotion Vector. Developed in close collaboration with pet behaviorists, the model is built upon insights from more than 900 peer-reviewed studies in animal behavior and trained on behavioral data from over 2 million dogs. Designed to deepen emotional bonds between pets and their owners, the Cognitive Smart Collar helps pet parents gain a richer understanding of their dogs’ emotional well-being and overall mental health.
The collar's most remarkable feature lies in its ability to autonomously identify and track kinds of emotional states in dogs, including joy, anxiety, excitement, and distress. Translating them into human-understandable language enables real-time, meaningful two-way communication — including vocalizations, emotions, behaviors, and even what the dog “wants to say.” Using a combination of sound analysis from the dog’s barks and vital signals like heart rate, temperature, and body movement, Traini’s AI-powered collar interprets these data points to provide a comprehensive emotional profile of the pet. Traini's collar offers actionable feedback that can enhance the bond between owner and pet.
Beyond everyday care, the collar also has significant potential in the realm of pet healthcare. By continuously monitoring a dog’s emotional and physiological signals, it can help detect early signs of health issues.
On the user side, Traini has already served over 2 million dogs and received widespread coverage from major media and influencers. Its videos on YouTube have surpassed 70 million views. In 2023, Traini launched PetGPT, a natural-language and behavior-analysis model with 99% coverage across users, improving service engagement by 70%. Earlier this year, it unveiled T-Agent, an AI-powered product recommendation and purchasing system that autonomously identifies dogs’ real needs—allowing pets to make decisions, with their owners acting as co-pilots. Traini now collaborates with nearly 40,000 local pet stores across the U.S.
To realize this vision, Traini had to overcome two major challenges: data scarcity and environmental complexity. Traini addressed these through multimodal fusion, integrating image, video, and audio data to train its PPI (Pet Perception and Interaction) model. Powered by transformers, PPI combines real-time perception, adaptive reasoning, and feedback generation.
Traini also introduced a “train-as-you-use” mechanism, where every anonymized user-pet interaction feeds back into model training, enhancing personalization and continuously expanding the behavioral dataset. This approach strengthens the model’s precision while building a sustainable data moat. Traini has achieved breakthroughs at both data and model levels, developing a rich, high-quality pet audio-video dataset across multiple breeds, scenes, and emotions. As a result, users can detect early signs of pet stress, loneliness, or health risks, transforming Traini from a passive tool into an emotional interface for daily companionship.
Investors describe Traini as a breakthrough example of AI in vertical applications—where technology, empathy, and consumer experience converge.
The Traini cognitive smart collar, designed for real-time human–dog conversational AI. https://traini.app
