How I’m Using AI With My Newsletter

Prefer to listen to this story?

I have clients who are all in on artificial intelligence, building a suite of tools to help them write, build, and even sell advertising within newsletters. I have others who are just dipping their toes into the AI waters.

But everyone wants to know: What am I not doing with AI that I should be?

I’ll tell you one thing I’m not doing yet: Using AI to generate newsletters with lots of original writing and voice. Maybe I’ll feel differently about it in a year — I know I’ll get a few emails from readers telling me I’m missing out (and please do email me! I’d love to be wrong about this!) — but for now, I think that voice is one of the biggest differentiators with a newsletter, and I’m not ready to hand that over to AI just yet.

What do I use AI for instead? Three things:

• Process — AI is great for creating custom apps or workflows to speed up repetitive processes.
• Ideation — AI can be a great brainstorming partner. You can give it the outline of an idea, and it can help you improve and expand upon it.
• Analysis — AI is really good at sorting through tons of data — from numbers to the open-ended feedback you collected in a survey — and identifying trends that a person might miss.

Can it be used for some types of writing? Absolutely. I even shared an example below. But I’ve had the most success so far with using AI for one of these three outcomes.

Let’s go deeper into how I’m using AI with my newsletter — and what you might want to try, too.

Coding a section of a newsletter

This fall, I worked with a team to refresh its daily newsletter. We decided that every newsletter should start with what we called our “quick hits” section — a few bullet points to provide an overview of what was in the newsletter, upcoming webinars or events, and a glance at the state of the stock market.

We explored a few different ways to make this happen:

• Email service providers, or ESPs, like Beehiiv have built-in integrations that allow you to easily add stock quotes to a newsletter. However, this client wasn’t interested in switching to a new ESP.
• A third-party email builder like Chamaileon would’ve allowed more design options within the template, but the client wanted to stick with the drag-and-drop tool in their current email platform.
• We could’ve spent hundreds of dollars on an app that would automatically add a stock ticker to the template, but it didn’t seem worth the cost.
• An editor could manually input the data, but it would’ve been so time-consuming.

Instead, I coded something entirely using Claude, which I’ve found to be the most effective AI tool for coding.

I gave Claude some parameters, explaining what fields we needed to input and what colors and fonts we wanted to use. I even used Claude to find a free API to pull in the stock data. (We used Finnhub.)

After a few minutes of going back and forth with Claude, plus some feedback from the client, I’d created a tool that generates the entire section for the client. An editor pops in short descriptions of four things in that day’s newsletter. (Down the road, we might ask AI to generate this for us.) They click a button, and the latest stock market updates appear automatically. If they have an event to promote, they can add that. Then they click another button, and Claude spits out a bit of email-friendly HTML, which the editor copies and pastes into the template. The rest of the newsletter is built using their ESP’s drag-and-drop builder.

A tool that took me less than an hour to build, without requiring any coding knowledge, will save the client a few minutes every day and cost them $0 to use going forward.

Coding an entire newsletter

When I worked at BuzzFeed, I came in with a few ideas for our new daily newsletter. I sketched out some ideas in a little red notebook and gave it to a designer, who immediately made some terrific improvements to my scribblings. But it still took us a few weeks to turn that template into something I could send via our email platform.

Why? We had lots of amazing people on our product team, but none focused on email design, and coding for newsletters is a bit different than coding for the web or apps.

What’s amazing is that now, thanks to AI, we could’ve created a template in an afternoon, just from that wireframe sketch.

If you have a basic sketch of the newsletter — images go here, headlines go here, etc. — you can take a photo of it, upload it to a tool like Claude, and it’ll build an entire template from scratch. You’ll have to offer some edits and improvements from the initial template, but you can at least create something to show your colleagues. For a non-designer, non-coder like myself, AI lets me quickly turn the ideas in my head into semi-functional newsletter templates. 

I usually build the final version in the client’s email platform so we can utilize the ESP’s native email builder, but this allows us to speed up the process of creating mock-ups to share with the client.

Some clients take things a step further and create entire, ready-to-use templates with AI.

Here’s an example: Let’s say you’re a local newsroom that sends out a breaking news alert when there’s big news in your community. The template is pretty basic: 

• Your logo at the top
• The headline for the story
• 1-2 sentences about the story
• A button to click to read more
• Your footer

This is the kind of thing that AI is perfect for.

First, you’d create a wireframe of what you want the template to look like or take a screenshot of another newsroom’s breaking news template and upload that to Claude. Then you’d give it some key information, like your logo or correct footer information. Once you’ve got things the way you want, you’d create more instructions. You’d tell Claude to create a generator that feeds into that template. You’d tell it that some elements of the template — in this case, the logo and the footer — should remain exactly the same every time. Then you’d tell Claude that you want to be able to input the URL for the story, and that it should copy over the headline and the first sentence of the story into the template. You’d tell it to add the link to the URL in the headline and in the “Read more” button.

What it creates is a custom app for building that breaking news email. You’d insert the URL into the app, and AI automatically adds the right pieces to the template. Click a button, and it spits out the HTML for you to copy and paste into your ESP.

