Intro to Mods
How to move beyond the search query mindset and transform prompts to into reusable mods
I've been low-key moderating ChatGPT conversations for the past year as I've been crafting a dynamic system that I'm calling base architecture (more on that in later posts).
Before I dive into how the system works, I have to peel back so many layers of what I've been working on this entire time. I rode the train all over the 5 boroughs of NYC showing people Custom GPT's that I've created.
I've watched closely at how people have interacted with this technology and it is honestly fascinating to watch (my UX Researcher hat showing).
It didn't take too much time to notice people's patterns with prompting the system. I noticed that a large segment of those that I showed it too used it like a next generation search engine (fascinating!).
This also made me realize that what I want people to see or to focus on isn't my Custom GPT but the modular architecture that it uses to get the results that it does.
Things I've done with my system:
- Vibe code and build software/dashboards
- Synthesized large datasets, cleaned and visualized
- Generated Reports, Proposals, Email Sequences, and Micro-copy
- Generated Product Ideas, Marketing Campaigns, and Professional Services
- Generated detailed images, illustrations, and info graphics
- and honestly, so much more including meta-prompts!
This also leads to the types of prompts that are being used. Here's another post where you can tryout [copy/paste] your first mod: How a simple prompt can help prevent corrective loops
Here's a mirror reflection:
When most people think about prompts, they treat them like search queries. Type a phrase, hit enter, get an answer. Quick in, quick out. Disposable.
But that’s the shallow view. And honestly, it’s the biggest misconception.
Prompts aren’t queries. Prompts are conversations.
A query assumes the system is just a database: you ask, it retrieves. Generative AI doesn’t work like that. It doesn’t just retrieve. It engages, adapts, and builds on what you give it. And when you see prompts through the “query” lens, you miss the deeper potential of what’s happening: you’re not pinging a static system for answers, you’re co-directing a dialogue with a dynamic model.
This is why I’ve never been satisfied with the idea of prompts as “cheat codes” or plug-and-play shortcuts. At their best, prompts aren’t tricks. They’re the starting moves in a longer arc. They set tone, establish boundaries, and provide context that shapes everything that follows. They’re scaffolding for a dialogue, not keywords for a search engine.
But here’s the problem: even when you understand prompts as conversations, they’re still fragile. You write one, you get a result. You write another, maybe it works, maybe it doesn’t. No consistency. No compounding. No way to reliably carry forward what worked last time.
That’s where the next layer comes in.
That’s where mods come in.
The Case for Mods
A mod is a repeatable framework, a reusable component. Something you can drop into your workflow again and again, knowing it will deliver.
Where a prompt is reactive, a mod is intentional.
Where a prompt is fleeting, a mod is durable.
Where a prompt is a spark, a mod is a framework.
Example — Writing Headlines
- Raw Prompt:
“Write five headlines for a product launch.”
The result? You’ll get five headlines. Some good, some not. Try again tomorrow and you’ll get a completely different set. Disposable. - Modded Conversation (Reusable Framework):
“Generate five product launch headlines. For each one:
– Provide a bold version for social media,
– A clear version for email subject lines,
– A persuasive version for press releases.
Rank them by clarity and emotional pull, and suggest a way to A/B test them.”Now you don’t just have five headlines. You have a system for how headlines are produced every time: consistent angles, structured outputs, built-in checks. If you use this mod again next week, it will give you results you can trust — not randomness.
That’s the difference. Prompts = one-offs. Mods = architecture.
From Education to Collaboration
I can sketch the idea of mods. I can write examples like the one above. But I can’t decide on my own which mods matter most.
That has to come from you — from the real needs you’re facing in your work.
For some people, the most valuable mod might be a framework for drafting sales emails. For others, it might be a structure for synthesizing research or analyzing customer feedback. The possibilities are endless, but the real question is: which ones will make the biggest difference for you?
That’s why I built the Prompt Playground survey.
It’s not just a form. It’s a way for you to help shape this next layer of design. Every response helps identify the most important needs, the kinds of mods that will actually make your work more productive, efficient, and creative.
By contributing, you’re not just filling out a survey — you’re co-authoring the system itself. You’re investing in something that can serve you directly, now and in the future.
Why This Matters
Years from now, I don’t want readers of this newsletter to say, “I got prompt tips in my inbox.”
I want them to say, “This is where I learned to build my own system.”
Not hacks. Not tricks. Not disposable inputs. But frameworks that last.
That’s why I write this newsletter. That’s why your input matters.
Closing Reflection
Fifteen years ago, when I first started studying design, I didn’t realize how much it would ripple outward. I shared what I was learning with a friend. Years later, he told me it inspired him to become a designer. That early influence shaped his whole path.
I think the same thing can happen here. By sharing these reflections on prompts and mods, and by building these systems together, we’re laying down something that people will look back on and say: “That changed the way I work.”
This is just the beginning. Let’s build it together.
