LogoThread Easy
  • 発見
  • スレッド作成
LogoThread Easy

Twitter スレッドの万能パートナー

© 2025 Thread Easy All Rights Reserved.

探索

Newest first — browse tweet threads

Keep on to blur preview images; turn off to show them clearly

this is a really cool thread to think about:
1. what should a realtime omnimodal multiagent experience feel like for the user and the agents
2. why is it hard today and how will we temporarily do a fix until it does

if you’ve messed around with realtime APIs, you quickly see that the multi-channel observe/output is difficult, lots of awkward uncertainty in when to act and when to observe.

this gets even harder when there’s A LOT of different actors producing outputs to be consumed, how do you manage all of these channels well?

so today we fix it with coordinated turn based agents and shared systems of record which is great bc it’s nice that stuff works today!  but we still have lots of cool stuff to explore around agents working managing multichannel info in realtime

also tldraw always drops bangers

this is a really cool thread to think about: 1. what should a realtime omnimodal multiagent experience feel like for the user and the agents 2. why is it hard today and how will we temporarily do a fix until it does if you’ve messed around with realtime APIs, you quickly see that the multi-channel observe/output is difficult, lots of awkward uncertainty in when to act and when to observe. this gets even harder when there’s A LOT of different actors producing outputs to be consumed, how do you manage all of these channels well? so today we fix it with coordinated turn based agents and shared systems of record which is great bc it’s nice that stuff works today! but we still have lots of cool stuff to explore around agents working managing multichannel info in realtime also tldraw always drops bangers

agents, harnesses, and evals @LangChainAI, prev @awscloud, phd cs @ temple

avatar for Viv
Viv
Sat Dec 20 23:59:21
RT @shao__meng: 评估 AI Agent 的上下文压缩策略

在处理软件开发等需要连续性的任务时,单纯追求上下文高压缩率并不足够,更重要的是压缩后是否保留了 AI Agent 继续有效工作的关键信息。 @FactoryAI 提出的结构化总结方法,在实际任务表现上优…

RT @shao__meng: 评估 AI Agent 的上下文压缩策略 在处理软件开发等需要连续性的任务时,单纯追求上下文高压缩率并不足够,更重要的是压缩后是否保留了 AI Agent 继续有效工作的关键信息。 @FactoryAI 提出的结构化总结方法,在实际任务表现上优…

邵猛,中年失业程序员 😂 专注 - Context Engineering, AI Agents. 分享 - AI papers, apps and OSS. ex Microsoft MVP 合作 - 私信/邮箱:shaomeng@outlook.com 📢 公众号/小红书: AI 启蒙小伙伴

avatar for meng shao
meng shao
Sat Dec 20 23:57:57
RT @shao__meng: Google 性能优化之道:实战技巧与工程思维

Google 两位传奇杰出工程师 Jeff Dean 和 Sanjay Ghemawat 撰写的经典技术指南,基于 Google 多年来在构建高性能软件中的实战经验,总结了一套关于性能优化的原则和…

RT @shao__meng: Google 性能优化之道:实战技巧与工程思维 Google 两位传奇杰出工程师 Jeff Dean 和 Sanjay Ghemawat 撰写的经典技术指南,基于 Google 多年来在构建高性能软件中的实战经验,总结了一套关于性能优化的原则和…

邵猛,中年失业程序员 😂 专注 - Context Engineering, AI Agents. 分享 - AI papers, apps and OSS. ex Microsoft MVP 合作 - 私信/邮箱:shaomeng@outlook.com 📢 公众号/小红书: AI 启蒙小伙伴

avatar for meng shao
meng shao
Sat Dec 20 23:57:18
RT @shao__meng: 最近在集中看「上下文工程」的优质博客,除了 Anthropic Engineering 必看之外,还有几个官方博客推荐给朋友们(周末了,都别休息 😂)

Anthropic Engineering
https://t.co/bIFvBhVa0E…

RT @shao__meng: 最近在集中看「上下文工程」的优质博客,除了 Anthropic Engineering 必看之外,还有几个官方博客推荐给朋友们(周末了,都别休息 😂) Anthropic Engineering https://t.co/bIFvBhVa0E…

邵猛,中年失业程序员 😂 专注 - Context Engineering, AI Agents. 分享 - AI papers, apps and OSS. ex Microsoft MVP 合作 - 私信/邮箱:shaomeng@outlook.com 📢 公众号/小红书: AI 启蒙小伙伴

avatar for meng shao
meng shao
Sat Dec 20 23:57:07
Friendly reminder that people who literally spend every waking minute analyzing stocks have a hard time reliably beating index funds. 

If you happen to make a nice bit of money and let your ego convince you you’re an investing god, NGMI.

Stick in index + increase net income.

Friendly reminder that people who literally spend every waking minute analyzing stocks have a hard time reliably beating index funds. If you happen to make a nice bit of money and let your ego convince you you’re an investing god, NGMI. Stick in index + increase net income.

Founder & CEO @userp_io | The team behind the growth of monday,com, Robinhood, Wiz and 100s more. @forbes 30 Under 30 (2023)

avatar for Jeremy Moser
Jeremy Moser
Sat Dec 20 23:47:18
RT @tomieinlove: To give you an idea of how far AI has advanced, this is when an early neural network tried to come up with new names for c…

RT @tomieinlove: To give you an idea of how far AI has advanced, this is when an early neural network tried to come up with new names for c…

avatar for Joscha Bach
Joscha Bach
Sat Dec 20 23:46:40
  • Previous
  • 1
  • More pages
  • 238
  • 239
  • 240
  • More pages
  • 5634
  • Next