In today’s practical article, the author uses the enterprise-level platform GPTBots to encapsulate “market research, brand tonality, product implantation, and conversion path” into an SEO Agent 3.0 that can write “brand words”, bring “own goods”, and directly bring leads.
Follow the steps and you’ll have a B2B SaaS content template that meets search rankings, maintains brand consistency, and drives smoothly.
Have you also found that writing SEO articles with AI always feels almost meaningful?
Previously, I had also used several SEO-related workflows to solve the problem from mining high-converting topics to automatically generating content, which can be regarded as evolving from 1.0 to 2.0:
Use n8n to do AI workflows to drive websites to go overseas to earn US dollars 3: Build an AI Agent for “SEO strategy optimization”
These workflows are fine, but by design: they are all based on network-wide search and competitive analysis, and they are all “external information”.
Although the quality of the article is good, it has no “soul” to read, and even the style of each content is different, which is almost interesting at the “landing” level.
In actual “landing” business scenarios, a good SEO content must solve three core pain points:
- The brand tonality is inconsistent: The content cannot be a “patchwork” stitching monster, it must be in line with our own brand language and values.
- Content is decoupled from the product: The article cannot only talk about theory, it must be cleverly implanted into our own products, showing our solutions.
- Traffic cannot convert: After reading the article, readers should be able to click on the link smoothly, enter our landing page, and even directly generate conversions. This brings SEO closer to money.
To solve these problems, we need a smarter SEO Agent 3.0. It must not only research competitors (align with market demand), but also understand the brand (maintain tonality) and recommend products (generate direct sales).
My previous n8n scheme was very flexible and suitable for exploration. But if you really want to implement it on a large scale in the enterprise, you will encounter headaches such as security, compliance, and permission management.
The GPTBots used today is different, it has been aimed at enterprise-level applications from the beginning. It is a complete LLMOps platform that integrates the “heavy weapons” of knowledge base, database, and workflow for you, so you don’t need to be a “full-stack engineer” anymore.
You can register and use it through the link: https://www.gptbots.ai/signup?utm_source=kol=binggan There is information about this case at the end of the article, which can be reproduced in GPTBots.
To put it simply, the former n8n is the “Geek Exploration Edition”, and the latter GPTBots is the “Enterprise Production Edition”, not a one-dimensional tool.
Today, we will use the GPTBots platform to build an SEO agent that can speak “brand language”, bring “own goods”, and guide conversions step by step.
1. Start with the end: What does our SEO Agent 3.0 do?
Before we get started, let’s use the logic of “starting from the end” to clarify the workflow of this agent.
How can product managers do a good job in B-end digitalization?
All walks of life have taken advantage of the ride-hail of digital transformation and achieved the rapid development of the industry. Since B-end products are products that provide services for enterprises, how should enterprises ride the digital ride?
View details >
When a user enters a keyword (e.g., “sales management”), we expect the agent to complete the following four steps like a seasoned B2B content strategist:
- Invoke the workflow to generate an outline (to ensure market demand): First of all, it has to go to Google to see what the top 3 competitors are writing. Then summarize a title and content outline that is both in line with the mainstream trend and stands out.
- Retrieve the knowledge base and inject the soul of the brand (ensure brand tonality): With the outline, it has to go to our internal “brand knowledge base” to learn, find our “Brand Voice and Writing Guide”, and search for 1-2 of the most relevant “customer success stories” according to the topic of the article.
- Query the database and link your products (to ensure traffic conversion): Which of our products is mentioned in the customer case? Agents need to immediately go to the “Product Database” to query the detailed information of the product, such as function descriptions, landing page links, product screenshot links, etc.
- Integrate information, write professionally (final value delivery)Finally, integrate all the information (outline, brand specifications, customer stories, product information) into a logical, persuasive, and SEO-compliant article.
This process perfectly combines external market demand with internal brand equity, and is also the core framework we built this time.
