A large language model (LLM) runs on a statistical program, not a brain. But AI agents can still help us write — with the right prompts. Working with an LLM is as much about us speaking its language as the other way around.

General tips

• Use individual queries rather than putting many questions into one query. Use a back-and-forth technique, like a conversation, to build context.
• Provide statements (“Create a…”) rather than questions (“Can you create a…”).
• If you don’t get what you need the first time, correct and iterate.
• Tell your AI companion to clarify the request if it doesn’t understand what you’re asking for.
• Tell it to cite sources for research and verify the results before using information.
• Delete UTM tags from any links you include in your piece. (This is the “?utm_source=chatgpt.com” at the end of the URL.)
• Save your most successful and frequently used prompts in a library.

The following sections cover some of the most common AI use cases for writers and suggested prompts that we’ve tried, refined, and found helpful.

Research and interview prep

Understanding the context of a topic or how something works

“Explain [topic] to me as if I were five years old.”

Preparing for a SME interview

“I’m preparing for a SME interview for an article on [topic]. Give me some context on [X] theme and key points to explore in that interview.”

Identifying different perspectives (These can lead to potential “challenge questions” during an interview.)

“What does this mean for…”

“What are some possible objections or alternate beliefs?”

Avoiding detail overload

“Now that we’ve talked about all these topics, summarize where we stand on this theme.”

Tailoring your content for the audience

Creating personas

“At a [size/annual revenue/industry] company in [region/country/state], what role(s) or job title(s)
would most likely be concerned with [topic]?”

“What are some of the toughest problems or challenges faced by a [job title] at a [size/annual
revenue/industry] company in [region/country/state]?”

“Build me a persona of a [job title] at a [size/annual revenue/industry] company in [region/country/
state]. This person is looking for help with [problem], and they are considering [purchasing this product/
service]. List their hopes/dreams, fears/concerns, emotional triggers, and decision criteria.”

Identifying a compelling lede

“Why would someone who [has this job title/works in this industry] care about [topic]?”

Foundational content creation

Creating an outline or a rough first draft

“I’m writing an article for [audience] about [theme]. It should be between [X and Y] words long and cover [topics] and [approach, tone, call to action]. Provide me a rough first draft that does not have an introduction or a conclusion, but is more of a ‘data dump’ covering the relevant topics. I’ll use this to iterate and refine and will use it as a starting point for deeper research.”

Fine-tuning

Improving organization and flow

“Does this example fit best in this paragraph or this
other one?”

Creating transitions between paragraphs

“Suggest a transition from this section to the following one.”

Creating headlines and titles

“Suggest some catchy, newsy titles for this article. Do not use colons. Title should be 70-100 characters, including spaces.”

Ensuring you’ve responded to client feedback

“Here’s my original paragraph, here’s the client feedback, and here’s my revision. Did I capture everything?”

Creating a fake color-team reviewer for a proposal

Before you drop the RFP into the LLM, scrub any proprietary or CUI data.

“Is this section responsive to the requirements listed in the RFP? If not, please identify the elements that
are missing and recommend ways to address those requirements.

Creating derivative content

Creating social media posts

First, share some examples of client social posts.

“Pasted below are some examples of this company’s past social media posts. Using these as examples of voice and tone, provide three LinkedIn posts based on this article, with length around 100 characters.”

“Create three LinkedIn posts based on the quotes included in this article.”

Creating one-pagers, FAQs, slide decks, or lists from a case study (Great for permissions follow-up.)

“Create a list of the main points made in this case study.”

“Create a list of all products/companies/people mentioned in this study, with page references.”

Creating a proposal compliance matrix from an RFP

Drop solicitation in (scrubbed of any proprietary data).

“Build me a table with each requirement listed on its own line, and then tell me where you find that
requirement in the document.