The Era of No-Code AI: What You Need to Know

查看原文 / View Original 编辑 / Edit

The Era of No-Code AI: What You Need to Know

无代码 AI 时代：你需要了解的一切

Artificial Intelligence 人工智能

The Era of No-Code AI: What You Need to Know 无代码 AI 时代：你需要了解的一切

If you are a programmer and you don’t feel “special” anymore, you are not alone. 如果你是一名程序员，并且感觉自己不再那么“特别”了，你并不孤单。

Intro 引言

Coders grew up in a world where knowing a programming language was considered “hot”. If you knew how to program, you had an advantage: you could build software and automate tasks, while others depended on you. Now, the world has changed, and everyone can create AI without a single line of code. 程序员们成长在一个掌握编程语言被视为“热门”的世界里。如果你会编程，你就拥有优势：你可以构建软件并自动化任务，而其他人则依赖于你。现在，世界已经改变，每个人都可以在不写一行代码的情况下创建 AI。

AI has evolved far beyond chatbots. Just a few years ago, learning how to use ChatGPT effectively was enough to stand out. In 2025, building local Agents still largely meant writing Python code, with developers turning to tools like LangChain to run open-source models directly on their own computers. AI 的发展早已超越了聊天机器人。就在几年前，学会如何有效地使用 ChatGPT 就足以脱颖而出。在 2025 年，构建本地智能体（Agents）在很大程度上仍意味着编写 Python 代码，开发者们转向使用 LangChain 等工具，在自己的电脑上直接运行开源模型。

However, since the beginning of 2026, the AI landscape has accelerated dramatically. We have now entered the era of no-code AI, where anyone (without a technical background) can quickly create, deploy, and manage multiple custom Agents. But fear not. In this article, I shall break down what skill set you need in order to gain an advantage in this new era as well (so you can feel “special” again). 然而，自 2026 年初以来，AI 领域的发展显著加速。我们现在已经进入了无代码 AI 时代，任何（没有技术背景的）人都可以快速创建、部署和管理多个自定义智能体。但不必担心。在本文中，我将剖析你需要掌握哪些技能，以便在这个新时代获得优势（这样你就能再次感到“特别”了）。

Prompting 提示词工程 (Prompting)

Every interaction with an AI model starts with a prompt. The difference between average users and advanced users is not the model itself. As much as it pains me to say it, writing good prompts is the new coding. If you want to use AI products, then you need to know the industry standard for prompting. 与 AI 模型的每一次交互都始于提示词。普通用户与高级用户之间的区别不在于模型本身。尽管我很不情愿这么说，但编写高质量的提示词就是新的编程。如果你想使用 AI 产品，那么你需要了解行业标准的提示词技巧。

Over the years, we have seen many prompting techniques, like Zero-Shot, ReAct, Chain-of-Thoughts… Today, there are two main prompting frameworks: 多年来，我们见证了许多提示词技术，如零样本（Zero-Shot）、ReAct、思维链（Chain-of-Thoughts）等。如今，主要有两种提示词框架：

  1. TCRF (the most frequently used):
  2. TCRF（最常用的框架）：
  • Task (T) – The explicit actionable instruction (i.e. “write an email to an applicant”).
  • 任务 (Task, T) – 明确的可执行指令（例如：“给申请人写一封邮件”）。
  • Context (C) – Background information and constraints (i.e. “after 2 weeks of CV screening, you found a young talent. Don’t be too formal but keep it professional”).
  • 背景 (Context, C) – 背景信息和约束条件（例如：“经过两周的简历筛选，你发现了一位年轻人才。不要太正式，但要保持专业”）。
  • Role (R) – The persona the AI should adopt (i.e. “you are an experienced HR manager”).
  • 角色 (Role, R) – AI 应采用的人设（例如：“你是一位经验丰富的人力资源经理”）。
  • Format (F) – The desired output structure (i.e. “the email must have three paragraphs, use the following example…”).
  • 格式 (Format, F) – 期望的输出结构（例如：“邮件必须包含三段，使用以下示例……”）。
  1. TCREI (introduced by Google as an iterative and advanced extension of TCRF):
  2. TCREI（由 Google 引入，作为 TCRF 的迭代和高级扩展）：
  • Task (T) and Context (C) are the same as before.
  • 任务 (T) 和背景 (C) 与之前相同。
  • References (R) – Role + Format (i.e. “you are an experienced HR manager. The email must have three paragraphs, use the following example…”).
  • 参考 (References, R) – 角色 + 格式（例如：“你是一位经验丰富的人力资源经理。邮件必须包含三段，使用以下示例……”）。
  • Evaluate (E) – This is the addition: ask the AI to critically assess its own output based on specific criteria (i.e. “after writing the email, evaluate it on a scale of 1-10 for: Clarity, Engagement, Persuasiveness, and Alignment. Point out specific weaknesses”).
  • 评估 (Evaluate, E) – 这是新增部分：要求 AI 根据特定标准批判性地评估自己的输出（例如：“写完邮件后，请从清晰度、参与度、说服力和一致性四个维度进行 1-10 分的评估，并指出具体弱点”）。
  • Iterate (I) – Instruct the AI to improve the output based on the evaluation (i.e. “then rewrite an improved version”).
  • 迭代 (Iterate, I) – 指导 AI 根据评估结果改进输出（例如：“然后重写一个改进版本”）。

