Latest

Every published item, in order · 5280 items
Nov 10
How-toBuildersMulti-source05:13
Chinese contentNVIDIA:比价目表没用,token 效率才是选模型的关键指标

NVIDIA 的人在 Open by Design NYC 活动上说,选模型别只看单价,token 效率才是决定真实成本的关键,训练自己模型的团队也适用。现场还有 Reflection 和 Together AI 的人一起聊开源模型和 post-training,做技术选型的可以听听他们的算法。

Event
Official accountTogether AI@togethercomputeSource
How-toBuildersMulti-source05:09
Chinese contentAnthropic 发布选型指南:Fable 5.2、Opus 5.5、Sonnet 与 Haiku 该怎么选

Anthropic 官方把四个 Claude 模型的适用场景讲清楚了,最实用的一条是子智能体用小模型跑、主任务留给旗舰模型来省成本。正在做多智能体或批量任务的开发者,照着配一遍就能省不少 token 钱。

Event
claudedevs@claudedevsSource
How-toBuildersMulti-source04:25
Chinese contentAnthropic 的 Alex Albert 分享一条模型验收提示词:让新模型通读项目找问题

Anthropic 员工 Alex Albert 的一条实用提示词,让新模型通读你的项目代码库,修掉有把握的问题、其余列出来给你审。比起跑基准分数,用真实项目验收新模型更直观,写代码的人都可以直接抄走用。

Event
Alex Albert@alexalbert__Source
How-toBuildersOfficial04:20
Chinese contentAmazon Bedrock Managed Knowledge Base 推出单次检索与 Agentic 检索模式

AWS 给 Bedrock 知识库加了个很实用的开关:简单问题走单次 RAG,复杂问题才让智能体多轮检索,用 maxAgentIteration 控制。做企业知识库问答的工程师可以照着文中的银行政策助手示例改,能明显省延迟和调用成本。

Primary sourceAWS Machine Learning Blog@Siddhant SahuSource
How-toBuildersOfficialSelected04:20
OpenAI dot workflows for testing iOS apps across Macs

The subtle point is that dot's value here comes from owning a second machine, not raw model quality — delegating tests to separate hardware avoids the cursor and focus contention that plagues single-machine agent use. iOS developers juggling multiple SDK versions and QA leads comparing regressions should try the cross-machine repro workflow first. Compared with manual device farm setups, this needs only a spare Mac and an OpenAI account.

Event
Primary source@OpenAIDevs@OpenAIDevsSource
How-toBuildersMulti-source81°03:25
Chinese contentAnthropic 发布 Claude Fable 5.2 与 Mythos 5.2,附高效使用技巧

Anthropic 刚发了 Fable 5.2 和 Mythos 5.2 两个模型,作者还附了实操建议:让它先挑毛病再干活、日常用中档 effort 省算力、整项目托管跑几小时。做长任务的工程师可以直接照抄这套用法。

Event
Alex Albert@alexalbert__Source
Tomorrow 08:00, the daily lands in your inboxTelegram channel or email. Same screening as this page, no ads.
How-toBuildersMulti-source00:36
Chinese content教程:5 步在 Claude 中用 Plaid 关联银行账户做个人理财

Anthropic 给 Claude 加了个人理财功能,靠 Plaid 连银行账户后能自动汇总消费、揪出隐藏手续费、预测下期账单。这篇教程拆成 5 步照做就能跑通,还讲清了数据权限和怎么解绑。想管钱又懒得记账的人可以直接试试。

Event
Techsauce@ Techsauce Team Source
Chinese contentถอดสูตร ttb สร้างดิจิทัลแบงก์กิ้งแบบ Humanized Digital Banking

ถ้าอยากรู้ว่าธนาคารไทยอย่าง ttb ออกแบบแอปที่ลูกค้าใช้แล้วรู้สึกว่ารู้ใจทำยังไง บทความนี้เล่าตรงจุด ตั้งแต่หา pain point ของลูกค้าก่อน แล้วค่อยเอา Digital, Human และ AI มาผสมกัน ไม่ใช่เอาเทคโนโลยีนำ ทีมสินค้าดิจิทัลหรือคนทำแบงก์กิ้งในไทยและอาเซียนน่าจับไปอ่านเทียบกับวิธีที่ทีมตัวเองทำอยู่

