本文来源:https://mp.weixin.qq.com/s/6v8nd6gE3ZZyq4mJ5R970w
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。
目录
1.Agent适合包揽的公众号文章类型
2.Workbuddy写文发布的整套流程
3.小结
写新媒体文案、科普文、小说、工作总结.......想必大家都用过大模型。从刚开始的deepseek、豆包,到后来公认写作出众的Gemini。所以,当我现在提出WorkBuddy这类AI Agent时,大家也会想问,用AI Agent写公众号文,相比用一般的LLM模型写,有什么优势吗?AI Agent的优点很多。从执行逻辑到生成结果都能令用户质效双收:
AI Agent产品也很多啊,为何选中了WorkBuddy呢?
主要还是由其技能库、运行效率、价格和个人需求共同决定。
所以你用适合你自己的其他Agent,也是可以的!
本来下面该直达主题,但为了让Agent扬长避短,让大家使用更得心应手,先给大家看看Agent在写文方面的能耐。
Agent适合包揽的公众号文章类型
不是所有的文章都能被Agent hold住。
像以个人经历、个人情感情绪为重点的故事,或者浓浓创意味儿的文章,Agent就驾驭不了。
它就适合撰写结构规范、依赖信息归纳与模板化的文章。
据研报统计,微信公众号中,知识科普类和社会文化类的内容更受欢迎,比如干货、资讯等。
所以两者结合,Agent适合写的微信公众号文章大概形成以下矩阵:
Workbuddy写文发布的整套流程
尽管Agent有公众号文章处理的全流程技能,但整套流程大家还是要心里有数,能提前准备最好,避免消耗不必要的tokens。
Agent、LLM与其Skills的关系,就好比全自动的智能汽车、负责看路做决定的虚拟司机与车配置的关系。
有限预算内性能达到最优,主要还是靠Skills发力了,毕竟顶尖LLM很多都要自己接API。
此处奉上技能安装的技巧:
workbuddy公众号写作相关的技能很多,全流程的,不同文风的,去AI去味儿,大家选择适合自己的即可:
说一下总原则,就是安装技能之后的每一步都要自己审核无误后再进行下一步,当然一般来说agent会主动让你先过目。
接下来就是选题。选题操作更简单一些:
这里我选的是上市公司财报分析的案例。
写作环节,要求的分析含量越高,输入的指令就越是细致,人工修改部分可能就越多。
指令案例:
你是一名财经分析大师,现在需要你写一篇以《金价狂飙,百年老凤祥的营收、净利缘何大幅双降?》为题目、以老凤祥2025年财报数据为指引、以财务逻辑为基础的上市公司的公众号分析文章,文章的具体要求如下:
1.文章结构如下,写的时候按照这个框架写:
(1)老凤祥营收净利双降与金价走高的矛盾背景
(2)几个原因分析
(3)总结
2.一个点不要只写一段话,一段话最多3句话。
3.字数:1500-2000字之间
4.文章风格:@wechat-article-pro ,用刘润风格
5.面对群体:大众
插图环节注意选择好用的生图模型。
排版若你没有公文一般的高要求,就会比较省心。
贴图和图文都能推送到公众号后台形成草稿,但图文推送后可能格式会变。
我的老凤祥财报分析的公众号文章,最终就是这个样子:
小结
从 0 到发布,流程Agent全包了,门槛它拆了。但有意思的是,门槛越低,脑子里有没有东西就越藏不住。
所以,你觉得内容的护城河该是什么呢?欢迎在评论区留言参与讨论哦~
Source: https://mp.weixin.qq.com/s/6v8nd6gE3ZZyq4mJ5R970w. Contact: 13951431913#139.com for copyright concerns.
Table of Contents
1. Types of Articles Suitable for Agent Handling
2. Complete Workflow of Workbuddy Writing and Publishing
3. Conclusion
Write new media copy, popular science articles, novels, work summaries... I'm sure everyone has used a large model before. Starting with deepseek and Doubao, then moving on to the universally acclaimed Gemini for writing. So, when I now introduce AI Agents like WorkBuddy, everyone will likely wonder: what's the advantage of using an AI Agent to write a WeChat post compared to using a regular LLM model? The benefits of AI Agents are numerous. From execution logic to the end result, users can experience both quality and efficiency gains: There are many AI Agent products out there, so why choose WorkBuddy?
There are many AI Agent products available. Why did I choose WorkBuddy?
You are welcome to use other Agents that fit your requirements!
Before diving into the details, let's examine what types of articles are best suited for Agents to ensure optimal usage.
Agents are not suitable for all types of articles.
They struggle with content that is heavily reliant on personal emotions, experiences, or high levels of creativity.
Agents excel at writing structured, data-driven articles that require information aggregation and follow specific templates.
According to research reports, knowledge-based and social/cultural content are popular on WeChat Public Accounts. Therefore, the ideal articles for Agents to write would be:
Although Agents have a full-lifecycle skill set for handling WeChat articles, it is important to understand the workflow to avoid unnecessary token consumption.
Tip for Skill Installation:
Workbuddy offers a variety of skills for WeChat writing, including full-process management, style selection, and debiasing. Users should select the skills that best meet their needs.
The key principle is to review the output of each step to ensure accuracy before proceeding.
For this example, I selected a case study on analyzing a public company's financial report.
Example Prompt:
Article Structure: 34. Background: The contradiction between rising gold prices and decreasing revenue/profit. 35. Analysis: Reasons for the decline. 36. Conclusion.
You are a maestro of financial analysis. Now, you must write a public account analysis article for a listed company titled "Gold Prices Soar, but Old Phoenix's Revenue and Net Profit Plummet — Why?" This article will be guided by the financial data from Old Phoenix's 2025 report and grounded in rigorous financial logic.
1. The article structure should follow this framework:
(1) The contradictory background of Old Phoenix's revenue and net profit decline despite rising gold prices.
Constraints: 38. 3-4 sentences per paragraph. 39. Word count: 1500-2000. 40. Style: Similar to Liu Run's style. 41. Audience: General public.
(3)总结
2. Each point should be covered in a single paragraph, with no more than three sentences per paragraph.
3.字数:1500-2000字之间
4. Writing style: @wechat-article-pro, in the style of Liu Run.
For image selection, it is crucial to choose a reliable image generation model to avoid visual artifacts.
If you have specific formatting requirements, the layout step might require manual adjustment. However, for standard needs, the system usually handles it well.
Conclusion
In summary, AI Agents have significantly lowered the barrier to entry for creating structured content. They handle the entire workflow from conception to publication, allowing users to focus on the quality of the content rather than the mechanics of production. The "moat" now lies in crafting insightful and engaging narratives that leverage the data provided, rather than simply generating reports.
From zero to publication, the process agent handles everything, breaking down barriers. Interestingly, the lower the barrier, the more what's in your head becomes apparent.
So, what do you think the moat for content creation should be? Feel free to share your thoughts in the comments section!











