A Note on How today's Blog/ Moment Was Created: Human Reading Meets AI in Action
This blog is also part of my experiment with the ideas presented in the book.
I did not simply ask AI to summarize the book for me.
I first read the book myself, became familiar with the author's way of thinking and writing, and prepared my own reading notes as I progressed through the chapters.
Then I brought those notes into two different AI environments:
- Google Gemini Notebook
- OpenAI ChatGPT
I gave both tools the same essential material and asked them to work with my notes.
Why?
Because for me, the purpose of this exercise is no longer simply to read a book about AI Agents.
I want to see the tools themselves in action.
The book argues that the future of working with AI is not about repeatedly opening a chat box, asking a question, receiving an answer, and starting again from zero. It is about creating a process in which humans and AI each contribute what they are best at.
So I decided to apply that idea to the reading and writing process itself.
My role: read, understand, question, select, organize, and decide what is worth sharing.
The AI tools' role: process the material, help structure it, identify connections, organize information, and transform my raw notes into different possible forms.
This means the article you are reading is not simply a book summary, and it is not simply AI-generated content.
It is an AI–human collaboration experiment.
I am using the book's own philosophy as a lens through which to test the tools.
And I am deliberately using two different AI environments rather than relying on only one, because I want to see where they produce similar results, where they differ, and where the human still has to make the important decisions.
The Experiment
The process is intentionally simple:
For the moment, I am not trying to prove which tool is "better."
I'm more interested in a different question:
What happens when we stop treating AI as a machine that gives us answers and start treating it as part of a workflow?
That is, after all, one of the central questions raised by the book.
So while I am reading a book about building AI Agent workflows, I am also experimenting with my own workflow for reading, understanding, comparing, and creating.
The book is teaching me about Agents.
At the same time, the tools are giving me a chance to observe the methodology in action.
And this experiment is still ongoing.
AI Agents Are Actually Simple: A Zero-Base Introduction to Building AI Agent Workflows
A structured reading guide, chapter-by-chapter notes, and a comparison with the Jotform Business Development AI Agent.
About This Book
What is this book about?
This book teaches you how to upgrade AI from a disposable, temporary helper into a reusable and iterative AI Agent workflow.
Instead of simply chatting with AI, you learn to build automated systems that can run on their own.
The book is structured into four parts:
- Part 1 — Chapters 1–3: Concepts — AI, Agent, Workflow, and the mindset shift from "handicraft" to "factory" thinking.
- Part 2 — Chapters 4–6: Building methodology — the Five-Step Workflow Method, Agent job descriptions, and iterative trial runs.
- Part 3 — Chapters 7–11: Practical scenarios — short video, e-commerce, education, knowledge work, and entrepreneurship/management.
- Part 4 — Chapters 12–13: Moving from scattered processes toward a unified Personal Intelligent Workflow System.
Who Is the Author?
Long Xiaokang (龙晓康)
- Previously worked at Baidu and Qunar.
- Spent eight years as an entrepreneur.
- Founded the education company Shunzhong AI, which has served more than 100,000 students.
- Created AIGC Chengye, a technology blog with more than one million followers across the internet.
- Founded Yipai AI, a short-video marketing Agent product.
Why Did the Author Write This Book?
The author has also explained the reason behind the book through his established WeChat account. The following section presents his explanation in his own words.
"What ordinary people lack is not AI tools, but an Agent workflow."
1. What Ordinary People Lack Is Not AI Tools, but an Agent Workflow
Over the past few years, I have worked on both AI training and AI content, and now I am also involved in developing AI products. In theory, the people I encountered should be among the most enthusiastic proponents of AI.
This is indeed the case. Many people share AI-related information daily, save AI tools, purchase AI courses, study prompts, and immediately try new models and tools as soon as they become available.
But what makes me increasingly uneasy is this: most people appear to be making a concerted effort to learn about AI, but their work methods have hardly changed at all.
They will use AI to write a piece of copy, have AI revise a resume, have AI summarize an article, and have AI generate an event plan.
But each time, a dialog box is temporarily opened, a command is temporarily typed, and a result is temporarily waited for.
Unsatisfied with the results, they continue to ask questions. If the information is not accurate, they check it themselves. If the format is not appropriate, they modify it themselves.
In the end, AI has indeed been involved, but it doesn't seem to have made things much easier personally.
