Keep on to blur preview images; turn off to show them clearly
splitting one of these with a friend like divorced parents. training it to gaslight and respond to orders in exotic languages
investing @a16z // curating https://t.co/ssslqn6eo7
RT @yanatweets: My boss and big Satya 👀
Parable Founder & GP @parable_vc | x-a16z: @whatnot @livekindred @prismsofreality | Biggest fan @spenserskates | “It’s always Day 1”
RT @spenserskates: "The way you produce your artifact, your software, is changing. The product development process is completely getting ri…
Parable Founder & GP @parable_vc | x-a16z: @whatnot @livekindred @prismsofreality | Biggest fan @spenserskates | “It’s always Day 1”
You mad geniuses @daniel_eckler @WizLikeWizard
GM @speedrun. Investor at @a16z. Apply here: https://t.co/bbxTvgMK6w You've probably played a game I worked on. Husband, girldad x2, cook, point guard
最近在选择自己开发的 AI 工具栈，有一个感觉：Claude Code/Codex 等 AI Agent 的整体规划能力、架构能力都很不错，不过用来做前端开发，总感觉有些地方不太用得上力，比如设计团队的 Figma 设计稿还原，比如基于前端组件库和 CSS 样式来新增和修改产品设计。可能把 Spec 和 Plan、Skills、Agents md 等编写到位，多次迭代后，也能达到满意的结果。 不过感觉有些大炮打蚊子了，token 成本也是没必要的高！ 最近几天试用 Kombai 这个前端专精智能体，在前端开发方面值得试试，它通过领域优化（如内置 30+ 库 RAG 和任务特定规划），自研的 Figma 设计稿引擎等，直接生成高保真输出——少编译错、代码审查通过率高，还附带浏览器预览。用户能控制上下文，确保适配你的栈。 基准测试（200+ 任务）显示，它在 UI 实现和错误率上超通用智能体 2 倍左右，特别适合中大型组件。试了几个项目，确实高效，但也需熟悉其规划模式。 https://t.co/dSOvfD53YF
专注 - Context Engineering, AI(Coding)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 🔗 信息卡提示词 🔽
RT @lsanger: Not taking a position on controversial issues is what defines neutrality. Declaring only one position legitimate, as Wikipedia…
Subscribe https://t.co/Xm1OaUU8jk • seed investing • writing • ॐ •🙏• I use '—' • tweets saved 90 days • 📷