Products 产品

There are too many AI products. There is no official registry, but industry analysts estimate that thousands of new AI tools, wrappers, and applications are created every week. The total number of active AI platforms in the ecosystem is estimated to be around 90,000. AI 产品实在太多了。虽然没有官方注册机构，但行业分析师估计，每周都有数以千计的新 AI 工具、封装器和应用程序被创建。生态系统中活跃的 AI 平台总数估计在 90,000 个左右。

As of today, the market is still dominated by the “Big 4” cloud-based general-purpose Agents: OpenAI’s ChatGPT, Google’s Gemini, Anthropic’s Claude, X’s Grok. Then, there are specialized products for specific domains, like Perplexity for studying and research, and Cursor or GitHub–Copilot for coding (in fact, a growing trend is “Agentic Engineering” which is AI coding new software). A good cloud-based alternative to play, host, and share AI projects for free is HuggingFace-Spaces. 截至今日，市场仍由“四大”云端通用智能体主导：OpenAI 的 ChatGPT、Google 的 Gemini、Anthropic 的 Claude 和 X 的 Grok。此外，还有针对特定领域的专业产品，如用于学习和研究的 Perplexity，以及用于编程的 Cursor 或 GitHub Copilot（事实上，一个日益增长的趋势是“智能体工程”，即由 AI 编写新软件）。一个免费游玩、托管和分享 AI 项目的优秀云端替代方案是 HuggingFace-Spaces。

However, it appears that the market is recently shifting toward local models to ensure data privacy, eliminate recurring API costs, reduce cloud latency, and maintain control over proprietary workflows. We are talking about standalone closed-source products (i.e. Claude-Cowork and Claude-Code), and open-source solutions (i.e. OpenClaw and Hermes) that you have to pair with an LLM management app (i.e. Ollama). Please note that to run useful things locally, you need at least a machine with 16 GB of RAM and an 8 GB GPU (or a total of 24 GB of unified memory pool). 然而，市场似乎正转向本地模型，以确保数据隐私、消除持续的 API 成本、降低云端延迟并保持对专有工作流的控制。我们指的是独立的闭源产品（如 Claude-Cowork 和 Claude-Code），以及必须与 LLM 管理应用（如 Ollama）配合使用的开源解决方案（如 OpenClaw 和 Hermes）。请注意，要在本地运行有用的程序，你至少需要一台配备 16GB 内存和 8GB 显存（或总计 24GB 统一内存池）的机器。

At the moment, Claude is the smartest AI out there, so it’s important to understand the difference between the products in the Anthropic family: 目前，Claude 是市面上最聪明的 AI，因此了解 Anthropic 家族产品之间的区别非常重要：

  • Claude (web app) is the usual cloud-based agentic chatbot, no different than ChatGPT, Gemini, or Grok. This is for the average user.
  • Claude（网页版） 是常见的云端智能聊天机器人，与 ChatGPT、Gemini 或 Grok 没有区别。这是为普通用户准备的。
  • Claude-Cowork (desktop app) is for smart but non-technical users. It runs in a sandboxed environment on your PC with selected access to your folders. Ideal for workflow automation.
  • Claude-Cowork（桌面版） 是为聪明但非技术背景的用户准备的。它在你的电脑沙盒环境中运行，并对你的文件夹拥有选择性访问权限。非常适合工作流自动化。
  • Claude-Code (terminal app) is for developers. It has full access to your terminal, so it can execute code. Useful for building apps.
  • Claude-Code（终端版） 是为开发者准备的。它拥有对你终端的完全访问权限，因此可以执行代码。适用于构建应用程序。

Workflows and Apps 工作流与应用

We have moved from reactive AI to proactive AI. Before, it was you texting your chatbot asking questions. Now, the Agent pings you to tell you that the work you delegated to it is done. With the right setup, you won’t have to do anything (besides reviewing and validating the output). The AI autonomously researches, plans, executes, and deploys results. 我们已经从被动式 AI 转向了主动式 AI。以前，是你给聊天机器人发消息提问。现在，智能体会主动提醒你，告知你委托给它的工作已经完成。通过正确的设置，你无需做任何事（除了审查和验证输出结果）。AI 会自主进行研究、规划、执行并部署结果。

Local AI Agents unlock an entirely different way of working, and therefore, a new way of living. To put it in another way, in this new era, everything that doesn’t require a physical action can be automated with AI. So that’s what you should do… learn how to automate your life: 本地 AI 智能体开启了一种完全不同的工作方式，进而带来了一种新的生活方式。换句话说，在这个新时代，任何不需要物理动作的事情都可以通过 AI 自动化。所以这就是你应该做的……学习如何自动化你的生活：

  • all your daily tasks follow a workflow that can be automated by giving instructions (i.e. “research this topic, put it in Excel, and send it via email“)
  • 你所有的日常任务都遵循一个工作流，可以通过下达指令来自动化（例如：“研究这个主题，放入 Excel，并通过电子邮件发送”）。
  • all your ideas can be built in the form of an app by providing a goal (i.e. “I want a mobile dashboard for my investments“)
  • 你所有的想法都可以通过提供一个目标，以应用程序的形式构建出来（例如：“我想要一个用于投资的移动端仪表盘”）。

During your work, you most certainly need to connect your Agents to real-world tools, systems, and data. The best way to do that is through MCP Servers. MCP (Model Context Protocol) is an open-source standard framework introduced by Anthropic that enables AI systems to communicate with ex… 在工作中，你肯定需要将智能体连接到现实世界的工具、系统和数据中。实现这一点的最佳方式是通过 MCP 服务器。MCP（模型上下文协议）是由 Anthropic 引入的一种开源标准框架，它使 AI 系统能够与外部……（原文截断）

← 返回新闻列表 / Back to news list