Later, I realized that this is the biggest challenge many ordinary people face when it comes to using AI today: we believe we lack more sophisticated models, better tools, and more advanced prompts. But what many people really lack is a workflow.
I am increasingly convinced that the use of AI is overestimated.
Over the past few years, many people have asked me: "How should an ordinary person actually learn about AI?"
Initially, I would also recommend tools, explain prompts, and teach people how to get AI to write better, draw more accurately, and summarize more clearly.
These were certainly useful, but the more people I came into contact with, the more I found a problem: asking AI-related questions will not address the real issue of efficiency.
Because work is not a problem; work is a series of processes.
For example, making a short video is not simply asking, "Help me write a hit script."
At a minimum, it involves:
- Selecting the topic
- Setting the angle
- Creating the title
- Writing the script
- Finding material
- Designing the shots
- Preparing the dubbing
- Editing
- Creating the cover
- Publishing
- Reviewing the results
For example, doing an e-commerce startup is not simply saying, "Help me write a product copy."
Behind it are:
- Product information organization
- Selling-point refinement
- Main image requirements
- Detail image structure
- Headline keywords
- Marketing short sentences
- Platform listing
- User review analysis
- Follow-up review
An entrepreneur making a decision, for example, isn't simply asking, "Do you think I should or shouldn't do this business?"
Behind this are:
- Information gathering
- Competition analysis
- Scenario generation
- Risk derivation
- Resource evaluation
- Stop-loss conditions
- Team consensus
That's why many people find that using AI doesn't yield results. It's not that the AI is unable to respond; it's simply that you have condensed a complex process into a vague question.
What you provided to the AI was a jumbled mess, and of course, the AI returned a text that appears complete but is actually difficult to implement.
Over the past few years, I have become increasingly convinced that what truly sets apart individuals in the age of AI is not whether they know how to use AI, but whether they can break down their work into processes that AI can execute.
2. After Creating AI Products, I Developed a Deeper Appreciation for This Field
When I was involved in AI training in the past, I witnessed the difficulties faced by ordinary people in learning about AI.
Later, I worked on AI products, specifically in this area with Yipai AI, and I felt even more deeply about products that really deliver results to users.
Many users are not unwilling to use AI or unable to operate the buttons. The places where they really get stuck are often not "where the tool entrance is," but:
- What kind of data should I provide to the AI?
- What should the outcome look like?
- What steps can't be skipped in the middle?
- Where do I have to make a judgment?
- Where can I confidently entrust tasks to AI?
This realization made me understand that the biggest challenge ordinary people face when dealing with AI is not "not knowing how to use a particular feature," but rather "lacking a structured approach to work."
If there is no structure, the more powerful an AI becomes, the more likely it is to cause anxiety.
Because it answers everything, you don't know where to begin. It can generate anything, and you don't know what results qualify. It seems to know everything, and it's easier for you to hand over judgment that you shouldn't have given to it.
So when I wrote this book, I always had a very clear judgment in mind:
What ordinary people truly need is not another comprehensive guide to AI tools, but an introductory book that teaches them how to set up an AI workflow.
That's where the book, "The Intelligent Person is Actually Simple," started.
3. Why Didn't I Compile It into a "List of AI Tools"?
To be honest, if all you want to do is ride the wave of popularity, you could write this book as "100 Essential AI Tools," "50 Popular Prompt Phrases," or "Getting Started in 10 Minutes with AI Agents." It might be easier to look busy.
But I don't really want to write that, for a simple reason: the tools will become obsolete, the models will become obsolete, and the platforms will become obsolete.
Today, someone might find an AI product extremely useful, but within half a year, it could be replaced by another product. The operating tutorials for a platform today may be very new, while the position of a button may change in a few months.
If a book only writes about tools and operations, its life cycle will be short. But some things don't become obsolete that quickly.
For example:
- How do you break a vague task into clear steps?
- How do you define inputs and outputs for each step?
- How do you decide which things should be done by people, which can be handed over to Agents, and which are better handled by custom scripts?
- How would you write a clear job description for an AI digital employee?
- How do you distill a running process into an asset that you can use over and over again?
These are the fundamental skills that ordinary people truly need to possess in the age of AI.
So while this book appears to be about AI Agents, what I really want to talk about is:
How can ordinary individuals transition from simply "using AI tools" to "designing the way AI operates"?
That's why I talk about flowcharts, workflows, job descriptions, test runs, checkpoints, and iterations over and over again in my book.
Because I don't want the reader to just know a few concepts after reading a book.