Techsauce@ Techsauce Team Source
Chinese contentGoogle Gemini 产品负责人 Tulsee Doshi 谈 AI 时代产品经理的七个心得

Gemini 产品负责人自己的招人和做产品方法,全是实操细节:60-20-20 路线图怎么分、第一个 eval 用 5 条提示词就能开跑。做 AI 产品或想转 AI PM 的工程师直接照着用,比泛泛谈趋势的访谈实在得多。

Lenny Rachitsky@lennysanSource
How-toBuildersSelected00:01
Databricks cuts AI coding costs by up to 90%

The non-obvious lesson is that cost control came from routing and token overhead reduction, not from rationing access to expensive models. Engineering leaders and platform teams running AI coding tools at scale can apply the same four levers to their own budgets. The 90% figure offers a concrete benchmark to measure internal optimisation efforts against.

Official accountDatabricks@databricksSource
Nov 9
Chinese contentdair.ai 倡导企业自建模型评测体系:公共基准之外,生产环境需要 own evals

dair.ai 的 Elvis 在说:公共基准看看就行,模型能不能进生产得靠自己的 evals。他还把 evals 和 RSI 挂钩,说这是企业优化的地基,要自建不要外包。做模型选型的工程师和企业团队值得看看这条思路。

Event
elvis@omarsar0Source
How-toBuildersSelected21:26
Emil Kowalski releases /gestures skill for AI-built interactions

The skill encodes interaction-quality knowledge that code generators typically miss, so AI-built apps can feel closer to hand-polished native apps. Front-end developers working with Claude, Cursor or similar coding assistants can install it and apply the patterns directly to bottom sheets, drag handlers and other touch UIs. It complements Kowalski's earlier interaction skills, among the most widely installed skill sets for AI pair-programming.

Viking@vikingmuteSource
How-toBuildersMulti-sourceSelected21:25
Viking Shares AI Design Workflow for Landing Pages

The workflow shows why design refinement, not code generation, is now the bottleneck for AI-built products — even with a strong model, unconstrained output still looks generic. Builders shipping landing pages can copy the three-step pipeline directly: reference collection via Refero MCP, batch style generation with a design-token file, then small-scale animation polish. The Refero MCP integration makes the setup practical for developers without a design background.

Event
Viking@vikingmuteSource
How-toBuildersMulti-sourceSelected19:26
LangChain ToolRuntime Guide: Using Context, State and Store

The key security point is that user identity must come from an authenticated login system and be injected via context, never read from model output — a pattern that prevents prompt-injected privilege escalation in agent tools. Developers building LangChain agents in production can adopt this to separate runtime credentials from model-visible parameters and to decide what belongs in state versus store.

Event
掘金本周最热@Dragon_xjySource
How-toBuildersMulti-sourceSelected17:27
Garry Tan open-sources g-harness-evals for agent harness comparison

The result suggests harness design barely moves accuracy — the model does — but heavily moves cost and latency, so choosing a harness is mostly an efficiency decision. Practitioners building agent workflows can use the repo's Slack-based simulation and state-based scoring to replicate the setup on their own tasks. The author cautions the numbers come from a dev set approaching saturation, with some OpenClaw runs incomplete.

Event
shao__meng@shao__mengSource
How-toBuildersSelected15:23
Addy Osmani shares best practices for managing coding agents

The practical shift here is that supervising agents demands the same skills as managing engineers — scoping, review and verification — rather than writing code yourself. Developers running long agent sessions can apply the agents.md persistence trick and automated verification loops immediately. Anyone frustrated by agents repeating fixed errors will find the lowest-friction fix in the third practice.

Event
Viking@vikingmuteSource

Only the latest 2000 items are listed; for older ones see the daily archive