If you're a creative freelancer using AI and still not making more money, you're not imagining it — the math in fact doesn't work out the way everyone promises. This episode is the honest, unfiltered version of that problem, starting from a night I couldn't sleep doing that exact math in my head.

I'd been lying awake doing math in my head, realizing I was repeating the exact same overcommitting pattern I swore off in 2025 - saying yes to clients randomly, never actually productizing my own service even though I tell other creative professionals to do it constantly. And the math got worse from there: with everything AI has supposedly changed about how I work, I'm charging roughly the same as I was before any of it.

This one's for my editors, designers, writers, video people, social media managers - anyone whose work sits a step or two away from a client's actual revenue, which makes "just charge more" a lot harder to pull off than the advice makes it sound.

I get into why the work itself slowly gets worse when AI lets you take on more than you can actually hold, what I've started doing differently (including capping my client roster and bundling deliverables instead of taking on more people), and the harder question underneath all of it - whether the real fix is productizing, upskilling into something closer to the money, or stepping away from production work entirely.

I don't wrap this one up with a bow. But if you've had your own version of the 2am loop, I think you'll feel it here.

If this one hit home, hit subscribe on Apple Podcasts or Spotify so the next episode lands in your feed automatically.

The written version of this plus a couple of things that didn't make it into the recording — is up on my newsletter and Substack.

Come find me on Instagram or Substack.

If you've got a minute, a review on Apple Podcasts helps other creative pros find the show. Even just a rating helps more than you'd think.

• Instagram
• X
• Threads
• Substack

Here's my actual AI-assisted podcast production workflow - no editor, no team, recorded to fully scheduled in under two hours most weeks. If you're a solo creator wondering how AI tools actually fit into a real production process (not a fantasy one), this is the unglamorous, specific version. Okay so I'm gonna be real with you... when most people think about podcast production, they picture a whole operation. Studio time, an editor, someone writing show notes, someone else doing socials. Days between recording and publishing. And I get it, because that's how a lot of shows run.

Mine doesn't work that way. It's just me. No team, no production days, no back and forth with anyone. And from the moment I finish recording to the point where everything is scheduled — show notes, blog post, newsletter, the whole thing — it's under two hours most weeks.

In this episode I walk through exactly how I do it. The tools, the order I do things in, and why that order actually matters. I talk about how I plan episodes quarterly so I'm never starting from a blank page, the 70/30 split I keep between keyword-researched episodes and the ones I just make because I feel like it, and the Claude skill I built that takes one transcript and spits out every repurposing output in a single pass.

I'm also testing the whole thing on a second podcast right now in a completely different niche. Same system, different content, different audience. If it holds up there, it's actually a system. If it only works here, it's just something I got good at through repetition, which is a different thing.

The principles at the end apply no matter what tools you're using, so those are worth sticking around for.

RESOURCES & LINKS MENTIONED → Audacity – Free audio recording and editing software, used for noise reduction, compression, and EQ immediately after recording → Descript – Transcript-based podcast editing software for content polish, cutting dead air and tangents, adding transitions → Podbean – Podcast hosting platform for upload, metadata, scheduling, and distribution

SUBSCRIBE

If this was useful, hit follow on Apple Podcasts or Spotify so the next one shows up automatically.

The written version of each episode, plus the occasional thing that didn't make it into the recording, lives on the newsletter. Link below if you want in.

Come find me on IG - I'm @tiana.ned

And if you've got 30 seconds, a rating on Apple or Spotify helps new people find the show. That's really all it takes.

• Instagram
• X
• Threads
• Substack

"AI slop" isn't just a meme - it's a real design failure pattern, and it's got a 1985 ancestor most designers have never heard of. This episode breaks down the "ransom note effect," why AI image tools are heading toward smooth sameness instead of visible chaos, and what that means for anyone using AI tools for designers right now. Okay so you know that specific visual, the AI-generated poster with the floating shapes and the busy text that almost looks designed but doesn't quite hold together? There's actually a name for that. A typography term from 1985, coined the first time professional design tools landed in untrained hands. It's called the ransom note effect, and this episode is about why we keep living through the same moment over and over with different tools.

I counted four times it's happened (there's probably more but I focused on these four): desktop publishing in the '80s, early web in the '90s, Canva in the 2010s, and now ChatGPT Image 2.0 in 2026. The difference this time is that the output is smooth enough that we're not heading into the obvious chaos phase. We're heading somewhere more like WordPress sameness: everything competent, everything identical, and way harder to spot than blinking text and mismatched fonts.

I also get into the failure mode that I think is actually the most common one of all, and it's not chaos or sameness, but someone taking something the tool made pretty well and then going in and wrecking it through small edits they didn't know would cost them anything.

And for designers specifically, I talk about where I think the real value gap is sitting right now. Because AI image generation optimizes for looking complete. Design is something else - it's a series of decisions, each one ruling things out. Those two things can look similar on the surface and be totally different underneath.

RESOURCES & LINKS MENTIONED → ChatGPT Image 2.0 – OpenAI's image generation tool released April 2026; renders readable text, generates matching image sets, searches the web mid-process → PageMaker – Desktop publishing software from 1985 that triggered the original ransom note effect when combined with the Macintosh and LaserWriter → Canva – Design platform; My closest historical parallel to the current AI image moment

• Instagram
• X
• Threads
• Substack