This is a (sort of) cleaned-up version of the podcast episode transcript.
Listen to the full episode right here, or find it on Apple Podcasts and Spotify.
We’ve been hearing “AI” nonstop for the past few years, and I’ve been on the rollercoaster ride for a while now.
I’ve stayed up until midnight coding apps. I’ve despaired that everything is ruined forever.
And I know the minute I release this, something is going to change, and it’ll be out of date. But in some sense, I feel like I’ve settled into a steady AI rhythm, and I have a good sense for how it’s currently working for me.
So I wanted to share this with you. Because one of the biggest things I’ve found helpful (personally) is seeing examples of other people using AI in their business.
I get tired of people saying “just spin up some agents and create these layers and make sure you have the right skills.” I wanna know what work is actually getting done.
Like, what are you actually doing?
So today I’m sharing five ways AI is working for me in my marketing business, and three ways it is NOT.
First, the big question: do you use AI to write?
Let me say this. I am MOSTLY uninterested in reading others’ AI writing. I say mostly and not fully because there are some people who I really respect, and who I know have solid ideas, whose writing sounds very “AI” these days.
I still love to keep tabs on their thinking, and I can look past the AI tells.
I’m not against using AI to finesse sentences, or to help spin up rough drafts from a voice note. But what it essentially boils down to for me is that I don’t want to engage in ideas that a machine made up. I want to engage in ideas with other humans.
In general, I think there are much more interesting ways we can engage with AI.
Five ways AI is working for me
1. As a second brain
Running a business, co-managing a family of five, and trying to have a sliver of life outside of that is like trying to catch a million of those tiny little rubber bouncy balls. To make matters worse, I constantly add way more to my plate, like new business ventures and startups.
I have found doing less in some of these domains to be very helpful. (That is what “Do Less Marketing” is all about, after all.)
But there are certain areas where I simply do not want to give up the quality, and details can be a big part of that quality. Especially with certain client work: the messaging-heavy stuff, websites, and any place where I need to separate the signal from the noise and hold a lot of nuance.
This is where AI comes in. I have an organized desktop system, with folders that keep track of transcripts, discovery, and intake items. It’s often the small details that matter most, the specifics that help an organization stand out and sound like themselves in an age that’s increasingly boilerplate.
Those files live on my desktop, but I can use Claude Cowork to navigate everything. If you’re not familiar with Claude Cowork, it’s basically an AI assistant that can get right into the files and folders on my computer. It can read my documents, organize them, and pull details out of them, instead of just answering questions in a chat window.
So what does this actually mean?
I can be working on messaging and recall that a client said something really interesting about a differentiator, but not remember the details. Instead of scouring through a call transcript and my notes to find the right quote and information, I simply ask AI to find that information for me.
If I’m writing a page that requires a lot of ticky-tacky specific or technical information that I have already gathered, I don’t need to remember it, or even need to leave my workspace. I simply include it as a placeholder in my draft, and AI will fill in the most important details for me to review.
It’s not just helpful for client work. It’s also helpful for my own biz dev. Previously, all my ideas would die in a Google doc, or on a notepad somewhere, stashed under bills and empty coffee mugs. Now, I can put my ideas into my marketing assistant and have it file them away and categorize them for when the right time comes to act on them. It has better recall and can surface things that I’ve forgotten.
This is the game changer: using AI not to decide what’s important, but to fill in the things that are important that my brain doesn’t need to hold constantly. It leaves me space to deal with the nuance, whether that’s the dynamics with stakeholders, or the ways that we can creatively take this work and bring it to life in the world.
2. Navigating tech platforms and solving tech problems
A couple weeks ago, I had to set up some technical platforms, including hosting for some websites and payment processing. I just pulled up Claude Code and had it walk me through step by step in plain language. When I have questions, I literally drop in screenshots to get further direction.
I can finish this stuff in half the time I used to, without digging through some cluttered-up help center. You guys, there is so much in those articles that we do not need to know.
I used it recently to explore different features that Circle has to manage forums and communities. Way faster. I still will check the link to the source article to make sure Claude’s not spitting out something crazy, but it has been remarkably accurate in this area.
3. Mocking up strategy
One of the challenges with strategy work is that it’s almost all thinking. Of course, you end up with the deliverable at some point, but it can be hard to visualize the general direction or structure of a project without sinking a lot of time into mockups and prototypes.
I’m functional with Canva, and I can even dabble a little in the more complex Adobe tools. But generally, it would take way too much time for me to visually mock up my ideas, so I resorted to Google Docs and bullet points, and hoping that my verbal explanation would not be TOO confusing.
Now, with the right context and a few strategic prompts and edits, I can spin up so many things to illustrate a strategy. I can mock up recommended website flows, create a full prototype of a quiz. I even created a visual prototype of that Circle community that I was mentioning earlier, based on the limits of the actual Circle platform itself that AI helped me research.
4. Systematizing things…a little better
I’m a messy organizer when it comes to operations.
