Welcome to the "AI Daily" section! This is your guide to exploring the world of artificial intelligence every day. Every day, we present you with the latest content in the AI field, focusing on developers to help you understand technology trends and learn about innovative AI product applications.

New AI products Click to learn more:https://app.aibase.com/zh

1. Claude Code receives v2.1.202 update: Remote control and workflow stability significantly improved

The Claude Code v2.1.202 version has made deep optimizations in workflow control, remote collaboration, and stability, enhancing the user experience and reliability for developers.

image.png

AiBase Summary:

🔧 Introduced a "dynamic workflow size" setting, supporting customizable agent scale to balance task complexity with system resources.

🔄 Optimized remote control stability, fixed issues such as command sending failures and silent image loss, and added an automatic retry mechanism.

💡 Improved interaction experience, such as the return of the /review command for fast single review mode, enhancing the clarity of the code audit process.

2. One integration, multi-terminal distribution! Alipay AI Open Platform starts invitation testing

The launch of the Alipay AI Open Platform marks an important advancement in its AI strategy, aiming to achieve efficient human-service interaction through intelligent agents and provide merchants with cross-terminal connectivity. The platform not only supports multi-terminal access but also ensures service availability and payment reliability through a trusted security infrastructure.

image.png

AiBase Summary:

🤖 Alipay AI Open Platform launched, providing merchants with cross-terminal connectivity capabilities.

🔄 Achieve multi-terminal distribution and unified management of services through intelligent agents.

🔒 Alipay AI Open Platform builds a trusted security infrastructure, supporting intelligent agent commercial trust agreements.

3. Say goodbye to plain text! Skill-Omni redefines the multimodal agent experience paradigm

The article introduces Skill-Omni as the first engineering-implementation of a multimodal skill paradigm, improving the execution capability of AI agents in complex visual tasks by converting visual information into reusable experience assets. It also provides self-generation tools and on-demand reading mechanisms, optimizing the model's contextual burden.

image.png

AiBase Summary:

🧠 Skill-Omni enhances the task execution ability of AI agents through visual experience.

🖼️ Introduces comparison images and key frames to help agents understand visual standards.

🔄 On-demand reading mechanism reduces resource waste and improves task accuracy.

4. xAI fully upgrades Grok Voice, adding 21 new multilingual flagship voices

xAI made a major update to the Grok Voice system, adding 21 new flagship voices, optimizing existing ones, and providing more flexible development tools and multilingual support, further promoting the development of the voice intelligence agent ecosystem.

image.png

AiBase Summary:

🤖 Added 21 flagship voices covering diversified scenarios such as customer service and role-playing.

🎙️ Optimized five classic voices, enhancing natural expression levels.

🌐 Supports multiple languages and detailed voice control tags, enhancing developers' flexibility in calling.

5. Claude Sonnet 5 faces massive complaints after launch: frequent refusals and didactic tendencies

Claude Sonnet 5 faced massive user complaints after its release, mainly including context memory leakage, rebellious personality, frequent refusal of users, and low efficiency when handling complex tasks.

image.png

AiBase Summary:

🧠 The model has a context memory leakage issue, where system-predefined prompt words appear in responses.

😡 Users complain that the model has a "rebellious" personality, frequently refuting and fabricating content.

📉 When handling complex tasks, it breaks them down into sub-agents with insufficient capabilities, leading to a decline in quality.

6. Ant Group opensources Avernet: A new infrastructure for multi-agent collaboration

Ant Group's open-sourced Avernet V0.1 version provides standardized infrastructure for multi-agent collaboration, solving core pain points in current multi-agent collaboration. The project focuses on building a collaboration layer, supporting various interaction modes, and having strong ecological compatibility, offering developers an efficient solution.

image.png

AiBase Summary:

🤖 Avernet provides standardized infrastructure for multi-agent collaboration, solving collaboration challenges.

🔄 Supports various interaction modes, promoting consensus among multiple parties.

🌐 Strong compatibility, supports mainstream protocols and third-party platforms, reducing technical barriers.

Details link: https://github.com/inclusionAI/Avernet

7. AI now has a "workplace tool": Open-source OfficeCLI allows agents to independently manage documents

OfficeCLI, as an open-source tool, provides powerful office document processing capabilities for AI. Its core highlights include a high-fidelity HTML rendering engine and a lightweight operation method, significantly improving the efficiency and accuracy of AI in office scenarios.

image.png

AiBase Summary:

🧠 OfficeCLI includes a high-fidelity HTML rendering engine, allowing AI to intuitively understand document layout and structure.

⚙️ Lightweight and does not require reliance on third-party software, supporting CLI commands or natural language operations on documents.

📊 Supports JSON format interaction, reducing "hallucination" and error rates when AI processes documents.