Daily AI News Brief - July 28, 2025

July 28, 2025 presents ten significant AI developments spanning open-source platform releases, creative workspace innovations, video editing breakthroughs, office intelligence upgrades, brand transformations, investment initiatives, web intelligence solutions, industry leadership metrics, financial AI standards, and immersive 3D modeling.

1️⃣ ByteDance Announces Open Source of Two Core Coze Projects: Coze Studio and Coze Loop

ByteDance has open-sourced the core projects Coze Studio and Coze Loop from its AI Agent development platform Coze, providing developers with low-barrier, high-flexibility tools while marking its strategic layout in the AI field and driving industry innovation through comprehensive agent development capabilities.

Revolutionary Open Source Features:Visual Development Platform: Coze Studio is a visual AI Agent development platform supporting drag-and-drop workflow construction, suitable for rapid AI application development and deployment
Lifecycle Management: Coze Loop focuses on AI Agent full lifecycle management, covering prompt debugging, performance evaluation, and other functions to help optimize agent performance
Apache 2.0 License: Both projects adopt Apache 2.0 protocol, allowing free use, modification, and even commercial deployment, reducing usage barriers

Available on GitHub at https://github.com/coze-dev/coze-studio, Coze Studio provides a complete AI agent development environment from development to deployment, enabling developers to focus on creating AI core value through prompt engineering, RAG, plugins, and workflows. The platform has served tens of thousands of enterprises and millions of developers, with microservices architecture built on Golang backend and React + TypeScript frontend.

2️⃣ Kling AI Officially Releases Lingdong Canvas Creative Workstation with Major Multi-Image Reference Upgrade

Kling AI announced two important developments at the 2025 World Artificial Intelligence Conference: launching the new creative workstation Lingdong Canvas and upgrading the multi-image reference functionality. The global creator base has surpassed 45 million users, demonstrating widespread application and increasing market recognition of AI-generated content.

Creative Platform Enhancements:

Infinite Visual Space: Lingdong Canvas provides unlimited visual space, intelligent creative assistance, and multi-person real-time collaboration features, enhancing creative efficiency
Enhanced Multi-Image Reference: Upgraded multi-image reference functionality shows significant improvements in character consistency, scene unity, and art style maintenance capabilities
Global Expansion: Kling AI user base exceeds 45 million, covering 149 countries and regions with rapid B-end customer growth

The platform represents advancement in AI-powered creative tools, enabling seamless collaboration and maintaining visual consistency across complex creative projects while serving a rapidly expanding global user base across diverse markets and use cases.

3️⃣ Runway Launches Revolutionary Video Editing Model Aleph

Runway has released the innovative video editing model Aleph, dubbed the Kontext model of the video field. The model enables diverse video editing through natural language instructions, including content addition/deletion, style transfer, and green screen compositing, dramatically lowering video creation barriers and providing unprecedented creative freedom for content creators.

Advanced Editing Capabilities:

Natural Language Control: Supports natural language instructions for diverse video editing including content modification and style transfer
Environment Transformation: Achieves environment and atmosphere changes, such as converting sunny scenes to rainy or adjusting to nighttime settings
Integrated Post-Production: Built-in green screen compositing and re-lighting capabilities enhance post-production efficiency

Available at https://runwayml.com/research/introducing-runway-aleph, Aleph represents a paradigm shift in video editing by enabling natural language control over complex editing operations, making professional video editing accessible through conversational interfaces.

4️⃣ Kingsoft Office Releases WPS AI 3.0 Lingxi Version: Building Native Office Agents

Kingsoft Office officially released WPS AI 3.0 version - WPS Lingxi at the 2025 AI Productivity Forum, marking office AI upgrade from tools to intelligent assistants. This version adopts native Office agent design, supporting natural language multi-round dialogue for document creation, presentation generation, and voice assistant functionalities.

Native Agent Features:

Multi-Round Dialogue: WPS Lingxi supports multi-round conversations, achieving seamless interaction for document creation and presentation generation
Format Preservation: Lingxi retains original formatting, ensuring AI-generated content meets actual user requirements
Integrated Knowledge: WPS knowledge base and AI search functionality integration provides comprehensive intelligent office solutions

The platform transforms traditional office software into intelligent assistants capable of understanding context and maintaining document integrity while providing comprehensive AI-powered productivity enhancements across the complete office workflow ecosystem.

5️⃣ JD Announces Brand Upgrade to JoyAI and Launches New Intelligent Brand JoyInside

JD announced at the 2025 World Artificial Intelligence Conference the upgrade of its large model brand to JoyAI and introduced the new intelligent brand JoyInside, demonstrating powerful capabilities and innovation in artificial intelligence while showcasing comprehensive AI ecosystem development.

