Using AI to Create Insurance Marketing Content

Most agents who try AI for marketing run into the same wall. They open a chat box, type "help me write a social post about Medicare," and get back something generic that doesn't sound like them and doesn't say anything a senior would actually care about. Then they decide AI doesn't work for insurance.

The agents who get real value from AI use it differently. They don't ask AI to write their marketing for them. They use AI as a fast first draft, then they edit it down into something that sounds like them and earns trust. That edit is the difference between marketing that converts and marketing that gets ignored.

This training walks you through the practical workflow. What to make, how to prompt for it, and the editing pass that turns AI output into content you'd actually publish.

Here's why this matters. Marketing time is the bottleneck for almost every solo agent. You know you should be posting consistently. You know you should have a monthly email going out. You know a few short videos would help your local visibility. But sitting down to write any of it from scratch eats 2 hours every time, and after a long week of appointments you don't have 2 hours.

AI compresses that. A post that used to take 90 minutes takes 15. An email that used to take an hour takes 20 minutes. The volume of content you can produce goes up 4 or 5 times without working any extra hours. That's the real unlock.

The fear most agents carry is that AI content will sound robotic, or worse, that it will say something that crosses a compliance line. Both fears are reasonable. Both have the same fix. You never publish what AI gives you. You publish what you publish after editing AI's draft. The model is a writing partner, not a publishing tool.

Before any prompt, three rules. First, never let AI write plan-specific claims. Premiums, copays, benefits, drug coverage details. Anything plan-specific gets pulled from carrier-approved materials, not generated. Second, the TPMO disclaimer rules apply to AI-generated content the same way they apply to anything else, and any video, audio, or written ad that promotes Medicare plans needs that disclaimer. Third, never put a client's personal information into a public AI tool. Names, dates of birth, plan details, anything that identifies a real person. That data leaves your computer and you don't fully control where it goes.

With those rules in place, here's the content stack worth building with AI.

There are 5 marketing assets that pay for themselves for a Medicare-focused agent. Short social posts. Email newsletters. Short video scripts. Long-form blog content. And conversational ad copy. We'll cover the first 3 in detail, and the workflow for each follows the same pattern. Define your topic. Give the model your voice. Get a draft. Edit hard.

Start with social posts because they're the highest-frequency asset and the lowest stakes. The goal of an agent's social post isn't to sell. It's to be useful, be recognizable, and remind people you exist. A good agent post looks like a tip, a reminder, or a small clarification of something seniors get confused about.

Here's the prompting pattern. You give the model a topic, an audience, a length, and a sample of how you actually talk. The prompt sounds like this. "Write a 60-word LinkedIn post about the difference between the Annual Election Period and the Medicare Advantage Open Enrollment Period. Audience is people who are already on Medicare and are confused about the dates. Tone is plainspoken, not corporate, no exclamation points, short sentences. Here's an example of how I write." Then paste 2 or 3 short paragraphs from an email you've actually sent.

That last part is the magic. AI imitates whatever voice you feed it. If you give it nothing, you get the default LinkedIn voice that everybody recognizes as AI. If you give it 3 paragraphs of your real writing, the output sounds like you with about 80% accuracy. The other 20% you fix in the edit.

The edit is where most agents stop too early. Read the draft out loud. Cut anything that sounds like a brochure. Replace any phrase you'd never actually say. Take out at least one sentence — AI tends to over-explain, and the post lands harder when you trust the reader to follow you. Add one specific detail from your actual practice. The phrase a client said, the town you're in, the time of year. Specifics are what stop the post from feeling generic.

The second asset is the monthly email. The pattern is similar to social, but the structure is more important because email asks more of the reader's attention. A working agent email has 4 parts. A short personal opener. One useful insight. One soft mention of what you're working on. A clear sign-off.

The prompt sounds like this. "Write a 250 word monthly email to my Medicare client list. Topic is what to look for during the upcoming Annual Election Period. Audience is current clients who are happy with their plan and not actively shopping. Tone is warm, plain, like a letter from a trusted neighbor. Use 4 short paragraphs. Don't mention any specific carrier names or plans. Here's how I write" — paste a real email.

When you get the draft back, the same edit rules apply. Read it out loud. Cut the brochure phrases. Add one specific. But the email gets one extra pass that social doesn't need. You check the call to action at the end. If the email ends with a vague suggestion to reach out, you change it to something concrete. The line that works is, "if anything in your situation has changed this year, hit reply and tell me about it." That single line outperforms any other email closer for most agents because it gives the reader an actual job to do.