I would like you to immediately take one of the most annoying, repetitive, and time-consuming tasks at hand and break it into a process that can actually run.
The "Intelligent Person" Is a Digital Employee
Nowadays, many people view AI Agents as quite mysterious when they come across the term.
It seems like a super AI that can do anything. As soon as you utter a single sentence, it can independently comprehend the objective, break down the task, summon the necessary tools, and complete everything on its own.
But I want to make this a little plainer. In my opinion, the simplest way for the average person to understand intelligent Agents is:
Think of it as a digital employee.
What you truly need to learn is not "invoking an all-purpose AI," but rather mastering the art of managing AI, just as you would manage a real team:
- Give it a clear job.
- Give it clear materials.
- Give it a fixed process.
- Give it output standards.
- Give it limitations.
- Give it trial and error and iterations.
It is also impossible for a real employee to fully understand your standards on the first day of employment. You need to train them, correct them, give them feedback, and help them become familiar with the way you work.
An Agent is the same. An easy-to-use Agent is not simply written once; it is improved again and again.
This is also why I place such emphasis on "job descriptions" in my book. Many people struggle with AI because they fail to clearly define their tasks, not because the AI is not intelligent enough.
You tell it to "write a little professionally," and it starts piling up terminology. You tell it to "analyze," and it outputs a bunch of correct nonsense.
But if you tell it, like a recruiter:
- Who you are
- What your purpose is
- What materials it can use
- What format it must output
- What cannot be said
- What must be done when information is missing
Its stability will be completely different.
So I'd rather readers view this book as an "AI Digital Employee Management Manual."
5. This Book Is Not for Technical Experts
The subtitle of this book is "Building AI Agent Workflows from Scratch."
Because this book wasn't written for technical experts. It is not a book about the principles of large models, nor is it a book on multi-agent architecture. It is not written for developers who already know how to program and build complex systems.
It's more suitable for people who are genuinely consumed by specific work every day.
If you're a content creator, you're choosing topics, writing scripts, making covers, posting videos, and looking at data every day. What you're missing may not be inspiration, but a steady content-production pipeline.
If you are an e-commerce practitioner, you have to organize product materials, make main images, write selling points, analyze reviews, and complete new listings every day. What you may be lacking is not just a more conversational AI, but a standardized process that covers everything from product images to the launch of new items.
If you're a teacher, you're preparing lessons, asking questions, making corrections, and writing feedback every day. What you need is not for AI to replace you as a teacher, but for AI to assist you in generating structural drafts, topic outlines, explanations, and initial feedback drafts, freeing your time to focus on areas that truly require judgment and care.
If you are a working professional, you have to hold meetings, write weekly reports, and communicate across departments every day. What you need isn't for AI to write a few catchy phrases; rather, it's to establish a consistent format for meeting minutes, action items, task lists, weekly report templates, and communication structures, making your work clearer, more manageable, and easier to hand over.
If you're an entrepreneur or manager, you're gathering information, developing a plan, evaluating risks, and making decisions every day. What you need is not for AI to make decisions on your behalf, but for AI to help broaden your options, provide additional information, and analyze risks, thereby making your decisions more replicable.
That's why I wrote about five scenarios in my book: short video, e-commerce, education, the workplace, and entrepreneurial management.
They look different across industries, but the underlying logic is the same:
Steps are broken down, inputs and outputs are defined, executors are reassigned, and the process is condensed into a reusable asset.
Once you master this methodology, you can still migrate the underlying logic by changing an industry, a tool, or a model.
6. What I Really Want to Promote Is Not Just a Book, but a Way of Working
I wrote this book not to tell people how powerful AI is. This phrase has been said too much over the past few years.
What I want to emphasize more is this: Please stop viewing AI as temporary workers.
The characteristic of temporary workers is that they leave when they have done their work and have to be reassigned the next time.
The truly valuable ways in which AI will be used in the future will certainly not involve repeated, ad-hoc calls. Instead, integrate AI into your workflow so that it becomes an integral part of your system.
You can have:
- A fixed topic-selection process
- A defined writing process
- A defined review process
- An established preparation process
- A maintained meeting process
- A defined decision-making process
In every process, there are parts suitable for people to do and parts suitable for Agents to do.
People are responsible for goals, judgments, aesthetics, trade-offs, and responsibilities. Agents are responsible for searching, organizing, summarizing, drafting, formatting, and repeating.
