UX Experts Advocate for 'Agentic UX' Beyond Chatbots

- The shift to agentic UX requires designing for delegation, not just interaction; core principles include ensuring user control through visible override mechanisms, providing transparency into the agent's reasoning and status, and designing for asynchronous, multi-step tasks. This contrasts with traditional chatbots, which are typically rule-based and reactive, whereas AI agents are autonomous, goal-driven, and capable of planning. - Open-source multi-agent orchestration frameworks are becoming critical for development. Projects like CrewAI, Microsoft's AutoGen, and LangGraph provide architectures for coordinating specialized agents, enabling them to collaborate on complex tasks that a single agent could not handle alone. These frameworks manage challenges like task delegation, shared memory, and ensuring context is maintained as workflows are handed off between agents. - Research in AI agent architecture is heavily focused on improving reasoning, planning, and memory. A key challenge is effective context management, which allows agents to maintain coherent interactions over extended periods. Papers from institutions like the University of Greater Manchester and surveys on arXiv are exploring new architectures that combine Large Language Models (LLMs) with mechanisms for self-reflection and advanced task-planning. - Scaling AI engineering teams presents unique challenges beyond simply increasing headcount; leaders find that velocity per person can plummet without a deliberate focus on culture, clear ownership, and adapting development processes for non-linear AI projects. Successful organizational models include centralized Centers of Excellence (CoE), as seen at companies like JPMorgan Chase, and decentralized models where data scientists are embedded directly into product teams. - In China, the AI agent landscape includes consumer-focused startups like Manus and enterprise platforms such as GPTBots.ai. Major technology firms are also key players, with Tencent offering its Hunyuan AI platform, Baidu providing the Wenxin (ERNIE Bot) ecosystem, and Ant Group developing the Lingji platform for financial scenarios. - China's AI regulatory environment is managed by the Cyberspace Administration of China (CAC) and currently lacks a single comprehensive law. Instead, the government employs a strategy of targeted regulations, such as the "Interim Measures for the Management of Generative Artificial Intelligence Services," combined with national standards and pilot programs in major tech hubs like Shanghai and Shenzhen.