I’ll declutter the heck out of a message, website, marketing strategy. But when it comes to admin, I would describe my organizational style as “on-the-go.” I can’t remember which file is where, it all gets out of date, I create duplicates. And then I’m left remembering everything, or reinventing the wheel.
So I’ll ask Claude Cowork to take what I was working on and file it away FOR me. I can also easily create folder structures based on other projects, and more.
I’ll be honest here about how effective this is. I truly think your AI is only as organized, or slightly more organized, than you on your best day.
At least in my own experience, it can spin up a lot of clutter in your folder system, and I’ve been experiencing the downside of this. This use case has not been a silver bullet for me, and I’m still experimenting.
5. Building my own software
The final use case before I get into the ways it has not been working for me: I created my own software to do capacity planning.
I kept running into a challenge that I think is pretty familiar for a lot of solopreneurs. It’s pretty hard to know when I can start a project, and also to know if I’m projecting timelines and budgets right.
I’ll say yes to a project when I really should have waited a month or two to begin it, and then I end up scrambling to make everything fit and becoming a grumpy person through it all. Snapping at kids and working late.
You guys, I am my own boss. There is literally NO good reason for this.
So a while back, I started a mega spreadsheet, mapping out estimated hours for each week for each major project or area of life. So on any given week, I might have a certain number of hours for a client project, a certain number of hours for working out. I even started planning for holidays better. And I mixed in business and life together so I didn’t accidentally plan a huge client project on a week the kids are home for spring break.
Now, I wanna be clear, this is not an exact science. These are really just rough estimates. But it’s helped me see when I can actually slot in new things.
But the spreadsheet was getting a little unwieldy. And it couldn’t tell me whether my plans turned out to be right. I tried trackers like Harvest and Toggl, but none of them tied back to my original plan, so I couldn’t really compare. And without that comparison, the juice wasn’t worth the squeeze.
So I built my own app. It’s super simple: a grid of projects and weeks, so I can see my hours adding up, and then an easy way to track my time as I go. I can see where I can slot in new work, which projects are more profitable, and also where I’m draining hours so I can plan better for the next time.
And the most fun part about this is that I spun it up into a software platform that I’m now selling, called Margin.
The way AI has helped me to make this a sellable product is absolutely insane. If you would’ve told me a year ago that I would be setting up authorization, credit card billing, or even coding and hosting websites on my own, I would never have believed you.
But Claude Code has made it so easy. I know some of you won’t believe me. But you do not have to be super technical to figure this stuff out anymore.
Now, I have a caveat here, and that is that I have a software developer husband. And I think my confidence would’ve been a lot lower in launching this product without having him to gut-check major decisions. But there would have been no way I could’ve done this a few years ago without these tools.
…and three ways AI is not working for me
1. As a personal assistant or project manager
I’ve loaded up a ton of context. I’ve loaded instructions and guardrails.
But AI just does NOT know what I actually need to get done. At this point, at least, I just can’t rely on it to prioritize my days for me without a lot of oversight. And it takes more time than just writing my plan out on my planner.
I’ve been experimenting, and it’s ok at reminding me of things I need to do for a client if I’ve mapped out a large project plan. Which is great.
But even with guardrails, it will prioritize stupid things like “Hey, you need to respond to that one email.” And that one email was an SEO cold pitch from a very spammy email address. It did not even land in my main inbox.
It’s just not working for me right now.
2. Email communication
I wouldn’t mind AI drafting up to-do lists in an email (but again, even that has been rough).
But this goes back to my “robots talking to robots” comment. If a client or colleague asks me a question, you better believe I’m going to respond back with my own words and ideas.
If someone sends me an email that’s like, “So true. And that really hit me…” I’m gonna be like, “JUST TALK TO ME LIKE A REAL PERSON.”
It’s not that I wouldn’t love spending zero time in my inbox. But I think the communication worth sending actually requires you to, I don’t know, like, communicate?
3. Writing when I should be working out my own thoughts
And finally, I have been very intentional not using AI to write things for me when I should be working out my own, individual thoughts on a topic.
I will use AI like a Google doc or a thesaurus to type or speak my thoughts, but I am very careful that I don’t use it to draft things up and completely bypass the process of thinking.
I got into this line of work because I like to think. I also hate it, because sometimes it is hard. Sometimes I turn an idea over and over and over and over again and it just doesn’t seem to click. Working at an idea until things start to come together and it all becomes clear to you takes time, and doesn’t always look very efficient. Sometimes it isn’t efficient.
But I think, ultimately, we have to keep time for inefficient things in our lives, to stay thinking, to stay human.
Where I’ve landed, for now
I’m completely aware that I could be doing AI all wrong. I think we’re all in this boat of experimenting and figuring stuff out. I’ve spent more time on my computer than I think is probably good, which is one reason I planted over 10 different vegetables and herbs in my garden this year. I literally have to get outside to make sure I don’t kill them.
But I’ve been feeling recently like the things I’m good at are still the things I’m good at, and less like AI is this total threat that I’ve seen it as before.
This is a written companion to the first solo episode of Do Less Marketing. Listen to the full episode wherever you get your podcasts.