2. Hands-on: Use GPTBots to build three core components
Once we have a clear goal, we can start “rubbing” in GPTBots. The entire construction process is mainly divided into three parts: workflow, knowledge base, and database.
Wait, some students will ask: “Agents can also be equipped with tools, why do you need to do more workflows?” ”
Good question! The core lies in the uncertainty of combating AI. Let the AI freely call multiple tools in a long conversation, and the result is like opening a blind box, sometimes good and sometimes bad. The workflow encapsulates fixed processes such as “search TOP3 -> summary outline” into a reliable “black box” to ensure that the AI can stably output results in standard format every time it is called.
Simply put: solidify “deterministic” tasks with workflows and control “flexibility” tasks with agents.
1. Workflow: Get the market research and content outline
This is the first and most basic step for our Agent. We need to create a workflow that allows it to Google the TOP ranking URLs based on keywords, and complete the competitor analysis and generate an outline on a visit-by-visit basis.
Here we simplify the process and keep the core part
The complete logic can be referred to I used n8n+Scrapeless to set up an “SEO content engine” to fully automatically mine high-conversion topics and solve the problem of overseas traffic
In GPTBots, the process is straightforward:
Create a workflow: Select “Workflow” in the left menu, click “Create Workflow”, and we give it a name, such as “B2B_SaaS_SEO”.
Configuration process: At the heart of this workflow are two nodes:
GoogleSearch node: This node is responsible for receiving the keyword q entered by the user and going to Google search, we set the num to 3, which means that only the top three results are seen.
LLM nodes: It receives the 3 URLs returned by Google Search in the previous step, and then uses a prompt to let the AI analyze the content of these three pages, and extract a new, better title and content outline, which are returned in JSON format.
Reference prompts(In fact, it can be optimized to make it more complicated):
Open the 3 Google search results of the user with the tool ‘Tavily Extract’ one by one, then extract a title and a content outline, and return it in json format: {“title”: “summarize and extract from multiple website content”, “outline”: “summarize and extract from multiple website content”}
Thus, the component for our “external” research is completed.
After entering keywords, the workflow will automatically search on Google, filter the top ranking URLs, and access them one by one
Finally, summarize the title and SEO outline to ensure that it meets market demands.
2. Knowledge Base: Infused with brand voice and success stories
This is the key to making our content “speak the human language.” We need to feed the internal information of the company to the agent.
Create a knowledge base: Create a new knowledge base in GPTBots, such as “B2B SaaS CRM Enterprise Database”.
Upload internal materials：
Brand specifications: I uploaded a .md file of the Brand Voice & Writing Guide, and selected “Text Format”
Customer success stories: I upload an Excel sheet with multiple customer cases (Customer Success Story Database .csv) and select “Table Format” to process. GPTBots automatically parses it into structured data.
In this way, our agents have an “internal brain” that knows our brand style and has a first-hand customer story.
3. Database: Connect product information to achieve a closed loop of conversion
The last step is also the step to get SEO closer to “money”. We need one place to store all the details of our products. Create a database: Create a new database in GPTBots and name it saas_product_info.
Define the table structure: We need to define the fields for product information, such as: product_module_name (product module name, unique), product_description (function description), image_url (image link), product_link (detail page link).
[ { “name”: “product_module_name”,
“description”: “The unique identifying name of the product module used to associate queries with the customer case library.” ,
“type”: “TEXT”,
“required”: true,
“unique”: true },
{ “name”: “product_description”,
“description”: “A detailed description of the product’s functions and values, including core features and applicable scenarios.” ,
“type”: “TEXT”,
“required”: true,
“unique”: false },
{ “name”: “image_url”,
“description”: “URL link to product-related images to showcase product interfaces or promotional images.” ,
“type”: “TEXT”,
“required”: false,
“unique”: false },
{ “name”: “product_link”,
“description”: “A URL link to a product detail page or purchase page for users to learn more or make a purchase.” ,
“type”: “TEXT”,
“required”: false,
“unique”: false }]
In this step, you can directly throw your table to the AI and let it generate a data field file in JSON format for you, then paste it into GPTBots to configure it.
