Perplexity has expanded its multi-model synthesis feature "Model Council" to its Computer platform, enabling users to compose boards with 2-8 AI models and visualize consensus and differences.
Tencent integrates Hunyuan's multimodal and LLM divisions, appointing Yao Shunyu to lead the foundational model unit. Expansion of agent products like Miora and WorkBuddy accelerates AI implementation.
The smartphone AI agent "Open Minis" has released its full code as open source for iOS and Android, ushering in a new era where AI directly operates devices through an embedded Linux environment and browser automation.
Evaluation methods for AI agents are undergoing a transformation. Traditional scoring systems are giving way to comprehensive quality management systems that integrate task definitions, execution traces, and process compliance. Evaluation is becoming a key determinant of competitiveness for next-generation AI products.
"Caveman," developed by Julius Brussee, converts AI coding agent responses into "caveman speak," cutting output tokens by about 65%. It supports over 30 agents, including Claude Code and Codex.
Open-source AI agent "OpenClaw" is now officially available as an app for iOS and Android, allowing users to run the agent on their smartphones. The project, which gained attention during the MoltBook controversy, enters a new phase.
Offering guidance on architecture selection for AI agent development in 2026. Compares ReAct, Plan-and-Execute, and Multi-Agent patterns based on performance, implementation cost, and scalability.
A comprehensive explanation of AI agent definitions, operating principles, major types (single/multi-agent), real-world enterprise use cases, and the latest trends as of 2026.
This site uses cookies for access analysis and ad delivery. By clicking "Accept", you consent to the use of cookies. See our Privacy Policy for details.