Opus 5.5 一镜到底剪出 Actuity 产品宣传片
作者用 Opus 5.5 一次成片,为做 AI agent 可用性测试的 Actuity 拍宣传视频,并列出 agent 浏览网页的失败率数据。
opus 5.5 is insane, cooked this video in one shot your website might work perfectly for humans. but what happens when an ai agent tries to use it? we've been building @Actuity for the past 2 months to figure this out. some numbers that made us take this problem seriously: - 75–80% of ai-agent traffic lands on product and search pages - agents can read websites pretty well, but the best one only completes 46.6% of action tasks - agents handle sliders only ~59% of the time and toggles ~86% of the time so we started building Actuity. we send real agents to your product and give them tasks your users' agents might actually try: - find a return policy - compare plans - search your catalog - change a billing toggle - complete an action then we show you where they succeed, where they break and why. and if it's something you can fix, Actuity gives you a prompt to fix it and sends the agent back to verify the fix. agents are becoming a new type of customer. your product needs to be ready for them too. launching Actuity very soon.
这条是引用着另一条写的:「Anthropic 发布 Claude Opus 5.5 模型」