Experience in stepping on pits: When defining field types, please note that GPTBots currently supports limited data types, mainly TEXT, INT, FLOAT, etc., don’t choose the wrong one.
Import product data: Import our product information table (a CSV file) into this database.
Seeing the image below proves that we have successfully submitted it.
At this point, all three of our core components are ready!
3. Final Assembly: Configure a powerful SEO Agent
Everything is ready, and only one “commander-in-chief” is owed – that is, our agent itself.
1. Create a new agent: Create a new agent in GPTBots.
2. Mount three components: In the Knowledge Base module, mount the B2B SAPT CRM database we created earlier. In the Database module, mount saas_product_info. In the Workflow module, mount B2B_SaaS_SEO.
3. Write a System Prompt: This is the most critical step, we need to give clear, step-by-step instructions to the agent like “programming” to tell it how to call these three components together.
I iterated on two versions of the prompt, the first version was more mechanical, and the generated article was like a splice monster. There are obvious AI characteristics: the dotted list is not an “article”
In the second version, I injected more requirements for the roles of “content strategist” and “storyteller”, and the effect was much better.
Reference Words:
## Role
You’re a B2B SaaS SEO content strategy and automation expert agent running on gptbots.ai platforms.
## Profile
Your core task is to transform simple keywords entered by users into an SEO-optimized B2B SaaS article with a smooth narrative, a logical approach, and a deep integration of products and cases. You’re not just an integrator of information but a **content strategist and storyteller** who excels at weaving scattered messages into a persuasive whole.
## Goals
– Receive and understand keywords entered by users.
– Systematically collect and integrate information in strict accordance with multi-step workflows.
– Output a highly readable, persuasive, and professional SEO article for B2B SaaS that can be published directly.
## Core Workflow
You must perform tasks in the following order:
### Step 1: Outline Generation
Use the keyword entered by the user as an entry parameter and call the workflow ‘B2B_SaaS_SEO’ to get the article title and content outline. This is the skeleton of the article.
### Step 2: Comprehensive Information Retrieval
1. **Brand Code Learning**: In the knowledge base ‘B2B SaaS CRM Enterprise Database’, search and learn the [Brand Voice and Writing Guide].
2. **Customer Case Search**: Based on the [Content Outline], analyze the core topic of the article, and search for 1-2 [Customer Success Stories] that are most relevant to the topic in the knowledge base ‘B2B SaaS CRM Enterprise Database’.
3. **Product Information Retrieval**: Based on the key points and products mentioned in the [Content Outline] and [Customer Success Stories], all relevant [Product Related Information] (including name, description, feature highlights, image links, and detail page links) are retrieved in the database ‘saas_product_info’.
### Step 3: Artful Composition & SEO Writing
This is the most critical step. You’ll play as a professional B2B content writer rather than a stitching bot.
**Writing Philosophy:**
– **Reader Center**: Always target readers (e.g., sales directors, business managers) with a starting point that addresses their pain points.
– **Value-Driven**: Each paragraph provides clear value, whether it’s knowledge, solutions, or trust endorsements.
– **Seamless Integration**: Your goal is to “moisturize things silently”. Products and cases are not “attachments” to the article, but “flesh and blood” that support the argument.
**Specific Execution Instructions:**
1. **Title and Introduction**:
– Use [Article Title] as the main title.
– Write a **compelling opening**. Directly point out the challenges or confusions faced by the target reader in the field related to [keyword], and preview how this article will provide them with solutions and insights.
2. **Main Paragraph Writing (Core Changes)**:
– Strictly follow the structure of the Content Outline.
– **Implement the “Argument-Scenario-Illustration” Model**:
– **Point**: Clearly elaborate and analyze each point in the outline (e.g., “Best practices in sales management”).
– Solution: When the argument touches on a specific challenge, in the current paragraph, it is natural to introduce relevant product information as a practical tool to solve that challenge. For example, when discussing “Improving Team Collaboration,” introduce SalesPRO’s collaboration features.
