正在加载视频...
视频加载失败
“What’s blocking the launch?” Every team asks a question like this. The answer is usually spread across Slack threads and Linear tickets. So we gave it to an agent running Meta Muse Spark on Cerebras. It searched both and came back with the blockers. The whole thing took 36... show more
20,908 次观看 • 3 天前 •via X (Twitter)
12 条评论

Congrats on the quarter, Andrew. Beat the guide, raised for year, backlog grew to 26.1B. I’m more bullish. This post illustrates the value, the user can feel 3,000 tokens a second on a live agent workload is the whole thesis. Agents don't wait well. Whoever answers fastest wins the workflow, and right now that gap isn't close. One genuine question: are customers starting to contract for the speed itself, priced as its own tier?

@cerebras Cerebras 🚀

@cerebras I hope the shares cross the ipo price

@cerebras You don't say 🤔

@cerebras شيء مجنون حقا لماذا لا تفكر باطلاق وكيل خاص بكم؟

@cerebras 36 seconds to dig through slack threads and linear tickets and come back with what is actually blocking the launch, that is wild. 3,000 tokens per second is a fun speed to watch too 🦍

@cerebras Does it cite which Slack thread each blocker came from? For a launch review I care less about how quick the answer arrived and more about whether I can click through and check it myself.

@cerebras @alexandr_wang 🤔

@cerebras 36 seconds is the easy half. blockers get found, fixed, forgotten, then re-derived from scratch next week. what should survive is the decision and who owns it, and memoryrouter keeps that part.

@cerebras 36 seconds changes behavior more than the benchmark does. At that speed people ask the agent before the standup instead of digging through Linear after it.

@cerebras For a launch-blocker summary, I'd want 'still blocked' separated from 'was blocked but resolved in Slack.' Add the owner and latest source beside each item, and those 36 seconds could save a lot of people reopening yesterday's argument.

@cerebras Meta collab incoming???

