THIS WEEK’S STORY
Designers, Engineers, and Marketers Are Becoming One
We might be heading toward a world where the designer, the engineer, and the marketer become the same person. And that’s extremely exciting.
I was recently at a design engineering event hosted by Shopify in Montreal, and one thing stood out immediately; the traditional line between designers and engineers is disappearing, fast.
Powered by AI, engineers are now shipping beautiful, polished interfaces and exploring design variations in minutes. Designers are moving beyond static Figma screens and building interactive, functional prototypes that feel close to production. Both roles are expanding, not shrinking, because AI has leveled the playing field and dramatically reduced the friction between idea and execution.
What’s even more interesting is that Shopify is intentionally pushing this shift. Teams are encouraged to use AI, experiment freely, and blur disciplinary boundaries so they can build faster and think more creatively.
And it’s not just Shopify.
Tools like Cursor now let you design directly inside your codebase. You can select elements, tweak layouts and visuals, and have the frontend code update automatically. Design decisions turn into production code in real time.
This is creating a new kind of product builder. Someone who understands systems, cares deeply about design quality, and ships at high speed with AI as leverage.
If you learn to operate like this, you will stand out.
The future belongs to the hybrids.
Go build.
HEADLINES
OpenAI dominates the news this week:
Sam Altman declares “Code Red”
Inside OpenAI, “code red” means one thing; drop everything else and focus on ChatGPT. Altman reportedly called for an all-hands push as Google’s Gemini and Anthropic’s Claude Sonnet shipped serious upgrades over the past weeks. Faster models. Better coding. Stronger reasoning.
The message was clear; ChatGPT cannot afford to fall behind.
OpenAI fires back with GPT5.2
And the response came fast.
OpenAI launched GPT-5.2, its most advanced and most practical model so far. Instead of one giant model, GPT-5.2 ships in three variants:
Instant – fast responses for everyday tasks
Thinking – deep reasoning, coding, and analysis
Pro – maximum accuracy for complex workflows
This release follows Altman’s internal memo highlighting slowing ChatGPT usage and weaker benchmarks compared to rivals, especially in coding.
GPT-5.2 is designed to win that ground back.
Here’s why it matters if you build apps with AI:
Writes, fixes, and debugs code with fewer hallucinations
Understands screenshots, images, and long documents
Handles multi-step tasks without complex logic
Faster and cheaper than previous top-tier models
Easy to plug into apps with minimal setup
In short, it behaves less like a chatbot and more like a very sharp coworker who doesn’t get tired.
For non-developers, this is big.
It lowers the gap between having an idea and shipping a working AI-powered app even further.
THIS WEEK’S TIP
Choose your coding models wisely
When you use AI agents to write code for you, your job isn’t to type prompts all day; It’s to become the master orchestrator.
That means having a strategy, knowing your weapons, and choosing the right one at the right time.
Many builders using Cursor don’t realize this, but powerful agents cost money and tokens. If you activate them blindly, you burn credits fast with very little upside.
So here’s how I approach it.
Stop betting on a single model:
There are many models now, and each has strengths and weaknesses. No single one wins at everything.
This is what I personally use today inside Cursor:
Planning complex implementations: Opus 4.5 or GPT5.1 Codex Max
Planning simple implementations: Sonnet 4.5 or GPT5.1 Codex Mini
Implement the plan: Auto mode or Composer1
Quick questions on the code: Auto mode
UX and frontend design: Gemini 3 Pro
Complex debugging: Opus 4.5
Easy debugging: Sonnet 4.5 or auto mode
Treat prompts as cost control:
Structured, shorter prompts tend to perform better than long, verbose ones.
Be explicit about what you want. Reference the exact files. Share documentation when possible. This prevents the agent from over-searching and over-analyzing for no reason, which quietly burns a lot of dollars.
Use MCP servers sporadically:
MCPs let Cursor talk to external tools like Supabase or Figma. They’re powerful, but they’re also expensive in credits. Use them only when they unlock something you truly can’t do otherwise.
Use off-platform LLMs too:
Don’t rely only on Cursor’s built-in models. Use ChatGPT, Claude, and Grok independently to ask questions, sanity-check plans, and evaluate implementations.
I’ve found Grok especially strong for debugging when Cursor gets stuck. Let the models “fight” each other, then feed the best insights back into your main workflow.
AI performs best when you don’t put all your trust in a single brain.
QUICK HITS
Spotify is testing “Prompted Playlists,” letting users create deeply personalized playlists using long, detailed prompts tied to their full listening history.
Mistral’s Devstral 2, a 123B-parameter coding model plus its new Vibe CLI, is pushing French startup Mistral into the AI coding race, challenging U.S. rivals in “vibe coding” workflows.
Enterprise generative AI spend hit $37B in 2025 (3.2x YoY), with $19B going to copilots and sales agents; Anthropic has overtaken OpenAI in enterprise share, and 50+ AI tools now top $100M ARR.
Disney is investing $1B in OpenAI and licensing 200+ characters into ChatGPT and Sora, fueling automated storytelling and raising fresh ethical questions around IP and AI-generated media.
AI reasoning models and agentic systems now dominate 2025 hype, with tools like OpenAI’s o3 boosting complex “chain-of-thought” tasks as analysts predict 80% of enterprise apps will be multimodal by 2030.
And That is all for this week!
See you next Saturday!
Filippo
Filippo Pietrantonio
Whenever you are ready, here are 2 ways I can help you:
Become a member of the aiOS App builders community: In this group, I share content that you won’t find anywhere else + my bullet-proof workflows to build mobile apps with AI to escape your 9 to 5 and launch your dream app.
Get personal help and guidance from me: If you have a great app idea, you are eager to launch (timing is everything!), and you’ve tried different methods but still find yourself stuck with budget constraints… I can help you se yourself free and create what you truly want. Click here and let’s see how we can work together.