Some clients take things even further, adding integrations via tools like Zapier to automatically port over the HTML into their ESP, where their team can review everything and send.

And some go a step beyond that, asking AI to generate the subject line and preheader text for the email, porting everything into their ESP, and then sending the email to a specific list.

Analyzing or creating subject lines

Most email platforms allow you to download a .csv with your most recent newsletter sends. It’ll tell you the subject line for that email, as well as details like open rate. One interesting exercise is to take that .csv, upload it to an AI tool — I haven’t seen much of a difference between platforms like ChatGPT, Claude, or Gemini for this — and ask it to identify any trends. Do certain topics tend to drive higher open rates? What about subject line length? What about tone? If you give it enough data — and I’d recommend uploading at least a few months’ worth — it may be able to identify some best practices for you.

Also, once you’ve had it analyze enough data, you can ask AI to help you create subject lines. You’d give it a few paragraphs from your newsletter, and it can generate some ideas based on the best practices and results you’ve previously shared.

In fact, many ESPs have already added these types of AI tools to their platforms. AWeber has created integrations with ChatGPT and Claude that help you analyze email performance right from their platform; Mailchimp added a tool to suggest subject lines for the newsletter. You should absolutely test out these native AI tools and see if they work for you.

Digging into survey data

One of my favorite uses of AI is with surveys.

I love a good survey. I think it’s one of the most powerful ways to learn more about your audience. The next time you run a survey, ask something open-ended, like:

• Is there anything we could do to make this newsletter better for you?
• What would you like to see changed about our newsletter?
• If you could add one thing to our newsletter, what would it be?

There’s a reason why many teams don’t ask a question like this: It used to be a pain to sort through all the data. I’ve seen surveys where teams have to go through thousands of open-ended responses. Some of these replies are incredibly valuable — there’s absolute gold in there if you’re willing to dig through all the answers. But if you get a thousand replies, you might only get a few hundred answers that are really worth reviewing. (The rest are answers like, “Nope, your newsletter’s great! Keep up the good work!”)

But this is where AI really shines. Upload the responses (for privacy purposes, be careful to upload only the answers and not any of the user’s personal data) and then ask AI to break down the results. I like to ask AI things like:

• What are the big themes you’re noticing from these answers?
• Are there any unusual answers that stand out to you?
• What specific action items would you suggest I take based on the feedback here?

You can also ask AI to create documents based on all this feedback. Instead of sharing the link to the results page, ask it to create a Google Doc or a slideshow that shares some of the key data and takeaways. You can always edit it before you share it with your colleagues, but it usually does a good job of translating all the data into something easier to present to a larger group at a meeting or in an email.

Writing cold emails to sponsors

When I was building the Newsletter Advertising Playbook with Ian Barto and Nate Heintz of Ad Astra Media, we discussed how to reach out to potential advertisers. Ian and Nate spend a lot of their day getting in touch with advertisers who’d be a good fit for their clients, and that means sending multiple emails, over the course of a few weeks, to tons of people at lots of different companies. 

So we had a thought: Why not bring in AI to help?

As we wrote in the Playbook:

We also know that it can be tedious — and time-consuming — to write these emails. This is a case where it may make sense to use AI to write these emails for you. In our experience, the better the prompt you give the AI tool, the better the email it can write.

We built a template to collect key data about your newsletter, the sponsor, and the person you’re reaching out to. After inserting one of the prompts we created, you’d upload it to ChatGPT, and it would spit out the email copy.

As I said at the start of this article, I’m not ready for AI to write an entire newsletter (or a guide like this) just yet. But a cold email to a sponsor? I might need to send a few hundred (or more!) of those in a single month, and a dull email is one that gets deleted. (Even great emails get rejected more often than not.) But AI can be a useful writing partner for emails like these.

(And yes, when you buy the Newsletter Advertising Playbook, you get access to our cold email template, prompts, plus a lot more.)

Repurposing newsletter content for social media

Here’s one I like to call “The Thanksgiving Dinner Strategy.”

Say you’re hosting Thanksgiving dinner. What happens the day after the big meal? You open the fridge, look at what’s left over, and figure out how to make a meal out of it, whether it’s soup, a casserole, a sandwich, or just some tasty snacks.

The same philosophy should hold for anything you create for your newsletter. That article you wrote? That interview you conducted? You did great work with that, and you can get more out of it. You should repackage it and turn it into great social media content.

Whether it’s condensing a long article into a few bullet points for LinkedIn or turning that case study into a thread for Twitter, you should think about how to take the great content you produce and reuse it on social media to help you grow your audience and convert new readers into newsletter subscribers.

This is something AI is actually quite good at. I’ve found that certain tools, like NotebookLM, are excellent at identifying key points or themes in a long article. Others, like Spiral, are quite good at helping you write both in a specific voice and for specific social media channels.

Newsletter research and marketing

I think AI really shines as a research partner. If I’m brainstorming an idea for a new newsletter, I like to go back-and-forth in conversation with AI. I’ll walk through the basic idea for the newsletter and ask it to poke holes in the strategy. I’ll ask it to help me identify angles I might have missed. And then I’ll take all of that and put together the pitch for this new project.

