Happy Wednesday! Below, you’ll find:

Let’s get started!

You’ve probably heard the term “agentic AI” floating around as the next big thing. It sounds technical, abstract, maybe even a little overhyped or downright scary — which makes it easy to ignore if you’re focused on doing real communications work.

Here’s the plain-English version: agentic AI isn’t about a single prompt or a smarter chatbot. It’s about a system of AI agents. Think of it as a team of bots, each running part of a workflow on its own — pulling inputs, applying rules, producing outputs, and looping continuously — with another bot orchestrating the whole thing. We humans are there, too, providing direction, oversight, and judgment.

That distinction between a simple prompt and a network of AI agents is huge — and it matters for communicators. Most of our work isn’t one-off writing; it’s repeatable processes. We pull the same analytics for every campaign. Leadership asks the same questions after every launch. We go through the same reporting cycle every single time — it’s a process that absorbs our time, but rarely changes.

It’s something I needed to learn about.

A few weeks ago, I took the Harvard Data Science Review (HDSR) Agentic AI Intensive. It’s a 2.5-week course, and no, it’s not for coders. It’s for business leaders. It taught me how to map out all the steps needed to create an agentic system.

I focused on social media and internal comms channels, and specifically, how we measure success. Our analytics workflows are repetitive, rules-based, and time-sensitive. They trigger the same questions every time: What happened? How does this compare? What does it mean? What should we do next? And yet we still rely on slow, manual processes that almost guarantee the answers arrive too late to matter — if at all.

The framework I built in the course takes an agentic approach to that exact problem. Instead of humans doing the mechanical work of pulling data, benchmarking, summarizing performance, and rebuilding reports every cycle, specialized AI agents handle those steps continuously. And teams can add an LLM-powered semantic search layer, so if we decide we need to measure something else after the fact, we can. And yes, I added governance layers to ensure that private data stays private and the agents aren’t hallucinating. Humans stay in the loop — but our role shifts from production to judgment: validating outputs, interpreting results, and deciding what to do next.

The outcome isn’t just faster reporting. It’s clearer metrics, more timely insight, and analytics that actually support decision-making instead of post-hoc explanation. In other words: less time building reports, more time using them.

Is this the future of knowledge work? Honestly, I think it is, especially for the boring parts. Repetitive, time-intensive tasks — especially those that aren’t complex — are something we will increasingly offboard to AI agents and agentic AI systems. We’re already seeing software-as-a-service (SaaS) platforms emerge in the market, and I’m sure in-house technology teams are also focusing on similar, bespoke solutions.

There’s no magic here — just systems thinking applied to workflows you already understand. You don’t need to build the bots yourself, but you do need to know how to design the logic behind them. And the good news is that most communications processes are already structured and repeatable, which makes architecting agentic workflows far more approachable than it sounds.

Agentic AI won’t replace communicators — but it will reward the ones who know how to design workflows, ask better questions, and turn insights into action while the moment still matters. We’re not after faster writing. We’re after a deeper understanding, with humans deciding what matters and agents doing the rest.

Most people spend hours every day typing messages they could say in minutes. Wispr Flow fixes that.

Flow turns your voice into clean, polished text inside any app. Speak like you would to a colleague - tangents and all - and get professional output ready to send. 89% of messages go out with zero edits.

Use it for:

Millions of people use Flow daily, including teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, iPhone, and now Android -- free and unlimited on Android during launch.

Get Wispr Flow free →

I’ve seen the image above a half-dozen times on LinkedIn this week. It suggests something you probably already knew — most of the world hasn’t touched AI yet, and those of us who have are still mostly tinkering with the free versions of chatbots. (If you’re skeptical of the data, I was too; here’s a breakdown of sources.)

Approximately 1.5 billion people, constituting less than 20% of the world's population, use AI in some capacity. All of you are in that cohort. Some of us are more cutting edge, forking over some cash to get premium tools and trying things that the free versions don’t unlock, but don’t feel like you need to be there yet. And the two to five million using AI to help code? In the bigger picture, that’s a rounding error.

The key takeaway: You’re not behind the AI curve. You’re ahead of it.

According to my survey data, about 40% of you identify as AI beginners, 50%+ as intermediate, and a small remainder see themselves as advanced. From within our own bubbles, that’s probably true. But from a wider perspective, we’re all in the top 20% at least. And if you’re reading this, your curiosity and dedication will likely win the day. You’ve got this! 😀

Your next step: If you’re in the free user bucket and can afford the $20/month to try something more advanced, now’s a great time to try something new. I pay for ChatGPT because I wanted to make custom GPTs (and I’ve made three already!), but there are plenty of options out there. In fact:

⭐ Share what you’re investing in: Help me and the rest of The Comms Stack community level up! Reply to this email and let me know what paid platforms you’re using; I’ll share a recap of responses in future weeks.

Last week, I attended the prolific Amanda Coffee’s “Mastering GEO for Comms Pros” event in Manhattan. About 100 communicators gathered to learn best practices for ensuring that our clients appeared often — and correctly — in AI results. You can access the replay at the link above, and here are some key quotes and takeaways that resonated with me:

That’s a quote from Marcel Goldstein from Allison Worldwide, and he’s right — we can’t apply SEO learnings to GEO and expect it to work. If you’re trying to rank better on Google Search and you have a robust SEO effort already underway, changes to your core product, brand proposition, etc., are easy to adapt to. But as his co-panelist Jenny Force of Meltwater noted, GEO is different — it’s “not something you can just throw money or just update your site and 24 hours later, it’s fixed.” GEO is a lot more complicated — one of the panelists called the effort “monumental.”

Part of the reason why? Another panelist, Profound’s Joanie Sanders explained: “one team can’t own this alone — you need a cross-functional GEO team.” My view: it makes a lot of sense for communications to lead GEO efforts, but if you’re doing this without marketing and technology involved, you’re working against yourself.

This one pains me. I hate unpublishing things. I’m an archivist at heart.

Someone in the crowd asked a question about LLMs indexing incorrect information and what to do about it. The response: check your own website and other owned media, because there’s a good chance you’re the source of the wrong knowledge. We need to check those properties — and build workflows to do this regularly — to ensure that we don’t have outdated information on these channels. And if we do, we need to update or remove that content.

I don’t think any of the speakers used the phrase “narrative consistency” specifically, but the idea was everywhere. LLMs like to see the same idea stated the same way everywhere and anywhere — that’s a key signal that the models have found the “right” answer. As Eric Carlson, CEO of Notified, said, “at the end of the day, consensus is going to win the day.” The more consistently your narrative shows up across channels, the more likely AI systems are to converge on that version of your story.

(A quick aside — SEO needs to say similar things in many different ways to work; you need to have content across many, many keywords and phrases. That creates a tension with GEO. But there’s a path forward — I’ll have more to come on that in a few weeks!)

We all want to appear in AI chatbot results — but how we appear should be more important than how often we appear. Christine Bartlett, CMO of Hack the Box, said that beyond being present, you want your brand to come across accurately and authoritatively. Or, as Burson’s Steve Rubel said, “the most important metric that is often overlooked is, ‘do these AI engines understand you and the problem you solve for?’”

Quantity matters — there’s no substitute for visibility. But quality is what will drive your business or mission forward.

Photo credit: Andrew Xon

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

And don’t forget about MESSAGE, my custom GPT that works with you to create a great first draft for just about any comm.

Was this forwarded to you by a friend or colleague? Want to get this in your inbox next week? Click here to subscribe!

Until next Wednesday,

Dan