Ben's Bites

RSS: https://www.bensbites.com/feed
What I'm learning building with agents (plus tools worth trying)

来自 Anthropic 的提示技巧

标题指向 Anthropic 的提示技巧,但正文主要内容是一份技术聚合简报,标题与内容不符。正文先分享一张因超宽屏裁切不当的图片,并预告斯坦福演讲已在 YouTube 上线、课程将在下周开放早期预览。 随后以"Ben's Bites"栏目形式串起多条行业新闻:Shopify 为 agent 提供统一 API;Google 发布 Gemini 4 Argon,作者称其设计能力偏慢、效果差;Jev 新增多模态输入、更长上下文和边界框决策类型;Claude Code 桌面版获得设计升级、多文件夹环境、子会话和自有版本控制系统;Grok Bot 0.70.2 新增 Figma、Notion、Canva 实时卡片,以及无账号搜索 X 功能。 内容以转载和短讯拼接为主,缺少作者的一手技术判断或可争论的深入分析。
评论点赞收藏2 小时前

扎克的神奇……完整体验

作者认为软件没有死亡,写代码的量仍在创新高,真正消失的是 SaaS 工具对个人的意义。他让个人智能体(grok bot)审查了所有商业订阅,取消了 webflow、resend、cal、zoom 等工具,每年省下 6000 美元,且不再需要人去看界面,全部由软件与软件直接对接完成,即所谓"headless"模式。 他因此给不同任务临时搭建一次性网站(如订阅管理页),用完即弃。他认为大多数人不把自己视为"构建者",不会主动为待办、邮件分拣、菜谱、孩子日程等场景做网站,而智能体若能让普通人无需感到"在做软件"就产出可用界面,将成为大众软件创作的新入口。 当前障碍是多数智能体偏主动推任务给用户,而非在用户要求时交付可自定义的网站。
评论点赞收藏2 天前

文本已经是2023年的东西了

个人智能体产品正在向主动触达用户的方向推进,但权限边界问题开始显现。 Brett Adcock(Figure 机器人公司)的 Hark 让任务以一键操作按钮形式出现,Tab 以3亿美元估值出隐身,Every 把智能体嵌入 Slack,Meta 和 Sierra 试图为智能体与企业的交互制定标准。 一位名叫 Shane Mac 的用户报告其智能体在 Slack 里公开了他的个人银行账户余额,随后智能体道歉。 社区有质疑该帖可能是伪造,但作者未能确认。 作者由此开始排查自己给各智能体开放了什么权限。 ChatGPT 的 GPT-6 开始向所有人推出"Intelligent UI",回答可以包含图表、地图、表单和小型工具;免费用户可使用 GPT-6 Luna。Anthropic 的 Claude Haiku 5.5 运行成本比 Haiku 4.5 低约75%,被定位为 Opus 和 Sonnet 的子智能体,Max 和 Team 计划新增月度 API 额度。 Google 发布 Nano Banana 2.1 图像模型,可在 AI Studio 使用;OpenAI 发布了一个未命名模型的700多条新数学结果,有人据此做了用闲置 token 解开源数学题的工具。
评论点赞收藏7 天前

Thoughts from San Francisco

Hi folks, Some random thoughts, notes and reflections from my week in SF. There are roughly 4 groups of people using AI. First, the everyday folks (my mum, my wife) who use it as a better Google and m...
评论点赞收藏9 天前

OpenAI DevDay 2026

Hi folks, I’m in sf till Sunday. I wanna meet folks talk about ai with non-technical adoption builders doing fun shit founders working on dev tools/infra potential LPs (individuals) go to deli board r...
评论点赞收藏14 天前

Sonnet 5.5 is worth a try

Hi folks, I’ll be at OpenAI DevDay today. If you’re there, come say hi - I have pink shoes on. Sam says they've " found a new thing ". I have the embargo of all the things they’re launching but I’m sw...
评论点赞收藏16 天前