The third asset is short video scripts. AI is genuinely strong at video scripts because video scripts are short, structured, and benefit from a clear point of view.

Here's the format that works for 60 to 90 second videos. One sentence hook. One sentence problem. 2 to 3 sentences of useful content. One sentence call to action. That's it. 5 to 7 sentences total. Anything longer loses the viewer.

The prompt is, "write a 60-second video script about why somebody on Medicare should review their plan every fall, even if they think nothing has changed. Audience is current Medicare beneficiaries. Format is hook, problem, three insights, soft call to action. Tone is conversational, like talking to a neighbor over the fence. Here's how I sound" — paste a transcript of a real conversation you had with a client, or a short voice memo you transcribed.

Voice memos are the secret weapon here. Talk into your phone for 90 seconds about the topic. Run the audio through any free transcription tool. Paste the messy transcript into your AI prompt as the voice sample. Your real, conversational rhythm is now in the model. The script that comes back will sound like you talking, not like a corporate voiceover.

Here's what the full workflow looks like in practice. An agent named Jen sets aside 90 minutes every Monday morning. In that block, she produces her week's content. 3 short social posts, 1 monthly email if it's the first Monday of the month, and 1 short video script. She uses the same chat thread every week, with her voice samples and her 3 rules pinned at the top.

In 90 minutes she gets first drafts of everything. She edits each one for 10 to 15 minutes. She schedules the social posts to publish across the week. She sends the email Wednesday. She films the video script that afternoon on her phone, in front of her office bookshelf, no production. By Monday afternoon her marketing for the week is done. Total time from blank page to published, about 3 hours.

A year ago, that same output would have taken her 8 to 10 hours. The 4 to 5 hours she got back, she spends on appointments. That's the real win. Not better content. The same quality of content, in a fraction of the time.

Common mistakes to avoid. Number one is publishing the first draft. The first draft is never the post. The edit is what makes it yours. Number two is using one prompt forever. Save your prompts in a notes app, refine them every month, and watch your output quality climb. Number three is letting AI invent statistics or quote rules it isn't sure about. Anything specific — enrollment dates, eligibility ages, regulatory rules — you verify against the official source before it goes anywhere public.

Here's your action step today. Open a new chat with whichever AI tool you already use. Paste in 3 real emails or social posts you've written in the last 6 months. Tell the model, "this is my voice, study the rhythm and the word choices." Then ask it to write 1 sixty-word LinkedIn post about a topic your clients ask about all the time. Edit the draft for 10 minutes. Post it. The first one is the hardest. After that, you have a system you can run every week for the rest of your career.

Download

Because most agents prompt without giving the model a voice sample. AI defaults to the generic LinkedIn voice everybody recognizes as AI. The fix is to paste 2 or 3 paragraphs of your real writing — an email you sent, a transcribed voice memo, a real client conversation — into the prompt as a voice sample. With that anchor, the output sounds like you about 80% of the time. The other 20% gets fixed in the edit pass.

Three rules apply to every piece. First, never let AI write plan-specific claims like premiums, copays, benefits, or drug coverage details — those come from carrier-approved materials only. Second, TPMO disclaimer rules apply to AI-generated content the same way they apply to anything else, and any video, audio, or written ad that promotes Medicare plans needs the disclaimer. Third, never put client personal information into a public AI tool — names, dates of birth, plan details, or anything that identifies a real person.

Five to 7 sentences total. One sentence hook. One sentence problem. 2 to 3 sentences of useful content. One sentence soft call to action. Anything longer loses the viewer. Voice memos run through a free transcription tool and pasted as the voice sample produce scripts that sound like you talking instead of a corporate voiceover.

The edit pass. The first draft is never the post. Read the draft out loud. Cut anything that sounds like a brochure. Replace any phrase you would never actually say. Take out at least one sentence — AI over-explains. Add one specific detail from your actual practice. Specifics are what stop the post from feeling generic. Most agents stop the edit too early and publish content that still feels off.

About 90 minutes on Monday morning to produce first drafts of 3 social posts, the monthly email if due, and 1 short video script. Add 10 to 15 minutes per piece for the edit pass. Total time from blank page to published is about 3 hours per week — compared to 8 to 10 hours doing the same volume from scratch. The hours you save go to appointments.