There are two things to keep in mind if you do use AI for brainstorming. One is that you still have to show up with your own ideas. The more you bring to the table, the better AI can help you improve upon that idea.

The other is that AI tools tend to be relentlessly positive. You can show up with an idea that’s been done a thousand times, and it may tell you that it absolutely loves the idea. Don’t be afraid to push back, ask for criticism, and be willing to dig a little deeper to help improve your idea.

(This is also a good moment to note that AI is far from perfect. It can hallucinate or invent facts. It gets stuff wrong. If you’re using AI for writing, in particular, you’re going to want to carefully review and edit whatever it produces.) 

Another great use case: Using AI to help you build out a marketing strategy. Let’s say you want to launch a paid offering, like a membership or an ebook. Should you dig into the Inbox Collective archives for relevant ideas? I’d hope so! But you can also turn to AI for help. 

Explain to your AI tool what the new product is, who it’s for, and how it’ll help them. Then ask them to help you develop the marketing strategy.

The fun twist: AI knows a lot about some famous marketers through history. Instead of asking ChatGPT to give you generic marketing advice, tell AI to imagine they’re someone in the marketing space you respect. (This person doesn’t have to be alive or even real. Ask it to impersonate P.T. Barnum or Don Draper.) Then ask AI what that person would suggest you do as part of your marketing strategy. It’ll spit out suggestions for ways to market your product, sometimes even going a little too far over the top. (When I asked AI for Don Draper’s advice on selling an ebook, it opened with this line: “[Lights cigarette, stares out window].”)

From there, you can always ask it to clarify, to share examples, or to suggest additional ideas from another marketer. I’ve found that it does a good job of suggesting marketing angles I might not have thought of. As I worked on the Newsletter Advertising Playbook, for instance, it came up with a few original angles for webinars that we might try this year. In some cases, I’ve even given it content to help write the first version of a landing page or suggest themes for a post for LinkedIn.

Do I let AI completely take over the process? No. But it can certainly give you the kickstart you need.

I’ve shared a lot of ideas here for using AI. But there are two more things to consider as you figure out what AI strategies make sense for you and your newsletter.

Finding tools for your business

Ever since I launched Inbox Collective, my newsletter and my business have been linked. The newsletter helps me attract new clients, get people out to my events, and buy products that I sell. Here are three more AI tools that I use in connection with my business. They’re not things I’m using in my newsletter, but they help me with the business that comes from the newsletter.

Granola is an app that can listen in on my meetings (both on Zoom and IRL) and then suggest follow-up steps afterwards. I’ve tested a few of these, and it’s easily my favorite to use. I love that I can filter calls from specific clients into folders and search them to find items I might have forgotten from a previous call. (Plus, it doesn’t show up as a random black box on Zoom calls. Clients hate seeing a random AI bot show up on their video screen.) It’s been a huge time-saver for me.

Tella allows me to record videos and share them with clients and customers. Let’s say a client is having trouble with something in their ESP and asks to set up a quick call to troubleshoot. Since I have access to their accounts, I can pop up, use Tella to record a video explaining what to do and how to fix it, and send it over to them. I can also have Tella zoom in at various parts of the video (it follows your cursor), and it can quickly remove silences, filler words, or mistakes. We’ve been using Tella to record the customized ad assessments for everyone who buys the Newsletter Advertising Playbook. A year or two ago, a video like that would’ve taken me an hour or more to edit. Now I can record and edit the entire thing in less than 20 minutes.

WisprFlow is a voice-to-transcription tool. It’s so helpful for replying to my readers. In my welcome series, I ask readers to reply and tell me more about what they’re struggling with. I get a lot of emails about this — one out of every five readers writes back, and these notes are often thoughtful (and long). To speed up the reply process, I’ve used WisprFlow. I’ll hit a button on my computer — there’s also an iPhone app — and share a few words of advice. I can type pretty quickly, but an email that might take me 2-3 minutes to answer is now one I can reply to in 30 seconds. When you send a lot of emails, being able to speed up that process is a huge win.

Create your own AI policy

Regardless of how you use AI, I’d encourage you to create an AI policy and publicly list it in your newsletter and on your website.

Most publishers have a privacy policy that explains what they do with your data. Creating an AI policy is a similar move, but for AI, to help shield you from criticism (or, potentially, legal action) from readers.

This is a strategy that a client, Phil O’Brien at W42St, first introduced to me. On your AI policy page, you explain to readers how and when you use AI. For instance, do you use it to write stories? To edit stories? To create videos or images? You can tell readers that in your AI policy. Let’s say you use AI to come up with ideas for subject lines, but a human editor reviews everything before you hit send. That’s absolutely fine — just tell readers that in your AI policy.

Here’s W42St’s AI policy — I think it’s an excellent example. (Here’s the Inbox Collective AI policy, if you want a second example.)

There’s one more step: Once you’ve created this page on your website, add it to the footer of your newsletter. You might already have a link to a privacy policy down there, and you can add the AI policy next to it. That way, if readers do have questions about how you use AI, you’ve linked to a document that clearly explains your policy.