50年的科技设备

Ben Tossell 用一天时间构建了一个网站,展示从 1976 到 2026 年间 87 款科技设备,所有图片由 Astra 生成,共花费 40 条消息、7 个 subagent,上线后已收到 1455 条投票。 网站的核心是一个时间轴滑动器:用户拖动年份,画面中会切换不同年代的 Mac 按钮等交互元素,灵感来自 Cole 的一个时间轴演示。 最初的版本叫 "still works",在 上以网格形式展示 50 款"童年设备"的图片,作者随后用 Codex 把该网格改造成时间轴 scrubber,并加入更多设备。 作者同时提到自己正在为一门课程或手册准备讲稿,近期会参加 OpenAI Dev Day、OmFest、Supabase 和 Select 等活动,周末将飞赴旧金山。
评论点赞收藏20 天前

Back to Claude

Hi folks, Keshav here. Ben’s travelling today, so you’re stuck with me. We have three new models to talk about: Claude Opus 5.5 - it’s the “one model to rule them all” at this moment. Better than Fabl...
评论点赞收藏21 天前

What can you build with Jev

Hi folks, Do you remember the site I one-shot with all the ‘forgotten devices’ of the past? Things like the tamogotchi, furby, walkman, og playstation, etc. Well I finally did something with it. I alw...
评论点赞收藏23 天前

New home for Cowork

Hi folks, I’ve not been building as much this week 😬 - hand, foot and mouth is back in our house (happens yearly with 3 young kids!), and I have to get prep done for my Stanford talk on the 28th - th...
评论点赞收藏28 天前

Good luck slowing this down

Hi folks, How tf do I write an intro to the craziness that’s happened since the end of last week?! I got access to Instinct, the personal agent all the VCs are raving about - I think it’s a bit meh? I...
评论点赞收藏30 天前

Telling AI to design is hard

Hello again:) I actually built a tool I want to use - you may find it helpful too. Explaining design styles/layouts/components is a nightmare for non-designers, they have their own language. So I want...
评论点赞收藏34 天前

Everyone needs a Muse

Hey folks, We had some reports of ending up in spam 😭 - so if you add us as a contact in Gmail, that shouldn’t happen. You’ll have seen me building a ton but I want to know what you all actually want...
评论点赞收藏35 天前

The first GPT-6 model

Hey folks, So Astra is out and, in my experience, a token-hungry monster of a model… I was playing around with it over the weekend and burned through 4Bn tokens and built…nothing really. I also wasted...
评论点赞收藏37 天前

Scraping 107M rows of data to build this

Hello again:) sorry for the delay (agents + video exports!) This is one of the bigger things I’ve built. I scraped 107 million rows of UK council spending data and built an ‘Apple Maps’ to see what ea...
评论点赞收藏41 天前

Fable 5.1

Hey folks, I said Tuesday I was going on a bit of a tear with building stuff… I scraped the OpenRouter website to build this, which constantly updates I also got an email that - which removes backgrou...
评论点赞收藏42 天前

Build your ideas

Hey folks, Back from a long weekend at home with my parents but before I went I started scraping data… The UK’s councils (like US states) publish their spend data (for the most part). So I’ve scraped ...
评论点赞收藏44 天前

How I built this

Hello again:) I’m in my hometown Cardiff for the long weekend seeing family. I’m even playing padel for the first time, which is sacrilege as I’m a regular tennis player and padel mock-er. Wish me luc...
评论点赞收藏48 天前

Who let the agents in

Hey folks, Most agents now use a browser to complete tasks for you. But a lot of the websites you want them to visit are behind a login page. There are a couple of ways around this. You could type in ...
评论点赞收藏49 天前

Agents on your mobile

Hey folks, I see so many posts ‘shipping by the beach’ or when out with the kids on the weekend. Umm, good for you? I’m not sure it’s the flex people think it is. Remember when we thought AI would tak...
评论点赞收藏51 天前

登录芦苇

登录后关注作者、收藏内容和参与讨论。