AI 工具与 Agent 必读

过去 24 小时，31 个来源有 134 条相关新消息，这里是被报道最多的 8 条。

1. Meta releases open-weight Muse Spark AI model for coding

   Tech in Asia · techinasia.com · 另有 4 家在报

   They can offer businesses a lower-cost alternative to closed models controlled by their creators.

2. Amazon announces Bee AI wearables for chat to-dos

   Tech in Asia · techinasia.com · 另有 1 家在报

   Bee, a personal AI hardware startup that Amazon said on July 22, 2025 it planned to buy, developed the devices.

3. Trump posts AI-edited photo to brag about White House restoration work

   The Verge · theverge.com · 另有 1 家在报

   President Donald Trump posted a series of photos on Truth Social showing off renovations to the White House - but one of them appears to be modified with AI. The picture, which compares an old White House column to a…

4. Oslo-based AI-native engineering firm Solvera raises €5.62 million pre-Seed after July 2026 launch

   EU Startups · eu-startups.com · 另有 1 家在报

   Solvera, an Oslo-based company building an AI-native engineering firm for the construction industry, today announced a €5.62 million ($6.3 million) pre-Seed round. The round was led by Seedcamp, with participation from…

5. British Business Bank invests £180M across AI and deeptech funds

   Tech.eu · tech.eu · 另有 1 家在报

   The British Business Bank has today announced it is investing £180m across six AI and deeptech VC funds, as investors continue to pile into the two hot sectors.The state-owned investor is investing £3...

6. Muse起势、大厂谋变，个人智能体激战前夜

   虎嗅 · huxiu.com · 另有 1 家在报

   本文来自微信公众号： 来咖智库 ，作者：龚岩，编辑：G3007 新玩法和老问题并存。 个人智能体，正在迎来新的开局。 自一个月前，Meta推出个人智能体Muse，并且登顶美加iOS免费榜后，中国大厂也再次上演神级反应。字节跳动加速推进代号“Spell”的个人助理项目，腾讯上线LightVela，阿里千问即将发布“千问Go！”，蚂蚁“灵光”开启小范围内测对标Muse，另外还有独角兽Manus推出的独立个人智能体应用Cue…

7. ClawLoop: Self-Improving Agents via Harness-Native Training Loop

   ArXiv AI · arxiv.org · 另有 1 家在报

   The rapid advancement of Large Language Models (LLMs) has driven their deployment toward general-purpose harness environments, where LLM-based agents operate in sandbox workspaces and interact directly with file systems…

8. Should SLMs be in the Driver's Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs

   ArXiv AI · arxiv.org · 另有 1 家在报

   The most accurate agents executing user-provided tasks are driven by large, cloud-hosted models, which solve them by reading the private data of the user. Agents powered by models small enough to run on edge devices…