From Episode 3 of It's A Vibe podcast with Ariel Ibañez, Creative Director at Superside, one of the most AI-driven creative companies out there.
The lie everyone believed: AI adoption would happen overnight.
The reality: It's 2025, and agencies are still figuring out if they should even try.
I sat down with Ariel - a designer who's mentored 500+ creatives on AI - to understand what's actually working.
What I found wasn't about tools at all. It was about a complete mental rewiring of how we think about creative work.
Here's what surprised me most: The best AI creators are the ones who admit they're lazy.
The "Lazy" Advantage
"I've always been a bit lazy," Ariel told me, almost apologetically. "So I've always tried to find cheats or tricks to work faster."
Back in college, he was the Photoshop automation guy. While everyone else was clicking through menus, he was writing scripts. When his friends needed complex actions, they'd ask: "Can you create a folder of Photoshop actions for me?"
That same instinct - finding efficient shortcuts - is what led him to AI four years ago.
In 2021, his studio was stuck in a rut. They were tired of stock photography. They wanted edgy, experimental branding. But hiring a 3D designer? Not in the budget.
Then a colleague posted about something called DALL-E on Twitter.
"We started using that on branding," Ariel said. "And I guess my explanation could be: I'm still finding new ways of work within AI, within technology, within design - which is not far from what I've been doing my whole career."
The insight: The best AI adopters aren't learning new skills. They're applying their existing superpower - problem-solving - to new tools.
Stop Chasing Tools. Start Spotting Patterns.
Here's where most designers get it wrong.
They wake up to 10 new AI tools in their inbox. Runway drops a new model. Luma updates. Flora launches. Webby arrives. Figma adds features.
The instinct? Try to learn them all.
Ariel's approach is different: "I don't aim for tools. I aim for patterns."
Example: He recently discovered Flora and Webby - two AI tools that are node-based (think Comfy UI, but cleaner).
Instead of thinking, "Cool, two new tools to learn," he thought: "Node-based AI tooling is a trend. This is going to be big."
Why this matters:
- Tool chasers spend 10 hours learning software that's obsolete in 3 months
- Pattern spotters invest 10 hours understanding workflows that work across 20 tools for years
"When people ask me, 'What tool should I learn?' I tell them: Don't learn the tool. Learn the workflow."
The Mental Shift Agencies Need to Make
After mentoring 500+ creatives (plus CPMs, sales teams, CEOs), Ariel noticed something:
Everyone's obsessed with tool tutorials. Nobody's talking about mindset.
The old way: Linear workflows. Start → Design → Review → Deliver → End.
The new way: Become a researcher.
"You need to drop the former way of working that we knew from start to end, and try to be more of a researcher.
Try to find time to research - but not on tools. On patterns. On trends."
In practice, this looks like:
1. Allocate research time - Block off hours to experiment, not just execute
2. Document what you discover - Share learnings with your team weekly
3. Build a "helper network" - AI assistants that compound your skills (more on this below)
How He Built an Animation Tool in 2.5 Hours (and 2 Coffees)
A friend approached Ariel with a problem: "I need to display 50 portraits for my photography business on a website, but I want motion without hiring an animator."
Most designers would either:
- Decline (too complex)
- Spend days learning After Effects
- Hire someone
Ariel built a custom Figma plugin instead. In 2.5 hours.
Here's his exact workflow:
Step 1: Talk to his custom ChatGPT prompter
- He has a custom GPT called "Prompter" trained on best practices
- Shared a screenshot of a basic layout
- Asked: "I need a system prompt for Figma to create an app that does X and Y"
- The AI suggested additional features (corner radius, image blur) he hadn't considered
Step 2: Feed the prompt to Figma Make
- Used cursor.ai (an AI coding assistant) to write the code
- Built the plugin in one session
The result: A tool that auto-randomizes 20 different animations every 10-15 seconds, with full control over delay, rotation, blur, and corner radius.
"To me, it's like having Lego as a kid again," Ariel explained. "You can create whatever."
But here's the kicker: He immediately saw another use case. "I was thinking of doing something similar to
present branding projects to my customers. You want some motion graphics when you present, right?"
One tool → Two applications → Hours saved forever.
The Prompting Secret Nobody Tells You
Everyone talks about prompt engineering. "Learn to prompt!" they say.
Ariel's controversial take: "Don't prompt."
Wait, what?
"When I say 'don't prompt,' what I mean is: help yourself prompting with another LLM."
