How publishers can build AI workflows their teams will use

The more I experiment with AI, the more I discover it can do for the work we do every day. I still catch myself thinking, “I can’t believe I just got it to do that.”

One of my biggest takeaways lately is that the best ideas often come from listening to someone describe a task that takes too long, or information they wish they could get their hands on more easily. That’s usually where AI can make a real difference in someone’s workday.

We’re building AI workflows with several clients right now, and I want to share six that I think are especially relevant to publishers.

We first built many of these as custom GPTs in ChatGPT, but as the platform moves away from GPTs, we are now moving them into Projects and Plugins.

1. Meeting agenda tracker

Because it’s not necessarily easy to stay on top of every agenda for every public meeting, we built an agenda tracker that finds publicly available agendas on government websites and summarizes the upcoming meetings a newsroom may want to watch.

It also shows which agendas it reviewed without finding a potential story and which agendas it couldn’t locate. That way, editors can see what was checked and where they may need to look themselves.

Editors are telling us this helps them spot stories they might have missed and decide which meetings are worth a reporter’s time.

2. Podcast Shorts and posts

One of the best ways to promote a podcast episode is to pull short clips for YouTube Shorts and other vertical video formats. I’ve always found that these work better when the clip offers something useful on its own, rather than feeling like a commercial for the full episode.

The challenge is figuring out which moments from a 30-minute conversation will make the strongest clips. So we built a workflow tool that lets someone upload a transcript and get several clip ideas, along with timestamps showing where to find each moment in the video.

It also creates a short digital recap to pair with each clip as a post. Everything it suggests and writes comes from the transcript, so it stays true to what was actually said.

A screenshot of our image sizing tool

3. Image sizer

Many publishers need to size images for print and their websites to specific dimensions.

That often means resizing photos one at a time. Across dozens of articles a month, the time adds up.

So we built an image resizer with newsroom standards in mind. It keeps photos in proportion, avoids stretching or inventing parts of an image, and lets an editor check that the crop preserves the subject and important context.

Editors can upload photos in batches, select the sizes they need, preview each crop and download the finished files as a ZIP.

4. Google Search Console

Google Search Console has a ton of great information, but it honestly can be overwhelming to know where to start and what it all means.

That’s why we created a tool that reviews Google Search Console data to find stories appearing in search results but getting few clicks. It shows editors the search terms for which those stories appear and recommends specific title changes or content updates that could help more readers find them.

The tool gives the team a short list of specific actions, including suggested SEO titles. The idea is to help an editor move from looking at the data to knowing what to do next.

A screenshot of the side-by-side window

5. Side by side editing window

One of the AI tools we’ve built that gets the most use helps reporters and editors review a story before it’s published.

It doesn’t rewrite the story, but rather it suggests specific edits to improve the writing, grammar and style, and flags questions or gaps the reporter may want to address.

The most useful part has been seeing the original wording and suggested edit side by side, with a brief explanation of why the change might help. I shared an example above from a post a few weeks ago that I wrote so you can see what that looks like.

6. Branded content ideas

While it’s not necessary for a sales rep to nail down the exact topics of a branded content campaign, providing them with a few ideas and topics as part of the prospecting process can be really helpful.

This tool asks a sales rep to enter a little information about an advertiser. It then suggests branded content story ideas, questions readers might have about the advertiser’s industry, and related search terms. This has been helpful as it’s providing the rep something more specific to bring into that conversation with a client.

How we can help you build AI tools

Through our AI program, we spend two to three months working with a publisher’s team. We start with a detailed audit of how work gets done, then identify where AI could help with existing tasks and show the team other ways to use it.

We help write an AI policy, build tools in the publisher’s own system, refine them with the people who will use them and train the team to use them. If you’d like to explore what that could look like for your organization, I’d be glad to talk.

Need help growing audience and revenue?

🚀 Want to know how you can work with us? This page outlines our services, approach, current clients and more.

🌐 Follow our LinkedIn and Instagram pages to stay up to date on the latest in media, our blogs and recommendations.

📬 Sign up for our monthly LinkedIn newsletter which provides company updates, tips and more.

👋 You can always drop me a note by responding to this email, emailing david@davidarkinconsulting.com or calling 832 407 0188.