一种名为Moltbot的代理型AI助手在硅谷及科技圈迅速走红，用户将其深度接入个人与工作系统，自动化程度远超传统语音助手。该工具可常驻本地计算机，连接多种AI模型、应用与在线服务，通过WhatsApp或Telegram交互，执行跨应用的复杂任务。其病毒式传播始于1月初，用户反馈显示，许多人已将其用于日程管理、财务处理、信息汇总，甚至股票研究与自动下单，显示出高度“生活接管”特征。
Moltbot由独立开发者Peter Steinberger于2025年11月以Clawdbot之名发布，2026年初更名。其核心理念是将现有模型与服务“胶合”在本地运行，以减少对云端的依赖并增强数据主权。尽管架构并不复杂，但用户量在短时间内激增，新手大量涌入其Discord社区。该工具支持长期记忆与人格配置，使其在体验上更接近“真实助理”，推动了采用速度。
与此同时，风险与成本同样显著。安装需命令行操作、API密钥配置，部分用户报告误删数据或产生高额推理费用。安全隐患包括个人信息泄露与“提示注入”攻击风险。尽管如此，用户仍愿意在权衡后扩大使用范围，例如将其用于家庭企业管理。整体来看，Moltbot的爆红反映了2026年前后代理型AI的拐点：在效率与隐私、安全之间，越来越多用户选择了前者。
An agentic AI assistant called Moltbot has gone viral across Silicon Valley and the tech community, with users deeply integrating it into their personal and professional systems. Unlike traditional assistants, it runs persistently on a local computer, connects multiple AI models, apps, and online services, and communicates via WhatsApp or Telegram to execute complex, cross-application tasks. Since early January, adoption has surged, with users delegating scheduling, finances, information management, and even stock research and purchasing, indicating a high level of behavioral reliance.
Moltbot was released in November 2025 by independent developer Peter Steinberger under the name Clawdbot and rebranded in early 2026. Its design focuses on locally orchestrating existing models and services to preserve user data ownership while delivering powerful automation. Although technically simple in structure, demand spiked rapidly, flooding its Discord with new users. Features such as persistent memory and configurable personalities make it feel more like a real assistant than standard chatbots, accelerating uptake.
The risks and costs are significant. Setup requires command-line work and API keys, with reports of accidental data deletion and high inference bills. Security concerns include potential data leakage and prompt-injection attacks. Even so, many users accept these trade-offs and plan broader deployments, including small business management. Moltbot’s rapid rise highlights a 2026 inflection point for agentic AI, where users increasingly prioritize automation gains over privacy and security constraints.