Tag
Tencent QQ Mail update: Supports one-time email rename, and Agently Mail renamed to Agent Mail.
Tencent launched a mailbox service specifically designed for AI Agents. After registration, users can obtain a prompt for completing CLI setup via Codex or other Agents.
Tencent launches Agently Mail for QQ Mail, a dedicated email service tailored for AI Agents.
Tencent is testing an AI assistant called Xiaowei within its WeChat app in China, aiming to catch up with rivals in the AI market by leveraging its massive user base.
Fenng explained the positioning difference between Tencent's WeLM (closed-source large model) and Hunyuan (open-source large model), pointing out that the closed-source model will not disclose technical details or participate in evaluations, and suggested understanding it through product experience.
Tencent open-sourced the enterprise knowledge platform WeKnora, featuring three key capabilities: RAG intelligent Q&A, ReAct autonomous Agent, and self-maintaining Wiki + knowledge graph, transforming raw documents into inferable and growing knowledge assets.
Tencent launches WorkBuddy, an AI agent that autonomously completes complex tasks instead of just answering questions, with a demo showing parallel agent execution.
A user shares their experience using Tencent WorkBuddy for competitive research. The AI deployed a research team, and a dedicated review AI caught a number without a source, rejected the report, and had it redone. The final output was a presentation-ready slide deck.
Tencent open-sourced Hy-Memory, a memory plugin for AI agents that provides long-term memory with a 6-layer dual-reasoning framework, reducing token usage by 35% and memory bloat by 70%.
Tencent WorkBuddy is a new PC-based AI agent that reads files, calls tools, writes reports, and uses expert teams to handle complex tasks, connecting to various services via MCP.
Tencent has open-sourced TencentDB Agent Memory, which solves the AI agent long-context overflow problem through hierarchical memory management (symbolic short-term memory + hierarchical long-term memory). Benchmarks show token consumption reduced by up to 61% and task success rate improved by over 50%.
Tencent Workbuddy is gradually becoming a phenomenal product and may have significant implications.
This report comments on Tencent directly embedding the Agent into WeChat's homepage right-swipe entry, rather than a standalone app, and leveraging WeChat's mini-program ecosystem, with the intention of turning WeChat into an "Agent operating system."
The article observes that Tencent's Hy3 Preview open model performs surprisingly well in evaluations, narrowing the gap with top closed models, yet remains underdiscussed compared to Western AI labs.
BIS issued guidance requiring licenses for advanced AI chip exports to Chinese-headquartered firms located outside China, highlighting previous enforcement gaps for overseas subsidiaries like Tencent Malaysia buying Nvidia Blackwell chips.
A Chinese entrepreneur tells the story of a Beijing startup that created India's top news app with just 40 Chinese engineers, none of whom knew Hindi, and secured $50 million in investment from Tencent.
A mysterious model called Hy3 from Tencent has unexpectedly topped OpenRouter's LLM rankings by token usage, despite mediocre benchmark performance and low public awareness. The article investigates the anomaly using OpenRouter's public data.
Tencent tech expert releases a 37-minute AI programming tutorial, teaching zero-basis users to develop a local errand mini-program and APP, while also introducing the Rapid-MLX tool, claiming it is 2-4 times faster than Ollama on Apple Silicon.
Tencent's Hy-MT2 model has been relicensed under Apache License 2.0, granting full freedom for research, commercial use, and modification. The two model variants are currently ranked #1 and #4 on the Hugging Face trending leaderboard.
Tencent's App Bao team launches Marvis, an OS-level AI assistant that supports end-cloud collaboration and fully local privacy mode. It can deeply control the system, files, applications, etc., and features a multi-Agent parallel collaborative architecture, with 10 million tokens free daily.