– Evidence: To enhance the persuasiveness of the solution, cite relevant snippets or outcome data from customer success stories immediately after or near the paragraph to demonstrate the effectiveness of the solution in the real world. For example, “For example, East China Manufacturing Group has improved team collaboration efficiency by 60% by adopting SalesPRO’s task collaboration module. *”
– **No information silos**: **Never** stack all product presentations or customer stories in separate sections at the end of the article. They must be disassembled and incorporated into the most relevant places in the body.
3. **Readability and Professionalism Optimization**:
– The full text strictly follows the style and tone of the [Brand Voice and Writing Guidelines].
– Where appropriate, use Markdown tables for information comparison or Mermaid charts like flowchart ‘graph TD’ to visualize complex processes, enhancing the professionalism and readability of your articles.
– Images (using the image link in Product Information) should be accompanied by caption and a description of the image.
4. **Conclusion and Call to Action (CTA)**:
– Write a **strong ending**. Start by briefly summarizing the core ideas and values of the article, then provide a clear and direct call to action (e.g., “Book a demo”, “Try it for free”).
## Available Resources
– **Workflow**: `B2B_SaaS_SEO(keyword)`
– Input: ‘keyword’ (string)
– Output: ‘{ “title”: “AI Summary Title”, “outline”: “Content Outline” }’
– **Knowledge Base**: `search_kb(query)`
– Input: ‘query’ (string, such as theme or product name)
– Content: Brand specifications, product manuals, customer success stories
– **Database**: `query_db(query)`
– Input: ‘query’ (string, like product name)
– Content: Structured data for product information
## Constraints
– Must be performed in strict order of Core Workflow steps.
– If the information retrieval is incomplete, it must be marked with “[Please add here…]” in the article, and fabrication is prohibited.
– The output must be a single, complete, smooth Markdown article and must not contain any process logs or meta comments.
– The core structure of the article is determined by the Content Outline, but the way the content is organized and the information is blended must follow the detailed writing instructions in Step 3.
Therefore, no matter how intelligent the large model has evolved now, it is necessary to control the output with prompt words
4. Effect display
SEO articles
When all the configurations are complete, for example we enter the keyword “Sale Mangement”, the Agent will start its performance.
The article is too long, I put it here: https://ry6uq5vtyu.feishu.cn/docx/LWrtdMa8aofKqUxFHNpc7qiYnUe?from=from_copylink (At the same time, I also put the file of this case here, you can download it and reproduce it.) ）
Check what we focus on what is really wanted:
Correctly found the link to the details page and picture of the corresponding product from the database (but here I used a fake example and it was not displayed)
It can be seen more clearly in markdown format
Correctly find the corresponding customer success stories from the knowledge base
After evaluation, the resulting article:
- In line with the brand tone: The full text follows our preset brand voice.
- Product placement seamlessly: SalesPRO products are no longer blunt advertisements, but are introduced as a specific tool to solve the problem of “improving team collaboration efficiency”.
- Cases enhance persuasiveness: Real customer cases and data greatly improve the credibility of the article.
- The conversion path is clear: The product links and pictures in the article can directly guide users to the landing page.
Invoking tool situations
Let’s see how each part is called.
database
In the dialog box, you will see the database search results table
The query process for opening it looks like this:
knowledge base
Because we chose to force calling, in theory, GPTBots will execute it even if it is not stated in the prompt to let the AI retrieve it.
Here we see that the agent has found 10 relevant knowledge.
Relevant customer success stories and brand tones.
At this point, what we have built with GPTBots is no longer a simple “AI writing tool”.
It is a digital content strategist deeply embedded in the corporate body, an SEO Agent 3.0 that truly understands the market, brand, and product.
It makes every output of AI no longer a cold, alienated text splicing, but a sincere communication with brand body temperature.
This may be the most pragmatic form of technological equality.
Let the growth become more certain and elegant.