AI Brand Evolution:

Comprehensive Model Series: JoyAI large model series covers multiple modalities, meeting diverse user requirements across various applications
Innovation Technology: Introduces innovative technologies to improve reasoning efficiency and reduce training costs for enhanced performance
Diverse Applications: JoyInside applications span humanoid robots, children's toys, and various other carriers for broad market deployment

The brand transformation reflects JD's strategic commitment to AI leadership, positioning the company as a comprehensive AI solutions provider across consumer and enterprise markets with integrated hardware and software offerings.

6️⃣ Pudong Launches 2 Billion Yuan AI Seed Fund with 500 Million Initial Focus on Early Investment

Pudong has announced a 2 billion yuan artificial intelligence seed fund with an initial 500 million yuan focused on early-stage investment, aimed at solving startup financing difficulties and promoting AI industry development through comprehensive funding support for emerging AI enterprises.

Funding Program Details:

Comprehensive Fund Structure: Pudong AI seed fund totals 2 billion yuan with initial 500 million yuan focused on seed round investments
Industry Leadership: Pudong AI industry scale exceeds 160 billion yuan, accounting for 40% of the city total, forming significant cluster effects
Five-Pillar Strategy: Five major initiatives promote AI ecosystem including scenario expansion, infrastructure development, and financial support

The initiative demonstrates Pudong's commitment to maintaining AI leadership through systematic investment in early-stage companies, infrastructure development, and ecosystem building to support sustainable AI industry growth.

7️⃣ Alibaba Open Sources WebSailor AI Agent, Breaking Information Retrieval Barriers

Alibaba Tongyi Laboratory has launched the open-source AI agent framework WebSailor, demonstrating excellent performance in complex task processing while enhancing information retrieval efficiency through reinforcement learning and complex task generation technology, providing developers with expanded exploration opportunities.

Advanced Agent Capabilities:

Complex Task Processing: WebSailor is an open-source AI agent framework specializing in complex task processing and information retrieval
Enhanced Learning: Reinforcement learning combined with complex task generation improves model performance and intelligent processing
Information Fuzzing: Information fuzzing technology increases task challenge levels, improving model intelligence capabilities

Available on GitHub at https://github.com/Alibaba-NLP/WebAgent, WebSailor addresses limitations in traditional web navigation by providing sophisticated reasoning and planning capabilities for complex information retrieval tasks across diverse digital environments.

8️⃣ China's Large Model Count Exceeds 1,500, Highlighting Global Leadership Position

China has achieved remarkable success in artificial intelligence with globally leading large model quantities, widespread generative AI product applications, and rapid industry scale growth. Additionally, domestic AI products are deeply applied across multiple industries with accelerated internationalization, indicating strengthened competitiveness in global AI.

Global AI Dominance:

Model Leadership: Global large model count reaches 3,755 with Chinese enterprises contributing 1,509, leading worldwide
Market Penetration: Generative AI product user proportion reaches 80.9% with continued industry scale growth
International Expansion: High-quality domestic AI products accelerate overseas expansion, significantly enhancing China's international influence

The statistics demonstrate China's dominant position in AI development, with domestic companies leading in model development, user adoption, and international market expansion across diverse application domains and geographic regions.

9️⃣ Ant Group Partners with Multiple Institutions to Release Finova Large Model Financial Application Evaluation Benchmark

Ant Group officially released the financial reasoning large model Agentar-Fin-R1 at the World AI Conference forum, demonstrating excellence in financial professionalism, reasoning capabilities, and safety compliance. Through comprehensive financial task data systems and innovative training algorithms, the model enhanced complex task processing while collaborating with multiple institutions on the Finova evaluation benchmark.

Financial AI Excellence:

Superior Performance: Agentar-Fin-R1 demonstrates excellence in financial domains, surpassing similar models in professional applications
Comprehensive Framework: Constructed comprehensive financial task classification system covering multiple financial scenarios and use cases
Open Benchmark: Finova evaluation benchmark open-sourced to help industry improve large model application levels in financial services

The development represents advancement in financial AI applications, providing industry-standard evaluation methods and promoting responsible AI adoption in financial services through comprehensive benchmarking and professional model development.

🔟 Tencent Hunyuan 3D World Model Shocking Release: 360° Immersive Scenes with Free Experience

Tencent released the Hunyuan 3D World Model 1.0 at the 2025 World AI Conference with comprehensive open-source availability. The model supports high-precision 360° scene generation and interactive roaming functionality, dramatically reducing 3D content creation barriers while bringing new possibilities to game development, VR, and digital content creation.

3D World Generation:

360° Scene Generation: High-precision 360° scene generation with immersive visual experiences and interactive navigation capabilities
Interactive Roaming: Advanced interactive roaming functionality enables users to explore generated environments naturally
Industry Applications: Breakthrough applications in game development, VR experiences, and digital content creation with reduced technical barriers

The open-source 3D world model democratizes access to professional-grade 3D content generation, enabling creators across industries to develop immersive experiences without traditional technical expertise or resource requirements for comprehensive virtual world development.