His system:
1. Create a custom ChatGPT called "Prompter" with system instructions based on best practices
2. Talk to it in plain English: "I need ten prompts to create X that does Y"
3. The Prompter creates optimized prompts for you (even suggesting improvements you missed)
"I'm lazy. I don't like to write for two hours. Four years ago, we were crafting highly complex prompts and spending lots of hours. I learned a lot back then. So now I can detect a good prompt from a bad one - but I don't waste time writing them."
For beginners: Search for "MidJourney Prompt Helper" or similar tools in the ChatGPT store. They're already fine-tuned for specific tools.
For advanced users: Build your own by feeding ChatGPT examples of what works in your workflow.
The One Thing AI Can't (Yet) Replace
I asked Ariel the question everyone's thinking: "What can AI never replicate about you?"
His answer was immediate: "Ideas. And creativity."
"At some point, I think AI will replace everything. But right now, it's artificial narrow intelligence - it can't go beyond the data it's trained on. It's not creative."
He's even connecting AI to tools he doesn't know how to use:
- Figma → Cursor → Custom plugins (no coding knowledge needed)
- Blender → Cursor → 3D design (zero Blender experience required)
"I say in natural language, 'Hey, I need to create a scene that does X,' and Cursor just creates it by talking with Blender."
But here's the critical point:
"People think they can grab ideas from an LLM. That is not true. The LLMs are not going to give you good ideas - not without context."
The best AI-generated work he's seen - movies, series titles, teaser trailers, apps, products - all have one thing in common:
"The technical aspect doesn't matter. The idea is what makes it great."
The 5-Year Prediction That Should Change How You Work Today
"Creativity is going to be what we're talking about for the next five years. Ideas are currency right now."
What this means practically:
If you're spending time on technical execution:
- Layout design → AI will handle this soon (already possible with Figma plugins)
- 3D modeling → AI + natural language = no Blender knowledge needed
- Animation → Auto-generate with tools like Ariel's plugin
- Coding → Cursor, GitHub Copilot, and v0 are already there
Where you should invest your time:
- Concept development - What problems are you solving?
- Strategic thinking - Which patterns are emerging?
- Taste - What feels right vs. generic?
- Context - What nuance does your audience need?
"Whatever is technical is going to be done by AI, for sure. So you need to give context. You need to give ideas. You need to give human inputs. And then you can create amazing things with it."
The Most Dangerous Misconception About AI
When we started the conversation, I asked Ariel: "What belief did you have about AI that turned out to be completely wrong?"
His answer: "I thought adoption would be faster."
After four years at the forefront of AI design, mentoring hundreds of creatives, and building tools that make complex tasks trivial - he's still surprised how many agencies aren't using AI at all.
The irony? The barrier isn't technical. It's mental.
Agencies are waiting for:
- The "right" tool to emerge
- Someone to tell them exactly what to do
- Proof that it works
- Permission to experiment
Meanwhile, designers like Ariel are building custom tools in an afternoon. Not because they're smarter. Because they gave themselves permission to be lazy - and that laziness forced them to find better ways.
Your Next Step (Choose One)
Don't try to do everything. Pick ONE of these based on where you're stuck:
If you're drowning in tool updates: → Stop learning tools for one month. Instead, document 3 patterns you notice across multiple AI platforms.
If you don't know where to start: → Create a custom ChatGPT "Prompter" this week. Feed it examples of good prompts in your field.
If you're afraid AI will replace you: → Build something useless but fun with AI this weekend. Seriously. Make a Figma plugin that does something silly. Get comfortable creating, not consuming.
If you're a creative director or agency leader: → Give your team 4 hours per week to research AI - not learn it, RESEARCH it. Have them report patterns, not tools.
The future Ariel predicted isn't coming. It's already here.
The only question is: Are you still trying to compete on execution speed? Or are you ready to compete on ideas?
About This Interview
This post is based on a conversation from It's A Vibe, where I interview product leaders, designers, and builders about how AI is changing creative work.
Ariel leads design at one of the most AI-forward creative companies in the industry and has mentored 500+ creatives on AI workflows. His studio, Neofor Studio, specializes in experimental branding powered by AI tools.
Disagree with something in this post? Good. That means you're thinking critically. Tag me with your take - I'd love to hear it.
Time to read: 8 minutes Tools mentioned: DALL-E, Figma Make, Cursor, ChatGPT, Blender, Flora, Webby, Comfy UI, Runway, Luma
Key takeaway: Stop learning tools. Start spotting patterns.