E256 | Agents, robots and drug discovery: which gets rewritten by GPT-6 Astra first?E256｜Agent、机器人与制药，谁先被GPT-6 Astra改写？
E256｜Agent、机器人与制药，谁先被GPT-6 Astra改写？E256 | Agents, robots and drug discovery: which gets rewritten by GPT-6 Astra first?

Podcast硅谷1012026-10-19· 1:19:404

Guests: 许华哲 · 朱哲清 · 王晟 · 破壳机器人 / Pokee.ai / 华盛顿大学

Three researchers on GPT-6 Astra's impact on personal agents, robotics and AI drug discovery三位一线研究者谈 GPT-6 Astra 对个人 Agent、机器人与 AI 制药的冲击

三位一线研究者谈 GPT-6 Astra 对个人 Agent、机器人与 AI 制药的冲击Three researchers on GPT-6 Astra's impact on personal agents, robotics and AI drug discovery

Key points

Where personal-agent monetization is stuck: computer use unlocks legacy biomedical software; will Meta Muse repeat Manus’s path, and is hardware the key?
个人 Agent 商业化卡在哪：Computer Use 让生物医学“古董级软件”可被操作，讨论 Meta Muse 是否会走 Manus 老路、软硬件结合是破局关键。
Robotics: Astra topped robot benchmarks; the shock favors hardware makers, and the future needs fast decision models plus Astra-level reasoning.
机器人：Astra 空降机器人评测榜首，冲击利好本体硬件，未来需要“快决策模型 + Astra 推理能力”。
AI pharma: Astra completes only ~30% of experiment design; molecules are designed fast but validated slowly, and wet labs will connect to models via MCP/API.
AI 制药：Astra 的实验设计完成度只有约 30%，AI 设计分子快、真实验证慢，湿实验公司将用 MCP/API 直连大模型。

MuseAgentsResearchModels