While it’s true LLMs have changed how I work, it feels incomplete to call them indispensable to my workflow. My brain is no less tired at the end of the day. Humans can’t use server overloaded as an excuse to stop thinking.

I think I’m lucky that some of my most formative work experiences have been in situations with financial impact breathing down my neck. One of the benefits of working with earlier stage startups and bootstrapped/profitable businesses is that ideas are not worth much. Strategic thinking is cheap. Strategic thinking that’s been structured neatly through best practices from bull markets, made to jump through hoops of overworked execs who don’t have time for the fine print - is easy for a tool to emulate.

Whatever it spits out will never be enough. It’s not about the output of the tool; success still hinges on the outcome of the humans who bring their whole messy selves to work.

With that, a few field notes below on using GPT (or your LLM of choice, Claude, Gemini, etc)…

You can't rush the waiting. No tool replaces equity built through time.

H. asked me to estimate how much more productive GPT makes me, or how much more money I might’ve made if I had GPT back during my first independent stint a few years ago.

I might put productivity at 1.2x, but I say even that with hesitance - hard to compare kiwis to guavas. It doesn’t feel like doing more work but denser work. I haven’t been indie again long enough to try to estimate an revenue multiplier, but if there’s anything I’ve learned from sales/biz dev, it’s that no tool can replace the equity you build through time.

Equity in all senses of the word - socially, for relationships to mature; mentally, for applied knowledge to build; financially, for the market to value emerging skills; spiritually, for it all to coalesce into faith that ur gonna be ok, and for that sort of ease of being to attract the right partners.

We mistake AI-generated output as the product of writing or thinking when it’s actually communicating accepted paths of reasoning. So use it to test paths—not replace the hard work of thinking.

As seen through gateway use cases for AI, like writing emails, announcements, meeting agendas. Things that could fall in the corporate comms camp.

To demonstrate the flip side - here’s writer Henrik Karlson on how he thinks of metaphors:

“It's a bit like I'm doing the opposite of what a language model does. A language model plots the words you've said into an embedding space and then looks for words that are close to that point cloud in the embedding space. Whereas I am looking for points that are maximally far from that point cloud but still make sense.”

I used to poo poo on GPT until I realized as a writer, writing was far from where a GPT would be most useful to me. It’s the universe around the writing where it shines, for both business and literary settings.

I already have the insights I want to communicate. Now I have a co-wrangler to draw them out, point them in the field of directions I could travel. (e.g. in packaging, pricing, framing // or in tone, form, angle). Which brings me to…

See what happens when your bring your messiness to GPT.

The other gateway move is using AI for quick answers, summaries, and surface-level brainstorming. The productivity hacks camp - using AI to speed up existing ways of working. (Related to the efficiency & automation camp.)

But there’s a camp using AI to shape new ways of thinking. I’m not trying to sound dramatic but if we understand that with every tool comes the potential for it to help your nonlinear, conceptual work (say, Adobe Creative Cloud capabilities helping a retoucher or photographer redraw their creative evolution) - it’s a no-brainer. We all want to do more of that deep think work without the pesky little things getting in the way.

You don’t need to anthropomorphize tools as thought partners or therapists. Call it what you want, like a tool for intellectual collaboration. Call GPT a tool for adding structure to psychological elasticity.

Just give it more than “give me three ways to do X.” Ask the questions that are really a desperate cry of “how do I push my own thinking and practice forward?”

This is supremely helpful for me as an verbal external processor. My brain at any point in time:

Dancing at a DJ set while do i make eye contact / but do i really wanna interact w someone wearing an ironic balenciaga shirt / maybe just look at the plants climbing up the poles / feels like mexico city / did you see the high schoolers? the glow stick group? no / like patti smith i can’t even be in the moment i am observing / everything is movement everyone is moving past you all the time / british accent guy says i’m sorry excuse me thank you / very courteous for gowanus / sticky floor bathroom smells like urine / oh groundbreaking / where did all these ppl come from?

I won’t bother to share my work related furious meanderings but you can imagine the parallels. I give my scribbles a loose edit for structure and clarity, then I give them to GPT and see what it sees.

For writing & general storytelling, it’s been helpful for:

• Self articulation - Finding what you’re really trying to say
• Structural feedback - Helping your ideas land more powerfully without sacrificing your voice
• Dreaming & playing - Asking what happens if you remove the constraints you’ve assumed are fixed

For career & business development, it’s been helpful for:

• Strategic clarity - Seeing your big creative arc, spotting patterns, and pushing you toward what’s truly yours
• Business model brainstorming – Mapping how creative sustainability could work, with the context of past examples
• Unconventional angles & extrapolation – Asking what happens if you remove the constraints you’ve assumed are fixed (again!)

It’s not just pattern recognition across domains; it’s using that to synthesize your own thoughts at scale and point you to what doesn’t yet exist.

The side effect is that I can also move too fast with GPT. I usually need to take a step back, file away its outlines, let its affirmations quiet my heart, wait and see what deeper thing it prompts in me.

---

I would love to put these principles into practice with you and your team, through a 1 month AI strategy sprint. Interested? 🙂

I’m kicking off a new consulting package designed to help marketing, brand, and editorial leaders actually make AI work for them—in ways that are clear, strategic, and measurable. I’m releasing it to y’all on the newsletter to find the right partners to pilot it with me before expanding further. Sound fun?Email me to set up your AI exploratory chat.

My approach is inspired by a chat I had in Cannes with IBM’s AI/watsonx leader Jeremy Bassinder -

• Rather than shoehorning AI into process: "Implementing AI isn't process change; it’s organizational change."
• And rather than accidental anarchy: "Governance won’t kill your organization, but ambiguity will."

The high level:

• For: Marketing, brand, & editorial leaders looking to improve team effectiveness and energy. We’ll matrix out your needs, surface existing usage patterns, find opportunities/challenges in relational infrastructure (read: culture & change management), and see how your team can modernize creative ops + push innovation at the org level.
• Includes: 2 hour intensive working session with your team (in FigJam), internal qual/quant study w/ insights, next steps roadmap, before/after check-ins with you, and renewed energy on the team thanks to a clear path forward
• Could evolve into… a 1-3 month engagement where I support with implementation, adoption, training, and governance (in line with security standards)

I’ve seen how AI is like recruiting; they’re both things that get thrust on top of your work-work. While we may intuit the value (or have an exec mandate to carry it out), it’s hard to carve out the time to do it well. That’s where I come in.

Interested? Send me a note at [email protected], and I can set up a time for us to chat. If you have clear goals for this, bring them to our chat!