These methods separate the writer from the genAI assistant
As an experienced B2B content developer, I can tell you that generative AI is a lousy writer. While a great text generator, it lacks tone, style, and voice in its writing, even when told to match your own. But that doesn’t mean content developers—including creative writers—should avoid genAI applications. In fact, there are many effective ways to use generative AI for content development without letting it write all the text for you. These methods lean on the model’s other capabilities to organize data.
Here are six effective ways to use generative AI as a content developer.
1: Search Engine Optimization (SEO)
Generative AI models are very good at weaving SEO keywords into your existing content. That includes awkward keywords that are hard to fit into otherwise good writing. Instead of spending hours trying to make them fit, upload a file with a list of keywords to a GenAI application and tell the application to insert the words for you. I’ve found that Google’s Gemini has worked best at this task. It will leave most of your text alone and only add your keywords, even when the list is complex and the text is long. After uploading a list of keywords, prompt this: “Insert all the keywords in [Name of SEO document name here] into the most natural place they fit in the [blog/white paper/use case/landing page] called [target document name here]. At the end, make a list of any keywords you did not include.” Although I always ask it to provide a list of the words it didn’t use, it usually can place them all. It’s important to check their placement to ensure the application worked as you expected. I’ve used genAI to place SEO keywords in many of the blogs in my portfolio.
2: Document parsing and summarizing
If you’ve been a writer for a while, then you know that reference materials can be very long. Sometimes, all you need are two points buried on page 32 of an 80-page PDF. GenAI can analyze PDFs and slide decks to summarize the entire thing or find specific information you need. Behind the scenes, applications use computer programming to parse text and image documents. Using the prompt, the applications can then act as a searchable knowledge base. For example, to find a specific topic in a lengthy document, prompt this: “I have attached [file name of document]. Please extract all mentions of [specific topic] and extract them into bullet points. Provide the page numbers where this information was found so I can verify the context.”
3: Outline creation and text organization
More than just a structure, a verbose outline can give you a better map of how to write content. A genAI model can take a mess of text and create a structured outline to begin a blog, case study, or landing page. I’ve found OpenAI’s ChatGPT best at this task. Upload or paste the content into the prompt. Then prompt: “Create a verbose outline about [topic ABC] in the attached file called [document name] that includes [5] sections and [1,200] words.”
4: Research and citation assistance
You can simplify the research process by letting genAI rapidly scan large volumes of online sources to find relevant information for your content. With the right prompt, genAI can not only return important facts and statistics but also provide citations or links to sources, making fact-checking much easier. For example, prompt: “Research the latest statistics on [topic] from [name reputable sources, such as major newspapers, trade magazines, or scientific journals] and provide the source links. Summarize the findings in 2-3 sentences and list all sources below each point.” However, I have known genAI to hallucinate sources, so you must check each source it provides. Moreover, Carnegie Mellon University researchers said in July 2025 that large language models (LLMs), the heart of genAI applications, remained confident even when they were wrong.
| Example | Description | Recommended Prompt |
| SEO | Weaving complex or awkward keywords naturally into existing content. | Insert all the keywords in [Name of SEO document] into the most natural place they fit in the [document type] called [target document name]. At the end, make a list of any keywords you did not include. |
| Document Parsing | Analyzing long PDFs or slide decks to summarize content or act as a searchable knowledge base. | I have attached [file name]. Please extract all mentions of [specific topic] and extract them into bullet points. Provide the page numbers where this information was found so I can verify the context. |
| Outline Creation | Transforming text into a structured, verbose map for content. | Create a verbose outline about [topic ABC] in the attached file called [document name] that includes [5] sections and [1,200] words. |
| Research & Citations | Scanning large volumes of online sources to find facts and source links. | Research the latest statistics on [topic] from [reputable sources] and provide the source links. Summarize the findings in 2-3 sentences and list all sources below each point. |
| Table Generation | Converting data or existing text into usable tables for further editing in spreadsheets or word processors. | Scan the text in the attached document called [document name]. Determine what would make a good table. Then generate the table. |
| Math & Logic | Identifying errors in mathematical calculations and logic sequences within technical or educational reports. | Review the attached document called [document name] for equations and calculations and check them for accuracy. Explain any errors and provide corrected versions. Format the output in LaTeX. |
Note: This table was generated with Gemini.
5: Table generation
GenAI applications can be useful for converting data into tables—if you are not trying to reproduce infographics or charts. What’s even better? They can scan your text and determine what information would make a good table. Instead of manually organizing them, you can simply provide text, images, or descriptions and let the model build a table for you. While charts and tables created in AI can be further enhanced in design software, most genAI models cannot get important technical details right in a graphic. However, Anthropic’s Claude recently started producing technical graphics, and they are worth checking out. GenAI apps structure text well into a usable table that can be edited further in word processing software or spreadsheets. To have it find and create a table from your existing text, prompt, “Scan the text in the attached document called [name the document here]. Determine what would make a good table. Then generate the table.”
6: Math and logic proofreading
Mathematical calculations and logic sequences are native to genAI applications. For technical or educational reports, you can use genAI to identify errors and even explain mathematical reasoning. I’ve found that Claude works best for this task. If your content contains formulas or quantitative analysis, prompt: “Review the attached document called [name your document here] for equations and calculations and check them for accuracy. Explain any errors and provide corrected versions. Format the output in LaTeX.”
Leave a Reply