I used to have ten ideas and could only pick one

I've been in and around tech for nearly 20 years, yikes. I often screen prospective hires for having "shiny object syndrome", but maybe I do.  I started in law, on the Microsoft account right around the time that Xbox started spontaneously combusting in homes. I jumped to tech when text messaging was a new thing, right before feature phones and the internet collided. Moved from strategy into design when user experience became the innovation entry point. Left big companies for startups when that was where the biggest leaps were being made.

Every few years the ground shifts and I seem to have jumped at the other side. Not because I am brave (I will never swim off the back of a boat) but because the idea of staying still has always felt worse than the cost of being bad at something new. If that was the stick, the carrot was being closest to customers because it seemed like the best way to make things that made people I knew happy.

Those cycles came every few years, but the last 18 months has felt like a complete cycle a month. AI has made everything possible, and everything scary.

Everyone feels behind right now

A lot of conversation circles around whether engineers picked the right career path. Just a few years ago, I was convinced that not being an engineer was the biggest limitation on my future possible. As a marketer, there are as many people talking about one-person marketing teams and agents that replace entire functions as there are about developers losing relevance.

I don't think the CTO or the new ML grad honestly believes they have a confident view on this. There is a persistent, ambient feeling of "am I behind?" running through every level of tech right now, and the most useful thing any team can do is just name it rather than perform confidence about it. Bloomberg recently ran a whole piece on this feeling. I am quoted in it, somewhere between anxious and liberated.

I wrote a piece earlier this year about how the skills that made you successful might make you useless. Something along the lines of success hardening into instinct, instinct becoming identity, and that becoming the thing you protect. Even then, a few months ago, I didn't know how important that understanding would be.

How a non-technical person starts building

The first months were pretty unglamorous. ChatGPT for SQL queries (hit and miss). Lovable for spinning up landing pages. Playing inside Canva and Figma and random animation tools, where generative AI made me feel like a pilot, with the safety of knowing it was a simulator. That progressed into janky automations in Make and Zapier to pipe social signal into newsletters and morning reports. Most of what I built early on was bad and worked for a week at best. But I was touching the product directly, for the first time in 20 years, rather than directing it from a distance, and that felt liberating.

A lot of this learning happened outside of work, which I think is the best way to approach this moment. Rather than trying to impress your peers, try to satisfy yourself! I designed and sold t-shirts, built promotional microsites and created a color matching app that maps digital brand colors to physical paint swatches, because I was sick of buying the wrong paint. These were sacrificial lambs in the name of learning, and they were fun. When you are surrounded by brilliant engineers and you are the marketing person staring at a terminal for the first time, it can feel exposing. Building something small and personal, with no audience and no stakes, is a very good way to start.

Around the start of this year, I started using Claude Code and I kind of understood it. I know that LLMs will continue to tussle and this is only a moment in time, but in this moment, I finally feel "less behind". I've crossed the chasm of making things for fun, to making things for work, and now, making things other people at work can use.

What I have actually built

The first thing I built was a basic writing skill. Tone matters to me (probably an occupational hazard of working at a voice company and living in five countries). AI-generated text regresses to the mean and everything starts sounding like everything else (average). So I trained a skill on my own published writing, my editing patterns, my internal updates, to preserve something that felt like me rather than like a median of the internet. Notwithstanding that, I have written this with my voice (Wispr Flow), and later some of my trusty agents will tell me what's wrong with it!

I have lived in five countries. I know that a sentence that lands in San Francisco will not land in Auckland. So an obvious place to start building skills was with language - not just localization, but a deep appreciation for the moment, the zeitgeist, the things happening in any given place today.   AI enthusiasm in SoMa reads as obnoxious in cultures with tall poppy syndrome. British humor operates in a completely different register. These are not things you look up in a glossary (do these even exist any more?). They live in the texture of language people speak right now, and you either build for that or you produce content that feels foreign to the people reading it.

This basic entrypoint developed into primitive productivity aids - "don't miss this email", "here's your weekly report", and my first iPhone widget that I looked at far too often. I expanded my audience to my team with "the content banger score" and "partnerships shaman". Over time, these became genuinely useful tools for others: engines that combined an understanding of different types of customers with their local context and infinite pain points to suggest messaging and content we should test.

Now, the team runs a "Marketing-OS":  a shared marketing operating system built on Claude Code that the whole team uses. About 40 playbooks covering paid acquisition, creator partnerships, competitive intelligence, campaign launches, living in Git with direct access to analytics, Slack, Notion, email, CRM. A three-layer memory system so it carries institutional knowledge across sessions (this sounds like a lot because it is, but it grew piece by piece, not as a grand plan).

What changed about the work itself

The leadership model shifted so much these last six months. I used to obsess over effective delegation and operational velocity. Those still matter, but the job moved from directing the work to doing the work - at first directional, and then alongside the team. Building engines that collapse cycle times for everyone. Showing people what is possible instead of writing a brief about what should be possible. And then asking them to scale it.

Someone asked me if this was hard work. Yes and no. Remember, a lot of the early days were dedicated to fun creative explorations, building confidence. But for a few months I was effectively doing everything twice. Old workflow and new workflow, in parallel, comparing results. That is the only way to learn what you can trust and what you can not. By definition, doing everything twice is more work. But it's the only way to get to a place where you know what to trust and what not to, and eventually delegate things that make you more effective.

The taste problem

This comes up in every conversation about AI and marketing: the quality bar.

I sat through probably 50 agency and vendor pitches over the past year promising perfect automated creative workflows. Lots of them had great demos, but when you actually go to implement them, the output is not there. The pixel perfection, the brand alignment, the believability, the taste. It feels like content I would not want to put in front of people, especially as CMO of a brand that stands for amplifying their voice.

So I stopped outsourcing the problem and started building it internally. Right now (which might change tomorrow), I think the only way to make AI output achieve the standard required is to pour your own standards, context, and judgment into the system, with lots of gates for review, rather than relying on AI end to end. But every week, a new plugin solves for clunky handoffs in the end-to-end process from concept to design to published!

The anxiety changes form

I would be lying if I said the feeling of being behind has resolved. But it's my own choice - the reason I work in tech is that anytime something feels comfortable for a couple of weeks, you are guaranteed it will not be soon.

What changed is the nature of the feeling.

“It used to be dependency. I needed others to make the things I imagined. Now it is closer to: there is so much I could build that I have to be disciplined about what I choose.”

This week it was a daily recruiting digest, sweeping referrals out of Slack, inbound out of email and more busywork that keeps me from the deep thinking. It took a morning. A year ago it would have been a headcount request. 

I don't have a good benchmark for where other CMOs are on this (and would love to hear from others!). But the investment has been worth it, not because it solved the anxiety, but because it replaced a feeling of helplessness with a feeling of agency.

The tools are accessible enough now that the barrier is not technical skill. It is willingness to feel uncomfortable for a while. And honestly, the discomfort passes faster than you think.