This is how I understand AI collaboration.
It's not that humans will be replaced by AI, nor that people will continue to perform all tasks personally. Instead, people are upgraded from executors to system designers.
This is a very important identity change for the average person. Previously, you wanted to have a team that needed to be recruited, managed, and paid. Now, you can at least start with a group of AI digital employees.
They won't be perfect naturally, but they can be trained. They don't assume ultimate responsibility for you, but they can do a lot of repetitive work. They won't automatically understand all your standards, but you can use processes, job descriptions, and continuous iterations to make them more and more relevant to how you work.
When you really build a workflow that runs, you'll find that:
AI is no longer just a tool for answering questions; it has begun to become a lever for boosting productivity.
7. Three Changes I Would Like the Reader to Have After Reading This Book
First: Stop Running After Every New Tool
Of course, you should pay attention to tools, but don't be led by them.
The question you really want to ask is not: "What does this tool do?"
It is: "What kind of workflow does it fit into?"
Second: The Prompt Is No Longer the End Point
Prompts are important, but prompts are not the whole story.
Without a process, a prompt is just a one-time instruction.
If there is a process, prompts become part of an Agent job description and a long-term reusable asset.
Third: Start Building Your Personal Intelligent Workflow System
You don't need to build a complex system first. You just need to start with a real task.
Examples include:
- Your weekly report
- Your short-video scripts
- Your product launches
- Your meeting minutes
- Your course preparation
- Your industry briefs
Break it down into steps, define input and output, configure the Agent, run it again, change it again, and then settle it into a reusable process.
Once this is accomplished, you will have your very first AI workflow of your own.
With a first workflow, there will be a second and a third. Slowly, the way you work will change.
Instead of starting from scratch every time, you start to have your own:
- Process library
- Prompt library
- Agent library
- Workflow asset library
This may be the new asset that ordinary people truly need to accumulate in the age of AI.
Written at the End
The book "Intelligent Agents: Building AI Agent Workflows for Beginners" has been officially published and is now available on major e-commerce platforms.
If you're a regular content creator, e-commerce operator, teacher, working professional, entrepreneur, or anyone who is being consumed by repetitive work, I believe this book can be useful to you.
If you also feel like you've employed a lot of AI, but it hasn't actually made things easier for you,
If you also feel that you are being dragged down by a lot of repetitive work every day,
That's the book I wrote for you.
I hope it will help you set up your very first AI Agent workflow.
I also hope that by starting with this book, you won't just learn a few tools, but will truly acquire a new approach to work tailored for the AI era.
The Book's Philosophy: Architecting Custom Factories
The book guides you to become a "chief engineer," building custom workflows from the ground up.
You start with a blank canvas on a platform such as Coze or Dify and assemble your system by defining roles, selecting models, setting up knowledge bases, and connecting tools through prompt engineering and node-based logic.
The core skill the book teaches is workflow design: you draw the entire blueprint yourself.
This is the "intelligent factory" approach, where you own every piece of the system.
The Jotform Approach (AI Agent I have built Earlier, "Leila, Thinking Partner": Using a Pre-Built Machine for a Specific Job
The Jotform Business Development AI Agent is a "ready-to-use" product— closer to what the book calls an Agent Product.
It is a pre-assembled machine designed for a specific set of tasks: simplifying information collection and lead qualification.
You feed it your business context using natural-language instructions, QA pairs, and uploaded PDFs or URLs. It then handles conversations designed to qualify leads and capture information such as company name, job title, industry, and similar data.
Key Differences
Your Role
Book approach: You are the architect and engineer. You design the entire system.
Jotform Agent: You are the operator. You customize a pre-built system.
Starting Point
Book approach: A blank canvas on platforms such as Coze, Dify, or n8n.
Jotform Agent: A pre-designed conversational interface that is already working.
Core Function
Book approach: Open-ended. You can build many types of workflows.
Jotform Agent: Narrower and focused on business-development conversations and lead qualification.
Training and Configuration
Book approach: Advanced configuration involving prompts, RAG pipelines, tools, and node logic.
Jotform Agent: Simplified configuration through natural language, QA examples, and uploaded information.
Learning Curve
Book approach: Steeper, because you develop systems-thinking, logic, and workflow-design skills.
Jotform Agent: Lower, because the underlying infrastructure is already built.
Ownership
Book approach: You design and control the entire workflow.
Jotform Agent: Jotform provides and maintains the underlying infrastructure.
