How I think about building a content marketing function with AI

First of all, I dislike the word "content." It feels vague and comes across as a buzzword, much like when someone tells you their business is a "platform," or starts talking about "enablement." Like…what does that mean?

I take a pretty broad definition myself, but to me content is your evergreen, always-on organic marketing. A campaign has an end date, and paid stops the day you stop paying for it. But content is the blog post that still ranks 2 years later, the customer story your sales team forwards every week, or the LinkedIn post someone sends to their coworker a month later.

If content is always on, the team that produces it should be built like an operating system. No more trying to fill spots on a calendar.

Here's how I'd approach building a content marketing function in 2026:

1. Idea generation = AI + the real facts about your business. I don't think anyone should be starting at a blank screen to generate ideas in 2026. But I don't think anyone should be blindly prompting AI ("hey, Claude, what are some ideas for blogs that my customers would care about?") for ideas either.

The sweet spot is to meet somewhere in the middle. An always-updating feed of call transcripts, customer feedback, online comments, a daily drip of news from your industry, even internal meetings or objections from sales outreach – should be going into a project or artifact on Claude every day. THEN you could prompt your project if searching for new content ideas. Or better yet, set up an automation to surface what's going on at your company so you don't need to do a daily or weekly prompt.

The set up here is pretty quick, even though this would require some upkeep.

2. Use AI to refit and reformat content for as many channels as you can. The advice used to be to pick 1-2 social platforms and do them well. I think this is still true for some businesses, like lifestyle businesses where the founder really is the face of the brand.

But if you're a B2B brand, I see no reason why you can't use AI to clip your videos, rewrite your captions or generate resized images for three, four even five platforms. Maybe this grows over time, and you master each workflow one by one. But AI is perfect for this type of repeatable work.

I'd rather have fewer pieces of high-quality content shared across more platforms than many pieces of poor-quality content shared across a few channels every day.

Some people might say, "but then what about the quality?" my response is then #3…

3. Test every piece of content through a board of directors to evaluate quality. Create a GPT or skill composed of 3-5 personas based on your customers. 1-2 could be your "fans," so customers or prospects that are excited to hear what you say and love what you do, but others should be skeptics. People using a competitor right now. People who complain about price. People who question certain features.

The goal is not to "please" each of these personas, but to score and grade each piece of content on a scale from 1-100. Score each piece on how it answers questions like: Does it say something only we could say? Would this persona forward it to a colleague? Does it name the objection this persona actually has? Does it read like a person wrote it? Is the ask obvious? Publish at 70 and above. Revise between 50 and 69. Get rid of anything below 50.

Of course, you must understand what your customer pain points are and feed this into your model so the personas understand this and you can develop your quality scoring system. Which is why #1 on this list is so important.

4. Use AI so content marketers can run your experimentation arm too. Historically, content teams haven't done much experimentation or A/B testing. Tell me if you've seen otherwise. But in companies of 3+ marketers, I've seen experimentation typically fall to the "website" person, designer or occasionally paid media, since A/B tests traditionally required lots of new imagery, copy or a somewhat-complicated set up.

But it's time for content teams to take this on. The person responsible for creating the LinkedIn post or Instagram post should also be responsible for making sure the messaging on the website matches, the customer journey is seamless, and that the best variants are being put forward. And if we refer to #1, it only makes sense that the person who's closest to real customer insights translating these into the winning, durable marketing on a website.

Tools like Coframe make standing up tests so efficient, and don't require the support of multiple team members. You point it at a page, it writes and serves the variants, and it promotes the winner. There is no longer the operational headache of coordinating with lots of teams to run tests.

5. Approach content as a team sport if you're not already. Create a slack channel or teams channel (or whatever you use) for content. Set up an automation so everyone on your team will see every new post as it goes out so they can like and share. Encourage people to participate! Make it fun.

And yes it really is that important. There is nothing wrong with the majority of your "likes" coming from people who work at your company, especially as you're building. Because if the employees don't believe in what you're doing and saying, why should others?