What This Means
The Jotform Agent is not a replacement for the skills taught by the book. It is a different application of similar underlying principles.
Think of it this way:
The book teaches you how to build an entire factory.
Jotform gives you a specialized machine that is already built.
With the book, you learn to design assembly lines, quality-control systems, worker roles, logistics, and automation rules.
With Jotform, you receive a specialized machine for a particular business task, such as collecting and qualifying leads.
The Bottom Line
- Book approach: Build the factory.
- Jotform Agent: Use the specialized machine.
- Book's core value: Learn workflow architecture and systems thinking.
- Jotform's core value: Apply a pre-built AI system to a specific business need.
The book's real value is not learning one particular platform. Its deeper value is learning to see work as a system:
Task → Decomposition → Roles → Tools → Workflow → Automation → Reuse
Once you understand that logic, individual platforms become implementations of the methodology rather than the methodology itself.
To be continued as I read further...
Your Turn: What Do You See in the Two AI Approaches?
This article is also an experiment, and I don't want the experiment to end with my own observations.
I used the same essential reading material and my own reading notes with Google Gemini Notebook and OpenAI ChatGPT. The two outputs were not identical. They organized the material differently, emphasized different aspects, and created different reading experiences.
I am not asking you to decide which tool is "better." Instead, I would like to know what you see when you compare them.
Questions for Readers
- What is the biggest difference you notice between the Gemini Notebook version and the ChatGPT version?
- Which version helps you understand the book more easily, and why?
- Do you prefer the more condensed, article-style approach, or the more structured, chapter-by-chapter reading guide?
- Which approach makes you more curious about actually reading the book?
- Do you think using two AI tools on the same material adds value, or does it create unnecessary duplication?
- Where do you see the strongest evidence of human judgment in this experiment?
- After reading the author's philosophy, do you see AI differently? Is AI still primarily a "tool" for you, or are you beginning to see it as part of a workflow?
- Which idea from the book could be most useful in your own work: task decomposition, Agent roles, flowcharts, workflows, checkpoints, or reusable processes?
- If you could automate one repetitive task in your own work today, what would it be?
- And finally: Do you think this book could actually change the way you work with AI, or is it simply another AI book?
I'm especially interested in your answer to this question:
"After seeing both versions, what do you think AI is helping me do here that I could not do as easily with only one tool?"
The experiment is still ongoing. As I continue reading and compare AI tools, I will continue adding my observations and testing how the ideas in the book translate into actual AI-assisted work.
ملاحظة للقراء العرب: هذه ليست مجرد مراجعة لكتاب
هذه المقالة ليست مجرد ملخص لكتاب عن وكلاء الذكاء الاصطناعي، وليست أيضاً محتوى تم إنتاجه بالكامل بواسطة الذكاء الاصطناعي.
بدأت أولاً بقراءة الكتاب بنفسي، والتعرف على طريقة تفكير المؤلف وأسلوبه، ثم أعددت ملاحظاتي أثناء القراءة.
بعد ذلك استخدمت نفس المادة الأساسية مع أداتين مختلفتين: Google Gemini Notebook وOpenAI ChatGPT.
الهدف لم يكن معرفة أي أداة "أفضل"، وإنما مشاهدة ما يحدث عندما نضع أدوات الذكاء الاصطناعي داخل سير عمل حقيقي، ونقارن النتائج، ثم نستخدم الحكم البشري لاختيار ما يستحق الاحتفاظ به وتطويره.
وهذا يتوافق بشكل مباشر مع الفكرة الأساسية للكتاب: المستقبل ليس في استخدام الذكاء الاصطناعي كسؤال وجواب فقط، وإنما في تصميم طريقة عمل تجعل الذكاء الاصطناعي جزءاً من سير العمل.
لذلك أسألك أنت أيضاً:
- ما الفرق الذي تراه بين نتائج Gemini Notebook وChatGPT؟
- أي نسخة ساعدتك أكثر على فهم الكتاب؟
- هل ترى أن استخدام أداتين مختلفتين على نفس المادة يضيف قيمة؟
- ما المهمة المتكررة في عملك التي تتمنى تحويلها إلى AI Workflow؟
- وبعد قراءة هذه المقالة، هل ما زلت ترى AI كأداة فقط، أم بدأت تراه كجزء من نظام عمل؟
التجربة مستمرة، وسأضيف نتائج القراءة والتجربة مع استمرار الفصول القادمة.
